Opinion
23 Civ. 06188 (ER)
12-29-2023
FEDERAL TRADE COMMISSION, Plaintiff, v. IQVIA HOLDINGS INC. and PROPEL MEDIA, INC., Defendants.
OPINION & ORDER
The Court provided this opinion to the parties in final form on December 29, 2023. Public filing of the opinion was delayed, however, in order to ensure that no confidential business information that had been filed under seal was released. Based on input from the parties, certain competitively sensitive information has been redacted from this opinion. See generally FTC v. CCC Holdings Inc., 605 F.Supp.2d 26, 30 n. 1 (D.D.C. 2009) (describing similar procedure).
The Federal Trade Commission (FTC) brought this action to preliminarily enjoin the proposed acquisition of Deeplntent by IQVIA. According to the FTC, the merger will have the effect of substantially lessening competition in the field of programmatic advertising to health care professionals (HCPs). Specifically, the FTC alleges that the acquisition will cause the merger of two programmatic advertisers that compete directly with each other: Deeplntent and Lasso, a division of IQVIA. The FTC asserts that Deeplntent and Lasso, together with PulsePoint, are the three leading firms in the relatively young HCP programmatic advertising industry. Indeed, they have been referred to in IQVIAs internal business records as the "Big 3." Testimony from other market participants, such as pharmaceutical companies and advertising agencies, also suggests that Deeplntent, Lasso, and PulsePoint are the preeminent players in the industry. The vigorous competition among these three firms through the present day has not only resulted in lower prices, according to the FTC, but has also driven technological innovation in the field.
Defendant Propel Media, Inc., is a holding company that owns Deeplntent. Doc. 1 U 24. Throughout this opinion, the Court refers to IQVIA and Deeplntent as "Defendants."
The FTC commenced this lawsuit to put the acquisition on hold pending an in-house administrative proceeding. The FTC contends that this horizontal merger-that is, a merger between two firms in the same industry-will eliminate robust competition between Deeplntent and Lasso and will intensify concentration in the market for HCP programmatic advertising. In addition, the FTC has advanced a vertical theory of harm to competition based on IQVIA's role as a provider of essential data for HCP programmatic advertising. As a legal matter, the question for this Court is whether a preliminary injunction should issue to put the deal on hold while the FTC conducts an in-house proceeding on the merits. And while it does not affect the legal standard this Court is obligated to apply, Defendants have represented that the transaction likely will not go forward if the preliminary injunction is issued.
The parties have proceeded expeditiously with this litigation. Discovery took place on an accelerated schedule, and counsel for both sides worked diligently to prepare for an evidentiary hearing conducted over eight days in late November and early December. The Court heard testimony from twenty-seven witnesses, including industry participants, executives from the merging parties, and economic experts. The parties also submitted thousands of exhibits for the Court's consideration.
After carefully weighing all this evidence, the Court concludes that the FTC has carried its burden at this stage to warrant a preliminary injunction.
I. BACKGROUND
Healthcare advertising is a significant industry. Billions of dollars are spent each year on placing advertisements in front of not only patients but also HCPs-a category that includes doctors, nurse practitioners, physicians' assistants, and other medical professionals. In recent years, due in large part to the COVID-19 pandemic, the industry has shifted away from traditional forms of advertising, which included the familiar practice of sending pharmaceutical company representatives personally into doctors' offices to promote their products. Instead, companies have prioritized advertising through digital means. With that shift, a practice known as programmatic advertising has become widely prevalent in the healthcare industry. In simple terms, programmatic advertising provides an automated process for advertisers to target individuals-in this case individual HCPs-with specific ads on websites across the internet.
A. Digital Healthcare Advertising
Digital advertising is increasingly critical for pharmaceutical companies-for instance, household names such as Bayer and Merck-that wish to advertise their healthcare products or services. These companies often hire advertising agencies to manage their advertising campaigns. Tr. 449 (Margolis). Agencies help these companies make decisions about where to allocate their advertising resources across different "channels" that can be used to advertise to patients and doctors alike. Tr. (Redacted).
There are several digital channels from which those companies and agencies can choose. Social media platforms are one example: they include general-use sites such as Facebook and Linkedln, as well as sites that are healthcare-focused such as Doximity and Sermo. Tr. (Redacted); Doc. 284 (Parties' Joint Stipulated Glossary of Key Terms) at 4. Another channel is connected television, which refers to internet-connected smart television or streaming services that can be viewed on multiple devices. Doc. 284 at 2. Hulu is an example of connected television. Tr. (Redacted). Still another channel is directly buying digital advertising space on websites. Tr. (Redacted). This channel can be further subdivided into "endemic" websites such as WebMD or Medscape, which are healthcare-focused, and "non-endemic" websites such as CNN.com or ESPN.com, which are general interest websites. Doc. 284 at 3. And yet another channel is paid search, a form of digital advertising that places ads to users based on their search keywords and results. Doc. 284 at 4; Tr. (Redacted) PX0569 (Karlova Dep.) at 43. Finally, programmatic advertising, which employs automated technology to purchase digital advertising space across the internet, provides a distinct channel. Tr. (Redacted); Tr. (Redacted)? Tr (Redacted) (In determining how to distribute their advertising resources, companies and agencies evaluate the effectiveness of each channel based on a variety of factors, including cost and reach-that is, the number of potential consumers to see the Tr. (Redacted); Tr. (Redacted).
Doximity was described throughout the hearing as being like Linkedln for doctors, while Sermo was described as being like Facebook for doctors. PX0560 (Chase Dep.) at 65.
Companies use these channels to advertise directly to HCPs, a category that includes "[m]edical doctors, osteopaths, dentists, nurse practitioners, physicians' assistants and other medical professionals authorized to provide medical services to patients in the United States." Doc. 284 at 3. Any HCP with a "National Provider Identifier" (NPI)-an identifying number issued by the government to all HCPs, id.- falls into this category and is a potential advertising target. Tr. (Redacted). Advertisers can use an HCP's NPI number, which is publicly available on the Centers for Medicare and Medicaid Services (CMS) website, "to find them across their digital journey or digital footprint anywhere online that may or may not be a medical website." Tr. (Redacted); see also Tr. 296 (Lin).
The identifying information on the CMS website includes the HCP's name, NPI number, email address, postal address, and medical specialty. Tr. 296-97 (Lin).
Companies also use many of these channels to advertise directly to patients. This practice is known as direct-to-consumer (DTC) advertising. DTC advertising is more prevalent than HCP advertising. Tr. (Redacted); Tr. 461-62 (Margolis). There are far more patients than there are HCPs, so there are correspondingly more opportunities for advertisers on the DTC side as compared to the HCP side. Tr. 461-62 (Margolis). Due in part to the smaller universe of HCPs available to target, HCP advertising is typically more expensive than its DTC counterpart. Tr. (Redacted). The two also differ in that HCPs may be targeted on an individualized basis, while privacy rules and regulations impose more restrictions on DTC advertising. Tr. (Redacted)
B. HCP Programmatic Advertising
Within the broader world of digital healthcare advertising, this case is particularly concerned with HCP programmatic advertising. At a high level, programmatic advertising can be denned as an automated process that matches buyers and sellers of advertising space. Doc. 287 (PL's Proposed Findings of Fact and Conclusions of Law) (FOF) at 9 ("Programmatic advertising is an automated process for digital advertising that facilitates an auction process in microseconds across many digital advertising spaces, making the transaction nearly frictionless."); Doc. 288 (Defs.' Proposed Findings of Fact and Conclusions of Law) at 4 ("[M]any industry participants define [programmatic advertising] as the data-driven matching of an advertiser buyer and a seller in real time through the use of software technology."); see Tr. (Redacted)? Tr. 195 (Gerszke). This matching process occurs in real time: when an HCP visits a website, for instance, the HCP is recognized as such and the website puts up for auction the advertising space that HCP will see-demand-side platforms (DSPs) can then bid for that advertising space on behalf of customers who want to specifically target HCPs. Doc. 288 at 5.
Doc. 287 is the FTC's proposed findings of fact and conclusions of law, which are paginated separately. The Court includes an "FOF" parenthetical when citing the former and a "COL" parenthetical when citing the latter.
There are several players involved in HCP programmatic advertising. On one end of the transaction, publishers-for example, CNN.com or ESPN.com-make their ad space available for purchase. Doc. 284 at 4; Tr. (Redacted). Publishers then turn to supply-side platforms, which are auction-based technology platforms that sell publishers' ad inventory. Doc. 284 at 4; Tr. (Redacted). In the middle of the chain is the ad exchange, which is a technology platform that collects bids from a variety of advertisers and determines a winner by running an auction. Doc. 284 at 1; PX0568 (Gerszke Dep.) at 11-12. Ad inventory goes up for auction in these ad exchanges millions of times per second. Tr. 651-52 (Paquette). From the other side of the transaction, DSPs use software technology to participate in these auctions and purchase ad inventory on behalf of advertisers. Tr. 652 (Paquette); Tr. 801-02 (Lawson).
Healthcare companies work with DSPs, either directly or through advertising agencies, to run HCP programmatic advertising campaigns. Agencies often send out requests for proposal (RFPs) and requests for information (RFIs) to evaluate particular DSPs. Tr. (Redacted) Tr. 150-51 (Gerszke). Some DSPs focus almost exclusively on healthcare advertising. Tr. 257 (Colarossi). Hie three leading healthcare-specific DSPs-both by revenue and by reputation-are Deeplntent, Lasso (which is owned by IQVIA), and PulsePoint. Doc. 287 (FOF) at 27-31 (smnmarizing evidence identifying these three as the most prominent players in HCP programmatic advertising).Meanwhile, another group of "generalist" DSPs offers advertising seivices across multiple industries-such as automotive, financial seivices, and retail-rather than focusing only on healthcare. Tr. 733-34 (Harper). These DSPs include, for example. The Trade Desk and AdTlieorent. Tr. 733 (Harper); Tr. 258 (Colarossi). In recent years, some of these generalist DSPs have sought to increase their participation hi digital healthcare Tr (Redacted).
Lasso itself is not technically a DSP; it is a healthcare advertising platform that partners with a company called Xandr. which is owned by Microsoft, to use its DSP. Tr. (Redacted) Tr. 1228 (Resnick): Doc. 284 at 3. But industry participants view Lasso as a DSP partner. Tr. (Redacted) ad agency executive testifying that "how we see it internally is Lasso is our HCP DSP partner . . . because our primary relationship is with Lasso"): see also Tr. (Redacted).
HCP programmatic advertising campaigns include three principal stages. First, the advertiser identifies a target audience of HCPs to whom it wants to advertise. Advertisers want certain HCPs to see certain ads; for instance, depending on the product, an advertiser may wish to target only dermatologists or only oncologists. Tr. 158 (Gerszke). hi addition, advertisers may wish to target a certain geography-for example, cardiologists in the Houston, Texas area. Tr. 1296-99 (Jena). Targeting an audience also involves determining where an HCP is likely to be consuming content online. Tr. 158-59 (Gerszke). This allows advertisers to compare different environments in which HCPs might be consuming content in order to understand where they can find an HCP at an effective price point. Tr. 159 (Gerszke). DSPs permit advertisers to use a target audience that they themselves create or to use an audience created by the DSP. Tr. 161 (Gerszke).
The second stage is activation. Activation is the process of connecting the target audience to an advertisement in a specific context. Id. After the advertiser defines the target audience it wants to reach and the environment in which it wants to reach them- for example, the New York Times homepage-then the DSP finds that environment and places the advertisement for a specific price. Id. Advertising inventory goes up for auction in ad exchanges millions of times per second. Tr. 651-52 (Paquette). DSPs use software to participate in these real-time auctions for advertising slots, allowing advertisers to buy inventory-that is, advertising space for the actual digital ads-for their target audiences. Tr. 652 (Paquette).
Third, the advertiser measures the performance of the campaign. This measurement can occur in real time as the campaign is ongoing, which is sometimes referred to as optimization. Tr. (Redacted)- Optimization evaluates, on a real-time basis, how much money the ad campaign spent, how many impressions were served, and how many "clicks" the ad received. Id. The advertiser will analyze those results for each advertising partner or media channel-such as programmatic, social media, or search- with whom it is working. Id. As the advertiser receives the results, it may shift resources to those media channels that are working well and move away from those that are not, with an eye on efficiency and providing the best results at the lowest possible cost. Id.
An "impression" refers to "[e]ach instance a digital ad is shown to an individual." Doc. 284 at 3.
Measurement can be conducted at the end of the campaign as well. Tr. (Redacted). Advertisers want to understand the campaign's return on investment-that is, whether it resulted in more prescriptions being written or more products being sold. Id. For example, a pharmaceutical company may want to evaluate whether there was a change in prescribing behavior after HCPs were exposed to an advertising campaign for a certain drug. Tr. 164 (Gerszke). The measurement of whether advertising led to an increase in doctors' writing prescriptions is known as "script lift." Tr. (Redacted); see also Doc. 284 at 4 (denning script lift as "[t]he difference in the number of prescriptions issued by an HCP that is exposed to advertising for a particular drug before and after an advertising campaign"). The ultimate goal of measurement is to understand whether there has been a change in behavior as a result of the campaign. Tr. 165 (Gerszke); Tr. (Redacted). DSPs sometimes offer their own measurement product but also allow for third-party measurement that is connected in the platform. Tr. 161 (Gerszke).
Some DSPs, including Deeplntent, Lasso, and PulsePoint, provide all three of these services-audience-building, campaign activation, and measurement-within one integrated platform. Tr. 605-06 (Paquette); Tr. 253-55 (Colarossi). It is not uncommon, however, for customers-meaning healthcare companies and ad agencies-to "mix and match" these services. Tr. 268-69 (Colarossi); Tr. (Redacted). So, for example, an ad agency may use the audience-building services of a particular DSP but use a different firm to measure the results of a certain campaign.
C. Data Providers
Data is fundamental to HCP programmatic advertising. There are two categories of data that play a central role in the process: (1) identity data and (2) claims and prescription data. Identity data includes basic demographic information such as an HCP's name, email address, and medical specialty. Tr. 295 (Lin). Claims and prescription data refers to more specific information about, among other things, the prescriptions an HCP has written, the number of patients diagnosed, and the procedures performed. Tr. 375 (Brosso); Tr. 1301 (Jena); see also Doc. 284 at 2.
These categories of data are primarily used in the audience-building and measurement stages of an HCP programmatic advertising campaign. At the beginning of a campaign, advertisers rely on data to build the list of NPIs-which correspond to HCPs-that it wants to target with an advertising campaign. Tr. 266 (Colarossi); Tr. 1304-05 (Jena). Later, during the measurement stage, advertisers can use claims and prescription data to analyze script lift and determine whether the campaign was effective. Tr. 376-77 (Brosso); Tr. 1305 (Jena).
IQVIA licenses data to healthcare companies for digital advertising and is an industry leader in many categories of healthcare data. Doc. 288 at 11; Tr. 486 (Margolis). At the same time, there are several other companies that also provide data for both audience-building and measurement services. Doc. 288 at 30 (collecting testimony). That is because the underlying data is not owned by any single data provider; instead, data providers obtain data-such as, for example, information about how many prescriptions an HCP is writing-from third-party sources like insurance companies and pharmacies. Tr. 1301-04 (Jena). Data providers then curate the data to make it usable for customers. Tr. 371-72 (Brosso); Tr. 561-62 (Evenhaim).
IQVIA also maintains a "Third Party Access" (TPA) program, which provides a mechanism for data providers to share their data with third parties. Tr. 570 (Evenhaim). Under the TPA program, when a pharmaceutical customer with a license for IQVIA data used in audience creation or measurement wants to run that data through a non-IQVIA firm-such as a DSP-that firm must seek access from IQVIA pursuant to a TPA agreement. Tr. 162-64 (Gerszke); Tr. 571 (Evenhaim); Tr. 1185-86 (Kress). Other data providers operate similar third-party licensing programs. Tr. 570-71 (Evenhaim); Tr. (Redacted)
D. The Parties and the Proposed Transaction
IQVIA is a healthcare company that provides a variety of services to healthcare clients. Tr. 1204-11 (Resnick). In recent years, IQVIA has made a series of acquisitions connected to HCP programmatic advertising. In November 2019, IQVIA acquired MedData Group, whose primary business is selling HCP identity data used for the creation of HCP audiences. Tr. 286 (Lin); Doc. 284 at 3. IQVIA acquired another HCP data provider, DMD Marketing Solutions, (Redacted) in August 2021. Tr. 286, 336 (Lin). DMD Marketing Solutions also sells HCP identity data that is used for advertising to HCPs. Doc. 284 at 3. These acquisitions, according to one IQVIA employee, made IQVIA "by far the largest data provider in this vertical" and gave it "the industry's largest HCP reach." PX1140; PX1205-01.
Lasso was founded in late 2019. Tr. 520 (Field). In July 2022, IQVIA acquired Lasso (Redacted). Tr. 478-79 (Margolis). As mentioned above, while Lasso is not technically a DSP, customers view Lasso as a DSP partner. Tr. (Redacted) ? Tr. (Redacted) Deeplntent, Lasso, and PulsePoint are the three leading healthcare-focused DSPs in the industry and have been referred to internally by IQVIA as the "Big 3." PX1625-02; Doc. 287 (FOF) at 27-31.
Lasso is now known as the IQVIA Operating System. Tr. 244 (Colarossi); Doc. 284 at 3. For the sake of convenience, however, the Court refers to it as Lasso throughout this opinion, as the parties generally have done in their briefing and arguments.
Deeplntent was founded in 2015, and it began to focus on HCP programmatic advertising two years later. Tr. 644-45 (Paquette). In early 2022, around the same time IQVIA was planning its acquisition of Lasso, it also contemplated an acquisition of Deeplntent. Tr. 1223-24 (Resnick). IQVIA sought to acquire Deeplntent (Redacted) Tr. 337 479
As will be discussed in more detail, Deeplntent and Lasso have consistently viewed each other as significant competitors in the HCP programmatic advertising space. Doc. 287 (FOF) at 41-52 (summarizing evidence of head-to-head competition). Following the acquisition of Lasso and the planned acquisition of Deeplntent, one IQVIA executive wrote: "IQVIA has made a HUGE investment in this space over the last couple of years and it is profoundly obvious to everyone in the industry that IQVIA is leading a consolidation." PX1377-01. IQVIA and Deeplntent agreed on the terms of the proposed acquisition in July 2022. Doc. 1 ¶ 25.
II. PROCEDURAL HISTORY
On July 17, 2023, following a lengthy investigation, the FTC Commissioners by a 3-0 vote found reason to believe that the proposed acquisition of Deeplntent by IQVIA would substantially lessen competition in violation of section 7 of the Clayton Act, 15 U.S.C. § 18, and section 5 of the FTC Act, 15 U.S.C. § 45. Doc. 1 If 26. The FTC commenced an administrative adjudication proceeding to determine whether the acquisition is unlawful. Id.
The following day, the FTC filed this action seeking a temporary restraining order (TRO) and preliminary injunction to prevent Defendants from consummating the acquisition pending the outcome of the administrative proceeding. Id. at 1-2. On July 21, 2023, the parties agreed to a TRO blocking the acquisition until the earlier of November 22, 2023, or the third business day following the Court's ruling. Doc. 44. The Court subsequently extended the TRO until December 29, 2023, and advised the parties that it would issue a decision on the preliminary injunction by that date. Doc. 144 at 40. The parties agreed to an amended TRO blocking the acquisition until the earlier of December 29, 2023, or the third business day following the Court's ruling. Doc. 154. An administrative trial before an Administrative Law Judge (ALJ) was scheduled to begin on December 20, 2023. Doc. 1 at2.
During the evidentiary hearing, counsel for IQVIA stated that the parties had heard from the ALJ that "he does not intend to proceed with the [administrative] hearing until he receives a ruling from this Court." Tr. 945. Counsel said the ALJ had invited the parties to file a motion for a continuance, "which he noted is routinely granted in these circumstances." Tr. 945. And "[e]ven without a continuance he indicated that he will not hear any evidence until January 4, 2024." Id. Counsel for the FTC responded that opening statements in the administrative trial would proceed on December 20, 2023, and that she was" 100 percent confident that we will proceed with the evidence in January." Tr. 945, 947.
The parties proceeded with discovery on an accelerated schedule. On September 20, 2023, the FTC moved to strike several constitutional and equitable defenses raised in Defendants' answers. Doc. 146. The Court granted that motion in an opinion issued on October 31, 2023. Doc. 184; see FTC v. IQVIA Holdings Inc., No. 23 Civ. 06188 (ER), 2023 WL 7152577 (S.D.N.Y.Oct. 31, 2023). On November 13, 2023, the FTC moved in limine to exclude parts of the proposed testimony of one of Defendants' experts, Dr. Anupam B. Jena. Doc. 215. The Court deferred decision on the motion, concluding that it would hear Dr. Jena's testimony and then determine what weight to afford his opinions. Doc. 307 at 33-34.
An evidentiary hearing on the preliminary injunction began on November 20, 2023. The Court heard testimony from twenty-seven witnesses over the course of eight days. The parties submitted proposed findings of fact and conclusions of law on December 7, 2023, Docs. 287, 288, and the Court heard closing arguments on December 8,2023.
III. LEGAL STANDARD
A. Section 13(b) of the FTC Act
Section 13(b) of the FTC Act authorizes the FTC to "file suit in the federal district courts and seek a preliminary injunction to prevent a merger pending [an] FTC administrative adjudication' [w]henever the Commission has reason to believe that a corporation is violating, or is about to violate, Section 7 of the Clayton Act.'" FTC v. Penn State Hershey Med. Ctr., 838 F.3d 327, 337 (3d Cir. 2016) (second alteration in original) (quoting FTC v. H.J. Heinz Co., 246 F.3d 708, 714 (D.C. Cir. 2001)). The FTC may obtain a preliminary injunction "[u]pon a proper showing that, weighing the equities and considering the Commission's likelihood of ultimate success, such action would be in the public interest." 15 U.S.C. § 53(b). Courts thus follow a two-step inquiry that asks (1) whether the FTC has shown a likelihood of ultimate success on the merits in the administrative proceeding, and (2) whether the equities weigh in favor of an injunction. E.g., Perm State, 838 F.3d at 337; FTC v. Univ. Health, Inc., 938 F.2d 1206, 1217-18 (11th Cir. 1991).
In this case, there is some disagreement as to what is required for the FTC to demonstrate a likelihood of ultimate success. Relying on FTC v. Lancaster Colony Corp., 434 F.Supp. 1088 (S.D.N.Y. 1977), the FTC argues that it need only show "a fair and tenable chance of ultimate success on the merits." Doc. 287 (COL) at 1 (quoting Lancaster, 434 F.Supp. at 1090). The FTC further asserts that "it is not the role of the court at this stage 'to embark upon a detailed analysis' of the factual record or 'resolve these factual issues on this motion.'" Id. (quoting Lancaster, 434 F.Supp. at 1094).
Defendants, on the other hand, insist that Lancaster is an "outlier decision" and that the "fair and tenable chance" standard has been rejected by other courts. Doc. 288 at 49-50. Instead, appellate courts have asked "whether the FTC has raised questions going to the merits so serious, substantial, difficult and doubtful as to make them fair ground for thorough investigation, study, deliberation and determination by the FTC in the first instance and ultimately by the Court of Appeals." Id. at 49 (internal quotation marks and citation omitted). Defendants also contend that the FTC's request for a preliminary injunction should be denied "[u]nder any formulation of the standard." Id. at 51.
In opposing the FTC's motion to strike their constitutional and equitable defenses, Defendants argued that the FTC's likelihood of ultimate success must be measured by the final resolution of the proceedings rather than by chances of success on the antitrust merits in the administrative proceeding. Doc. 156 at 9. The Court rejected that argument, finding that the weight of the case law indicates that the administrative proceeding is the proper forum in which to measure the FTC's likelihood of success. IQVIA, 2023 WL 7152577, at *3-6; see, e.g., FTC v. Microsoft Corp., F.Supp. 3d -, -, No. 23 Civ. 02880 (JSC), 2023 WL 4443412, at *9 (N.D. Cal. July 10, 2023) (concluding that "the relevant forum for the question of likelihood of success is before the ALJ in the administrative proceedings"), appeal filed, No. 23-15992 (9th Cir. July 13, 2023); FTC v. Meta Platforms Inc., No. 22 Civ. 04325 (EJD), 2022 WL 16637996, at *6 (N.D. Cal. Nov. 2, 2022) (same). The Court also observed that Defendants had objected to the "fair and tenable chance" standard. IQVIA, 2023 WL 7152577, at *4 n. 1. But since the motion to strike would succeed regardless of the standard, the Court expressly noted that it did not need to decide which test would ultimately apply to the section 13(b) inquiry. Id.
The Second Circuit addressed the continuing viability of Lancaster in a case that involved injunctive relief under a different statute, the Flammable Fabrics Act (FFA). See United States v. Sun & Sand Imports, Ltd., 725 F.2d 184 (2d Cir. 1984). The Second Circuit observed that Lancaster had interpreted section 13(b) to require a showing of "a fair and tenable chance of ultimate success on the merits." Id. at 188. And the court explained that applying that standard to the FFA "strikes us as appropriate." Id. In a footnote, the court said it did not see "any significant difference between the 'serious question' standard and the 'fair and tenable chance' standard"-though it noted another court's conclusion that the "serious question" test places a higher burden on the government. Id. at 188 n.5 (citing FTC v. Nat'l Tea Co., 603 F.2d 694, 698 (8th Cir. 1979)). Other courts in this District have since invoked Lancaster's formulation of the standard. See FTC v. Verity Int'l, Ltd., 194 F.Supp. 2d 270, 281 (S.D.N.Y. 2002) (court must "determine that the FTC has a fair and tenable chance of ultimate success on the merits"); FTC v. Crescent Publ'gGrp., Inc., 129 F.Supp. 2d 311, 319 (S.D.N.Y. 2001) (same).
The Court concurs with the suggestion in Sun & Sand that there is no meaningful difference between the two standards. The specific words used to describe the test are not particularly important given that the Lancaster opinion itself included both "fair and tenable" language and "serious questions" language. In articulating the section 13(b) standard, the court explained that the FTC could meet its burden by showing "that it has a fair and tenable chance of ultimate success on the merits." Lancaster, 434 F.Supp. at 1090. The court continued:
In short, if it shows that the newly-minted "equities" weigh in its favor, a preliminary injunction should issue if the FTC has raised questions going to the merits so serious, substantial, difficult and doubtful as to make them fair ground for thorough investigation,
study, deliberation and determination by the FTC in the first instance and ultimately by the Court of Appeals.Id. at 1090-91. The Lancaster court, then, appears to have used the "serious questions" language to illustrate how the FTC could show a "fair and tenable chance" of success on the merits.
As Judge Flaum put it in FTC v. Rhinechem Corp., 459 F.Supp. 785 (N.D. 111. 1978), "[t]he phrase 'fair and tenable' seems to be less informative than the detailed elaboration of Lancaster Colony'?, 'likelihood of success' standard itself." Id. at 789 n.4. That articulation of the standard "has the additional virtue of keeping the courts' sight fixed on their primary responsibility in section 13(b) cases." Id. In other words, "[t]he central issue for us is not whether respondents have violated, or are about to violate, the antitrust laws, for adjudication of those issues is vested in [the] FTC in the first instance." Id. (quoting Lancaster, 434 F.Supp. at 1090). Instead, "the question for us is whether the FTC has shown prima facie that the public interest requires that a preliminary injunction issue to preserve the status quo until the FTC can perform its adjudicatory function." Id. (quoting Lancaster, 434 F.Supp. at 1090).
To the extent the "fair and tenable chance" standard could be construed as merely "rubber stamping" any request by the FTC for a preliminary injunction, several courts have rejected that approach. See, e.g., FTC v. Whole Foods Mkt, Inc., 548 F.3d 1028, 1035 (D.C. Cir. 2008) (opinion of Brown, J.) (noting that a court may not "simply rubber-stamp an injunction whenever the FTC provides some threshold evidence"); FTC v. Meta Platforms Inc., 654 F.Supp.3d 892, 911 (N.D. Cal. 2023) (explaining that the FTC is "required to provide more than mere questions or speculations supporting its likelihood of success on the merits"); FTC v. Sysco Corp., 113 F.Supp. 3d 1, 23 (D.D.C. 2015) ("Though more relaxed than the traditional equity injunction standard, Section 13(b)'s public interest standard nevertheless demands rigorous proof to block a proposed merger or acquisition."); FTC v. Arch Coal, 329 F.Supp.2d 109, 116 (D.D.C. 2004) (noting that "the FTC's burden is not insubstantial"). A court must exercise its "independent judgment" in analyzing whether the FTC has met its burden to show a likelihood of success on the antitrust merits. See, e.g., Whole Foods, 548 F.3d at 1035 (opinion of Brown, J.); FTC v. Microsoft Corp., - F.Supp.3d -, -, No. 23 Civ. 02880 (JSC), 2023 WL 4443412, at *8 (N.D. Cal. July 10, 2023), appeal filed, No. 23-15992 (9th Cir. July 13, 2023). And some courts, as Defendants point out, have explicitly rejected the "fair and tenable chance" standard. See FTC v. Freeman Hosp., 69 F.3d 260, 267 (8th Cir. 1995) (holding that the "fair and tenable chance" standard "runs contrary to congressional intent and reduces the judicial function to a mere 'rubber stamp' of the FTC's decisions"); FTC v. Staples, Inc., 970 F.Supp. 1066, 1072 (D.D.C. 1997) ("It is not enough for the FTC to show merely that it has a 'fair and tenable chance' of ultimate success on the merits as has been argued and rejected in other cases.").
Lancaster remains well supported, however, insofar as it emphasizes that "the scope of the Section 13(b) inquiry is necessarily limited and narrow." FTC v. Meta Platforms Inc., No. 22 Civ. 04325 (EJD), 2022 WL 16637996, at *5 (N.D. Cal. Nov. 2, 2022) (citing Lancaster, 434 F.Supp. at 1091). Courts have explained that the FTC's burden to obtain a preliminary injunction is lower than the burden it faces in the administrative proceeding. See, e.g., Penn State, 838 F.3d at 337 ("In its administrative adjudication, the FTC must show that the proposed merger violates Section 7 of the Clayton Act. ... At this stage, '[t]he FTC is not required to establish that the proposed merger would in fact violate section 7 of the Clayton Act.'" (alteration in original) (quoting Heinz, 246 F.3d at 714)); Staples, 970 F.Supp. at 1071 ("The determination of whether the acquisition actually violates the antitrust laws is reserved for the Commission and is, therefore, not before this Court.").
Accordingly, courts do not require the FTC to prove the merits of its case in a section 13(b) proceeding. See Whole Foods, 548 F.3d at 1035 (opinion of Brown, J.) ("[A] district court must not require the FTC to prove the merits, because, in a [section 13(b)] preliminary injunction proceeding, a court 'is not authorized to determine whether the antitrust laws ... are about to be violated.'" (omission in original) (quoting FTC v. Food Town Stores, Inc., 539 E2d 1339, 1342 (4th Cir. 1976))); Univ. Health, 938 E2d at 1218 (explaining that court's task is "to make only a preliminary assessment of the [acquisition]'s impact on competition" (alteration in original) (citation omitted)); Staples, 970 F.Supp. at 1072 ("[T]he FTC need not prove to a certainty that the merger will have an anti-competitive effect."); see also FTC v. Warner Commc 'ns Inc., 742 F.2d 1156, 1164 (9th Cir. 1984) ("Because the issue in this action for preliminary relief is a narrow one, we do not resolve the conflicts in the evidence, compare concentration ratios and effects on competition in other cases, or undertake an extensive analysis of the antitrust issues." (citing Lancaster, 434 F.Supp. at 1094, 1096)); Microsoft, - F.Supp.3d at -, 2023 WL 4443412, at *8 (noting that section 13(b) requires "rigorous analysis" but that the court "does not resolve conflicts in the evidence").
At bottom, the Court may not simply rubber stamp the FTC's request for an injunction upon the showing of a mere possibility that the antitrust laws will be violated. Instead, the Court must exercise its independent judgment to determine whether the FTC has raised serious questions about the antitrust merits that warrant thorough investigation in the first instance by the FTC. In doing so, however, the Court may not require the FTC to prove the merits of its case or to establish a violation of the Clayton Act. That inquiry is reserved for the administrative proceeding.
B. Section 7 of the Clayton Act
Section 7 of the Clayton Act prohibits mergers or acquisitions whose effect "may be substantially to lessen competition, or to tend to create a monopoly," in "any line of commerce or in any activity affecting commerce in any section of the country." 15 U.S.C. § 18. This provision "was intended to arrest a trade restraint or a substantial lessening of competition in its incipiency; it is not concerned with 'certainties.'" Fruehauf Corp. v. FTC, 603 F.2d 345, 351 (2d Cir. 1979) (citing Brown Shoe Co. v. United States, 370 U.S. 294, 317-18 (1962), and United States v. E.I. duPontde Nemours & Co., 353 U.S. 586, 589 (1957)). As the Second Circuit has explained:
Requiring a plaintiff to prove that substantial lessening of competition is inevitable would thwart the express intent of Congress to nip anticompetitive practices in the bud before they blossom into a Sherman Act restraint of trade, and would run counter to Congress' view that neither the Commission nor the courts should be charged with possession of powers of prevision that no one else has achieved.Id. (internal citation omitted).
Section 7 requires a "reasonable probability" of a substantial impairment of competition; a "mere possibility" is not sufficient. Id. Courts must evaluate the likelihood of anticompetitive effects "in the context of the 'structure, history, and probable future' of the particular markets that the merger will affect." New York v. Deutsche Telekom AG, 439 F.Supp.3d 179, 198 (S.D.N.Y.2020) (quoting United States v. Gen. Dynamics Corp., 415 U.S. 486, 498 (1974)). But again, at this stage, "[t]he FTC is not required to establish that the proposed merger would in fact violate section 7 of the Clayton Act." Penn State, 838 F.3d at 337 (alteration in original) (quoting Heinz, 246 F.3dat714).
Courts assess section 7 claims challenging horizontal mergers under a burden-shifting framework. See, e.g., id.; United States v. Baker Hughes Inc., 908 F.2d 981, 982-83 (D.C. Cir. 1990); see also In re AMR Corp., No. 22-901, 2023 WL 2563897, at *2 (2d Cir. Mar. 20, 2023) (collecting cases from other circuits that follow the burden-shifting approach). First, the FTC must make out a prima facie case that the merger is anticompetitive. Penn State, 838 F.3d at 337. If it does so, the burden shifts to the defendant to produce evidence rebutting that prima facie case. Id. And if the defendant is successful, "the burden of production shifts back to the Government and merges with the ultimate burden of persuasion, which is incumbent on the Government at all times." Id. (quoting Saint Alphonsus Med. Ctr.-Nampa Inc. v. St. Luke's Health Sys., Ltd., 778 F.3d 775, 783 (9th Cir. 2015)).
As one court has remarked, this approach "conjures up images of a tennis match where the government serves up its prima facie case, the defendant returns with evidence undermining the government's case, and then the government must respond to win the point." Univ. Health, 938 F.2d at 1219 n.25. But in practice, as in this case, "the government usually introduces all of its evidence at one time, and the defendant responds in kind." Id.
This burden-shifting approach is less entrenched in the context of vertical mergers. See Microsoft, - F.Supp.3d at -, 2023 WL 4443412, at *8 (noting that the Ninth Circuit has not addressed whether the burden-shifting framework applies to vertical mergers); cf. United States v. AT&T, Inc., 916 F.3d 1029, 1037 (D.C. Cir. 2019) ("There is a dearth of modern judicial precedent on vertical mergers and a multiplicity of contemporary viewpoints about how they might optimally be adjudicated and enforced."). The D.C. Circuit followed the burden-shifting approach in one vertical merger case, but it explained that "unlike horizontal mergers, the government cannot use a short cut to establish a presumption of anticompetitive effect through statistics about the change in market concentration, because vertical mergers produce no immediate change in the relevant market share." AT&T, 916 F.3d at 1032; see also Fruehauf, 603 F.2d at 351 (observing that "[a] vertical merger, unlike a horizontal one, does not eliminate a competing buyer or seller from the market" and therefore does not "automatically have an anticompetitive effect"). Instead, in a vertical merger case, "the government must make a fact-specific showing that the proposed merger is likely to be anticompetitive." AT&T, 916 F.3d at 1032 (internal quotation marks and citation omitted). If it does so, the familiar Baker Hughes framework applies: "the burden shifts to the defendant to present evidence that theprimafacie case inaccurately predicts the relevant transaction's probable effect on future competition or to sufficiently discredit the evidence underlying the prima facie case." Id. (internal quotation marks and citation omitted). And if the defendant is successful, the burden of producing additional evidence of anticompetitive effects shifts back to the government and merges with the ultimate burden of persuasion. Id; see also Microsoft, - F.Supp.3d at -, 2023 WL 4443412, at *12 (applying burden-shifting framework in vertical merger case).
IV. DISCUSSION
The FTC has challenged the proposed acquisition under both a horizontal theory and a vertical theory. In short, the horizontal theory asserts that the proposed acquisition will increase concentration in the market for HCP programmatic advertising and will eliminate the substantial head-to-head competition between Deeplntent and Lasso that continues to this day. The vertical theory posits that IQVIA's acquisition of Deeplntent will allow it to foreclose other industry participants from accessing IQVIA's data, which is a key input for HCP programmatic advertising.
Defendants challenge both theories on numerous grounds. Most fundamentally, they argue that the FTC has denned the market for HCP programmatic advertising too narrowly because other channels of advertising-such as social media and endemic websites-offer reasonably interchangeable alternatives. Even within the FTC's proposed market, moreover, Defendants contend that competition will remain vibrant post-merger in what they characterize as a dynamic and rapidly evolving industry. And with respect to the vertical theory, Defendants assert that IQVIA will have neither the ability nor the incentive to foreclose other companies from accessing HCP data.
After careful consideration of the evidence presented, the Court concludes that the FTC is likely to succeed on its horizontal challenge and that the equities weigh in favor of preliminary injunctive relief.
Because the FTC has shown that it is likely to succeed on the horizontal challenge and that the equities weigh in favor of preliminary injunctive relief, the Court need not decide whether it has also carried its burden on the vertical theory.
A. Likelihood of Success
For the FTC to make out a prima facie case, it must (1) define a relevant market, and (2) show that the effect of the merger in that market is likely to be anticompetitive. Perm State, 838 F.3d at 337-38. If it is able to do so, the burden shifts to Defendants to produce evidence rebutting the FTC's prima facie case. Id. at 347.
1. Relevant Product Market
The FTC's initial burden is to define a relevant market in which the proposed acquisition is likely to harm competition. See, e.g.,Meta, 654 F.Supp.3d at 911 ("The first step in analyzing a merger challenge under Section 7 of the Clayton Act is to determine the relevant market."); FTC v. Peabody Energy Corp., 492 F.Supp.3d 865, 884 (E.D. Mo. 2020) ("This initial step [of market definition] is mission-critical for all FTC merger challenges."). "Denning the relevant market is a 'necessary predicate' to finding a Clayton Act violation because the proposed merger 'must be one which will substantially lessen competition within the area of effective competition.'" FTC v. RAG-Stiftung, 436 F.Supp.3d 278, 291 (D.D.C. 2020) (quoting du Pont, 353 U.S. at 593). The scope of the relevant market "dictates the analysis of market power and the merger's anticompetitive effects." Id.
Market definition has two components: the relevant product market and the relevant geographic market. E.g.,Meta, 654 F.Supp.3d at 911. The parties here agree- or at least do not meaningfully contest-that the geographic market is worldwide. Doc. 287 (FOF) at 26; Doc. 288 at 54 (stating that "whether the relevant market is the United States or global is not dispositive of the Court's analysis"). But they forcefully dispute the scope of the relevant product market.
The FTC argues that the relevant product market is HCP programmatic advertising. Doc. 287 (FOF) at 9. The FTC defines programmatic advertising as "an automated process for digital advertising that facilitates an auction process in microseconds across many digital advertising spaces." Id; see also Id. at 13 (HCP programmatic advertising allows advertisers "to reach health care providers wherever they are on the internet across multiple media properties"). According to the FTC, HCP programmatic advertising has distinct characteristics and is not reasonably interchangeable with other forms of digital marketing. Id. at 18-22. Thus, the FTC's position is that channels such as social media and endemic websites should not be included in the market.
Defendants define the relevant product market more broadly. They argue that the FTC's market definition is arbitrary because the exact same advertisement could be sent to an HCP through different means-for example, through both a DSP and a social media website-but the FTC would divide those advertisements into separate markets. Doc. 288 at 54. Defendants further assert that the FTC's proposed market is unduly narrow because it excludes several reasonable substitutes to which advertisers can turn for HCP programmatic advertising. According to Defendants, social media platforms and endemic websites can offer programmatic advertising targeted at HCPs and thus are reasonable substitutes that should be included in the market. Id. at 55.
The relevant product market is denned by "the reasonable interchangeability of use or the cross-elasticity of demand between the product itself and substitutes for it." Brown Shoe, 370 U.S. at 325. The goal is "to identify the market participants and competitive pressures that restrain an individual firm's ability to raise prices or restrict output." Geneva Pharms. Tech. Corp. v. Barr Lab' ys Inc., 386 F.3d 485, 496 (2d Cir. 2004); see also RAG-Stiftung, 436 F.Supp.3d at 292 ("When aggregating products into a relevant market, courts focus on demand substitution because it illuminates whether customers can switch to one product and constrain anticompetitive pricing in another."). A relevant product market includes "all goods that are reasonable substitutes, even though the products themselves are not entirely the same." Sysco, 113 F.Supp.3d at 25; see also Peabody, 492 F.Supp.3d at 884 (observing that "a properly denned product market includes the functionally similar products to which customers could turn" in the event of a post-acquisition price increase); Staples, 970 F.Supp. at 1074 (explaining that "the general question is 'whether two products can be used for the same purpose, and if so, whether and to what extent purchasers are willing to substitute one for the other'" (quoting Hqyden Publ'g Co. v. Cox Broad. Corp., 730 E2d 64, 70 n.8 (2d Cir. 1984))).
Market definition is guided by the "narrowest market" principle. Sysco, 113 F.Supp.3d at 26. The product market "must be drawn narrowly to exclude any other product to which, within reasonable variations in price, only a limited number of buyers will turn." RAG-Stiftung, 436 F.Supp.3d at 292 (quoting Times-Picayune Publ'g Co. v. United States, 345 U.S. 594, 612 n.31 (1953)); see also Peabody, 492 F.Supp.3d at 886 (noting that "this Court's task is to identify the narrowest market within which the defendant companies compete that qualifies as a relevant product market"). While a firm may be a competitor in the overall marketplace, that "does not necessarily require that it be included in the relevant product market for antitrust purposes." Staples, 970 F.Supp. at 1075; see also FTC v. Advoc. Health Care Network, 841 F.3d 460, 469 (7th Cir. 2016) (noting that the relevant market need only include "the competitors that would 'substantially constrain [the merged firm's] price-increasing ability'" (quoting AD/SAT, Div. of Skylight, Inc. v. Associated Press, 181 F.3d 216, 228 (2d Cir. 1999))).
The parties in this case have fundamentally disagreed about whether certain products qualify as reasonable substitutes. Specifically, Defendants argue that generalist DSPs (as opposed to healthcare-specific DSPs), social media platforms, and endemic websites all provide alternative sources of HCP programmatic advertising. At a high level, Defendants' theory is that the proposed acquisition of Deeplntent would not allow IQVIAto raise prices because customers-such as pharmaceutical companies and their advertising agencies-could respond to a price increase by shifting their business both to other DSPs and to these alternative channels. Rather than pay a higher price for the merged firm's services, the argument goes, those customers would take their business to social media and endemic websites because they offer a comparable product.
The FTC maintains that these other forms of digital marketing are not reasonable substitutes for HCP programmatic advertising. It argues that social media and endemic websites do not offer the same level of functionality as HCP programmatic advertising and therefore are not meaningful competitive constraints. Doc. 287 (FOF) at 18-22. And the FTC contends that "[t]he existence of arguable functional substitutes to HCP Programmatic Advertising . . . does not affect the existence of a relevant market or submarket for the specialized versions of advertising." Id. (COL) at 6.
Courts consider both quantitative and qualitative evidence in denning the relevant product market. See, e.g., Peabody, 492 F.Supp.3d at 885; Sysco, 113 F.Supp.3d at 27. Here, the FTC relies on ordinary course documents-for example, internal memoranda, emails, and PowerPoint presentations-and other qualitative evidence to argue that HCP programmatic advertising is the relevant market. In addition, both parties have offered expert economic testimony to support their market definition arguments. The FTC's expert, Dr. Kostis Hatzitaskos, testified that HCP programmatic advertising is a relevant market. He tested a proposed market including Deeplntent, Lasso, PulsePoint, and several other DSPs that reported revenue from HCP programmatic advertising. Dr. Hatzitaskos did not include social media or endemic websites in the market he originally tested, concluding that they are not reasonable substitutes. Dr. Mark A. Israel, one of Defendants' experts, challenged Dr. Hatzitaskos's conclusions and opined that his market definition was too narrow.
After weighing all this evidence, the Court concludes that, for purposes of this section 13(b) proceeding, the FTC has met its burden with respect to market definition. At bottom, the Court's view is that social media and endemic websites are not reasonably interchangeable with the HCP programmatic advertising services provided by DSPs such as Deeplntent and Lasso. Testimony from industry participants, combined with ordinary course documents, indicates that programmatic has distinct characteristics and offers something meaningfully different than what is provided by those alternative channels. The Court's conclusion is also reinforced by the economic evidence that Dr. Hatzitaskos presented. Although Dr. Israel posed some valid criticisms of specific methods and inputs, the Court ultimately finds that the weight of the evidence tends to show that the FTC's candidate market is not impermissibly narrow. Accordingly, the FTC has carried its burden to "rais[e] some question of whether [HCP programmatic advertising] is a well-defined market." Whole Foods, 548 F.3d at 1037 (opinion of Brown, J.).
a. Brown Shoe Factors
The Court's analysis begins with the Brown Shoe factors. In Brown Shoe, the Supreme Court explained that the boundaries of a submarket "may be determined by examining such practical indicia as industry or public recognition of the submarket as a separate economic entity, the product's peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors." 370 U.S. at 325. The factors "are not rigidly applied," and the determination of the relevant market is ultimately "a matter of business reality-[] of how the market is perceived by those who strive for profit in it." Peabody, 492 F.Supp.3d at 892 (alteration in original) (citation omitted). Courts regularly rely on the Brown Shoe factors to help define the relevant product market. See, e.g., GenevaPharms., 386 F.3d at 496 ("Denning a submarket requires a fact-intensive inquiry that includes consideration of [the Brown Shoe indicia]."); Peabody, 492 F.Supp.3d at 892; see also Sysco, 113 F.Supp.3d at 27 (collecting cases). All the factors need not be satisfied for the Court to conclude that the FTC has identified a relevant market. See Staples, 970 F.Supp. at 1075 ("Since the Court described these factors as 'practical indicia' rather than requirements, subsequent cases have found that submarkets can exist even if only some of these factors are present.").
As noted above, courts analyzing market definition in section 13(b) cases have considered both qualitative evidence related to the Brown Shoe factors as well as quantitative evidence from economic experts. See, e.g., Peabody, 492 F.Supp.3d at 885; Sysco, 113 F.Supp.3d at 27. There is no uniform sequence in which courts approach this analysis. Compare, e.g., Peabody, 492 F.Supp.3d at 886 (analyzing the economic evidence first), with Sysco, 113 F.Supp.3d at 27 (analyzing the Brown Shoe indicia first). Here, the Court begins with the qualitative evidence and the Brown Shoe indicia before turning to the parties' economic evidence. In any event, the Court concludes that both types of evidence support the FTC's market definition.
Here, the Court analyzes the Brown Shoe factors with reference to the reasonably interchangeable substitutes proposed by Defendants-specifically, social media and endemic websites. In other words, the Court asks whether the programmatic advertising providers identified by the FTC offer a distinct product as compared to the substitutes identified by Defendants. This approach is consistent with how other courts have employed the Brown Shoe factors to inform their views of substitutability and the definition of the relevant market. See, e.g., Sysco, 113 F.Supp.3d at 27 (observing that courts have characterized the Brown Shoe indicia as "evidentiary proxies for direct proof of substitutability" (quoting Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 218 (D.C. Cir. 1986))); FTC v. CCC Holdings Inc., 605 F.Supp. 2d 26, 38 (D.D.C. 2009) (noting that the Brown Shoe indicia "can be applied to augment the analyses of interchangeability and cross-elasticity of demand"); see also United States v. Energy Solutions, Inc., 265 F.Supp.3d 415, 436 (D. Del. 2017) ("Factors for finding reasonable interchangeability 'include price, use, and qualities.'" (quoting Queen City Pizza, Inc. v. Domino's Pizza, Inc., 124 F.3d 430, 437 (3d Cir. 1997))).
The Court's analysis of the Brown Shoe factors shows that other channels of digital advertising are not readily interchangeable with HCP programmatic advertising. The first two factors-distinct characteristics and industry recognition-are the most illuminating in this case. The balance of the evidence reveals that HCP programmatic advertising has distinct characteristics that set it apart from channels such as social and endemic, and industry recognition of programmatic's distinctive features buttresses that conclusion.
i. Distinct Characteristics
First, the FTC asserts that HCP programmatic advertising has distinct characteristics because it offers "combined capabilities" and provides advertisers with "unparalleled inventory access, transparency, efficiency, and control." Doc. 287 (FOF) at 13-14. HCP programmatic advertising allows advertisers to deliver ads "across thousands of different publishers, determine which [HCPs] interact with the advertisements, and analyze whether those [HCPs] changed their prescribing behavior- all via 'a click of a button.'" Id. at 14 (citation omitted).
The FTC highlights several characteristics of programmatic advertising that distinguish it from advertising through social media and endemic websites. With respect to social media platforms, the FTC contends that they are not reasonably interchangeable with HCP programmatic advertising because social media sites are "walled gardens." Id. at 18. In essence, this means that when advertisers purchase ads on a social media site, their reach is limited to the users and data contained within the walls of that individual site. Tr. (Redacted); Tr. 802 (Lawson); Tr. (Redacted). As one witness explained, social media platforms "control all of the data and the content that exist within those walled gardens, and typically customers are not allowed to pull that data out." Tr. (Redacted). As a result, the FTC says, "advertisers cannot target HCPs or measure the effectiveness of an advertising campaign with [the] same granularity as in HCP programmatic advertising." Doc. 287 (FOF) at 19.
So too for advertising through endemic websites. According to the FTC, the "direct buys" available through endemic sites are not as efficient or cost-effective as HCP programmatic advertising. Id. at 21. Direct buys rely on a more labor-intensive, manual process of negotiating deals with publishers that is less efficient compared to programmatic advertising. Id. In addition, purchasing advertisements with an endemic website limits an advertiser to the fixed inventory on that website, which the FTC contrasts with the opportunity that HCP programmatic advertising provides to access inventory across myriad websites. Id. On the back end, moreover, programmatic advertising "enables faster aggregation and analysis of campaign metrics," which allows advertisers to adjust campaigns more quickly even while they are in progress. Id. at 21-22.
The Court finds that HCP programmatic advertising, as denned by the FTC, has distinct characteristics that differentiate it from the proposed substitutes identified by Defendants. Several witnesses testified about the differences between programmatic advertising and other media channels. As one advertising agency witness explained, programmatic advertising differs from social media "in the types of content that it delivers and the ad opportunities within whether it be video display or other opportunities." Tr. (Redacted). Another advertising agency executive confirmed that Facebook offers something "fundamentally different" from what a DSP like Deeplntent provides. Tr. (Redacted) (stating that "Facebook is a totally different business model" and that "on the physician side they don't offer the same type of experience on HCP"). And testimony from pharmaceutical companies likewise indicates that programmatic is a distinct form of advertising. A witness from one company stated that around 70% of (Redacted) division's HCP digital advertising budget is allocated to programmatic, 20% is allocated to paid search, and 10% is allocated to social. Tr. (Redacted). Programmatic receives the largest share, (Redacted) explained, because "[i]t's the channel where we have the most efficiencies, so we can get a lot more for our money with programmatic display." Tr. (Redacted).
Programmatic also differs from social media insofar as it allows advertisers to reach HCPs across the internet. Multiple witnesses explained that this feature distinguishes programmatic advertising from the "walled gardens" of social media sites. Tr. (Readcted). An advertiser seeking to target HCPs on Facebook, for example, would be "confined to the audience data that they have access to." Tr. (Redacted); see also Tr. (Redacted) (programmatic is different from Facebook because Facebook provides access only to what the platform offers for targeting, whereas with programmatic "the DSP has inventory from across hundreds if not thousands of sites"). As a result, the advertiser will not be able to reach an HCP that does not use Facebook. Tr. (Redacted). And even if the HCP does use Facebook, that individual HCP can be reached only while on the platform, even though an advertiser is "more likely to find those users across the open Internet many more times a day." Id. Programmatic advertising, by contrast, allows advertisers to reach HCPs more broadly across the web. Id. (explaining that "if you're confined to a social-media platform, you are not going to be able to scale your budget or reach users to be able to spend your marketing budget in the same way you would in the open web"); Tr. (Redacted) (compared to social media, programmatic "has a broader offering tied to a much larger set of publishers that offer different ad spaces, places, and advertising opportunities"); Tr. (Redacted) ("You can reach more HCPs with programmatic display than you can on a social program."); see also Tr. 603 (Paquette) (Deeplntent CEO agreeing that one purpose of HCP programmatic advertising is "to reach health care providers wherever they are on the internet across multiple media properties").
This limitation applies to endemic websites as well. Witnesses explained that advertising on endemic sites confines an advertiser to reaching HCPs on that site. Tr. (Redacted) (advertising on Medscape would be "direct advertising" rather than programmatic advertising because the agency would buy space, for instance, on a specific article); Tr. (Redacted) (observing a difference between Deeplntent and Medscape in that Medscape offers inventory on only its own website, while Deeplntent offers inventory on many websites, and characterizing the two as "different ways of buying advertising"); see also Tr. (Redacted) (noting that there has been "a big trend in the marketplace" toward non-endemic inventory-that is, "being able to find an [HCP] when they are just browsing the internet in general").
The data that advertisers can glean from advertising via social and endemic sites is also more limited. To be a "walled garden" means that a social media platform's data remains "within their one world" and is "closed off... to everybody else." Tr. (Redacted) (Redacted). Again, this characteristic distinguishes social media from programmatic. Tr. (Redacted) (explaining that programmatic is "a lot more fluid and open in terms of the data that is coming through"); Tr. (Redacted) (walled gardens such as Facebook "don't allow you to always take the data out in the form that you need it to do measurement or any kind of analysis that you might need for performance"); Tr. (Redacted) (asserting that advertisers cannot use the data generated from a platform like Facebook for measurement and that "if we can't measure it, we can't tell how it is performing, and that makes it really difficult for us to continue funding").
All told, this evidence suggests that programmatic advertising has distinct characteristics and is not reasonably interchangeable with social media platforms or direct buys on endemic websites. Tr. (Redacted) (an HCP advertising campaign would "definitely not" be as effective if an advertiser moved its entire spending with programmatic to other channels like paid search and social because "the reach would not be there always, so we would reach less HCPs because we have a budget constraint"); Tr. 805-06 (Lawson) (compared to direct buys on endemic websites, programmatic provides "a much more flexible real-time method of buying media than a direct buy with a fixed number of impressions on a set number of pages"); Tr. (Redacted) (compared to a direct buy on Medscape, an agency "can get more scale more quickly at a much cheaper cost by using programmatic to extend the reach, not to take the place of a direct buy, but to complement it"); see also Tr. 603 (Paquette) (agreeing that programmatic "offers an advantage over direct buys because it is not feasible to contract with hundreds of thousands of websites individually").
Defendants resist this conclusion, arguing that social media and endemic sites are reasonably interchangeable with programmatic advertising. Doc. 288 at 54-55. It is certainly true that some social media sites can be used to target HCPs with particular advertisements. Tr. (Redacted); Tr. 262 (Colarossi). Like DSPs, moreover, some social media platforms offer audience creation and campaign measurement services. Tr. 306 (Lin) (Doximity and Sermo have audience creation capabilities); Tr. 317 (Lin) (Facebook provides measurement tools for advertising campaigns). And the same is true of some endemic websites. Tr. 263 (Colarossi) (Medscape can target ads to HCPs).
As discussed, however, viewership on these social media platforms is confined within their walls, distinguishing this method of advertising from the kind of open internet access that DSPs provide through programmatic advertising. See also PX6504 (Hatzitaskos Reply Report) at 20-25 (summarizing third party testimony and other qualitative evidence indicating that social is distinct from programmatic); Tr. 331-32 (Lin) (agreeing that a Facebook ad would be confined "within the four walls of Facebook," though noting that the wall is "very large"). That distinction applies to endemic websites as well. See also PX6504 at 25-26; Tr. (Redacted) (ad agency executive noting that physicians might visit an endemic site only "once or twice a week" but use the internet "once or twice an hour," so "I can use programmatic to find them more times than waiting for them to go to the clinical sites once or twice a week"). Given the emphasis several witnesses placed on the reach that programmatic advertising provides, social media and endemic websites that limit advertisers to reaching HCPs only while they are using those sites do not provide a reasonably interchangeable substitute.
Defendants also rely on testimony indicating that agencies may shift portions of their budgets for particular campaigns away from DSPs to social or endemic if the latter will provide a better return on investment. One advertising agency executive testified that while a campaign is in progress, the agency will reallocate its budget among various HCP channels to maximize return on investment for its clients. Tr. (Redacted). For instance, it might move some of its budget to endemic websites from DSPs if it will result in a more favorable return on investment. Tr. (Redacted) (the agency will recommend reallocating from DSP to endemic if return on investment is higher, though "[s]ometimes there are barriers to doing that"). And during closing arguments, Defendants cited testimony from several advertisers purportedly indicating that they would not accept a DSP-imposed price increase but rather would take their business elsewhere. See IQVIA Closing Slides at 27-29; Tr. 1410-11 (IQVIA Closing); see also Doc. 288 at 29 ("[H]ealthcare companies and agencies testified that if a merged firm attempted to raise prices, they could and/or would simply shift their spend to other DSPs and/or platforms, including social and endemic").
But the fact that an agency might shift money around during a campaign does not establish that these alternative channels are substitutes for the distinct features that programmatic advertising provides. As the declaration of one pharmaceutical company executive explained, HCP programmatic advertising is "an important component of [the company's] overall advertising strategy, and it is not something the company would walk away from unless prices increased substantially." (Redacted). Defendants' examples, moreover, fail to demonstrate that customers would abandon DSPs providing programmatic advertising in favor of social media and endemic channels in the event of a price increase. One ad agency executive, for instance, stated that if Deeplntent were to raise prices, there is sufficient choice and optionality "such that agency or client partners would find another solution." PX0560 (Chase Dep.) at 30-31. But the alternatives he identified were other DSPs: PulsePoint and The Trade Desk. Mat 31. So the testimony does not necessarily suggest that advertisers would take their business to social media or endemic websites in response to a price increase. The same is true of another ad agency witness's testimony that, if the agency did not like the rates it was getting from a DSP, it either would negotiate or "would work with someone else, as long as it didn't sacrifice the services." Tr. (Redacted). Nothing in this answer demonstrates that the agency would take its business to social media or endemic sites instead of simply shifting to a comparable DSP. And given the same agency's testimony that programmatic provides advantages of scale and efficiency when compared to direct buys, Tr. (Redacted), it is hard to see how moving away from programmatic to social or endemic would not result in at least some sacrifice in services.
The same logic applies to testimony from another pharmaceutical company's witness that the company would reallocate some of its budget to other channels if the return on investment on programmatic was low. (Redacted) at 72. There is no indication that the company would move its entire budget away from programmatic; its testimony was simply that the company "can't rely on [programmatic] solely to drive our business." Id. (emphasis added).
Finally, Defendants note that some DSPs view social media platforms and endemic sites as competing for advertising dollars. An IQVIA executive, for example, testified that both general and healthcare-specific social media companies, as well as endemic sites, are "key competitors of Lasso." Tr. 263 (Colarossi); see also Tr. 527 (Field) (stating that Lasso competes against Medscape and Doximity for one ad agency's HCP advertising budget and is "evaluated by the same datasets"). Deeplntent's CEO, similarly, said Deeplntent has been competing with social media platforms since 2017, and he identified Meta (which includes Facebook) and Doximity as examples of specific competitors. Tr. 661 (Paquette).
To be clear, social media companies and endemic websites are competing with DSPs in a broad sense. An agency running an advertising campaign will not have an unlimited budget, so it must make decisions about how to allocate the advertising funds it has. But the fact that programmatic competes with these channels for advertising dollars in a broader market does not necessarily mean those channels are reasonably interchangeable substitutes that must be included in the relevant product market. See Peabody, 492 F.Supp.3d at 886 (although the record showed that coal competed with natural gas and renewables in a broader energy market, "the FTC has presented more than sufficient evidence that there is also a distinct competitive market among SPRB coal producers that satisfies the applicable criteria for market definition"); Staples, 970 F.Supp. at 1075 (even though a firm may be a competitor in the overall marketplace, that "does not necessarily require that it be included in the relevant product market for antitrust purposes"). Testimony from other DSPs, moreover, suggests that social media is not in fact a significant competitor. Tr. (Redacted) ("Social is completely different than what we are talking about with HCP programmatic campaigns."); Tr. 804-05 (Lawson) (stating that HCP programmatic advertising is not interchangeable with advertising on social media but rather is "a distinct and different method of advertising to health care professionals utilizing different data, utilizing different types of reporting and measurement"); Tr. 805 (Lawson) (generalist DSP AdTheorent is not competing with social media companies like Facebook, Sermo, and Doximity for programmatic media budgets, and RFPs from advertisers are directed specifically to programmatic open-internet advertising rather than social).
Aside from social media and endemic websites, Defendants have posited that Google is a reasonable substitute as well and should be included in the relevant market. Doc. 288 at 18-20; see also Tr. 1420-21 (IQVIA Closing). There is no dispute that Google qualifies as a generalist DSP, Doc. 287 (FOF) at 36; Doc. 288 at 18, and some generalist DSPs are included in the FTC's candidate market. But the parties have quarreled about whether Google engages in HCP programmatic advertising. The FTC suggests that Google does not compete in the market for HCP programmatic advertising (Redacted). Doc. 287 (FOF) at 36. Defendants respond that Google does offer HCP programmatic advertising and thus provides a competitive constraint. Doc. 288 at 18-20. Given the back-and-forth, a review of Google's testimony is warranted.
(Redacted)Google's Managing Director of Global Product and Sales Strategy, testified by deposition designations that were played during the evidentiary hearing. Much of his testimony centered on a declaration he submitted to the FTC in May 2023. PX0009. In that declaration, (Redacted) explained that one of Google's products, Display & Video 360 (DV360) is a DSP that "helps advertisers buy advertising space on digital platforms." PX0009-01. DV360 does not specialize in any particular category of advertising. Id. Advertisers using DV360 "can reach desired viewers using audience list targeting, which involves targeting the ads based on the characteristics of the user." Id. An advertiser may use Google's internally created audiences, or it can bring its own audience list. Id.
(Redacted) declaration noted that Google imposes some restrictions on advertisers. Specifically-and significant in the context of the candidate market here-Google "has a list of restricted audience categories for which no targeting is allowed, one of which is healthcare and medicine providers." Id. An advertiser may not target HCPs directly "if the advertising targeting relates to their role providing healthcare products or medicines." Id. In addition, Google has restrictions for advertising in "sensitive interest categories, including prescription medications and information about prescription medications." Id. (Redacted) stated: "To the extent that Deeplntent or Lasso allows personalized targeting of healthcare providers by NPI number, it is providing a service that Google does not (Redacted) provide. Google does not allow for personalized targeting of healthcare providers or for marketing prescription medication." PX0009-02. Google implemented these restrictions based on its users' privacy expectations. Id. (Redacted).
In his deposition testimony, (Redacted) confirmed that Google allows advertisers to run programmatic campaigns on DV360 and that advertisers can bring their own audiences to DV360. (Redacted) at 49-52. He clarified that an advertiser using Google "would be able to target an individual who is a healthcare provider; however, they wouldn't be able to target them with any advertising that is in a restricted or sensitive category." Id. at 149-50. For instance, a pharmaceutical company could run "a brand awareness campaign on the good deeds the pharmaceutical company does." Id. at 102; see also Id. at 182 (explaining that "you can have a first- or third-party list of individuals who are healthcare providers that allows you to reach those users with nonsensitive advertising categories such as a brand awareness campaign based on that [] user list").
As for the sensitive category restriction, (Redacted) stated that prescription medicines are a sensitive category and cannot be targeted to any audience list, healthcare providers or otherwise. Id. at 182-83; see also Id. at 189 ("[I]n no cases do we allow pharmaceutical and sensitive over the counter ads to be targeted to first party audiences."). He explained that "there may be the ability to provide personalized advertising for nonsensitive advertisements to user lists that are comprised of healthcare workers." Id. at 183. Ultimately, (Redacted) testified, "Google does not allow advertisers to target any healthcare provider or any person directly with advertising for prime products when the products and targeting relate to their role providing healthcare products or medicines." Id. at 213.
This testimony is not a model of clarity as to Google's programmatic offerings in the healthcare space. Still, the Court concludes that Google does not appear to be a reasonably interchangeable substitute for the services offered by healthcare-specific DSPs such as Deeplntent, or even for those provided by generalist DSPs such as The Trade Desk and AdTheorent, because it substantially limits the substance of what can be advertised to HCPs. (Redacted) stated that HCPs cannot be targeted using Google "when the products and targeting relate to their role providing healthcare products or medicines." Id. It is hard to see how this leaves much room for Google to serve as a substitute given that advertisements will almost always relate to an HCP's role providing such products or medicines. To be sure, there may be opportunities to target HCPs programmatically with a brand awareness campaign, but the Court fails to see how those opportunities provide a suitable replacement for the broader services that other DSPs in the candidate market offer.
The Court's conclusion is reinforced by the testimony of industry participants concerning Google. For instance, a witness from PulsePoint explained that he does not view Google as a significant competitor in the HCP space because "they tend to be very hesitant" on the HCP side of the market "for a number of reasons." Tr. 160 (Gerszke). An advertising agency witness, likewise, testified to his understanding that "what Google allows on their platform is extremely restricted from a pharmaceutical advertising space." Tr. (Redacted) . And a data company executive was "not aware of any health care professional campaigns that use Google DV360 as the DSP." Tr. (Redacted) .
Defendants point to their own evidence from industry participants to suggest that Google offers HCP programmatic advertising. One executive, for example, testified that his company has activated HCP audiences through an intermediary on Google's DSP and that, when it does so, Google is engaged in HCP programmatic advertising on a one-to-one basis (Redacted) at 25-26; see also Tr. 160 (Gerszke) (stating that Google has had "an HCP-related offering in the market" for "a long time"). But even if there are some programmatic advertising campaigns running on Google-and using IQVIAs data, Doc. 288 at 19-that does not mean Google provides a reasonably interchangeable substitute that could replace Deeplntent or another DSP's services in the event of a price increase. On the whole, the Court cannot conclude from the evidence in the record that customers would readily substitute Google for the DSPs in the candidate market-or that Google's policies would even permit such substitution for all the campaigns a customer would be interested in running.
Even if Google were included in the candidate market, it is far from clear that it would be a significant player based on the estimates of one of Defendants' experts. DX0076 (Israel Report) at 135 fig. 12 (estimating Google's revenue share (Redacted)).
Finally, Defendants contend that advertising agencies themselves provide an alternative to DSPs' programmatic offerings. Id. at 25-26. They point to evidence that these agencies are developing in-house technology for audience-building, campaign activation, and measurement. Id. These advancements purportedly allow agencies to negotiate directly with publishers rather than going through a DSP. Id. at 26.
Testimony from the agencies' witnesses, however, shows that the agencies do not provide a reasonably interchangeable substitute. One ad agency executive was asked directly why the agency purchases programmatic advertising services through DSPs "instead of just doing that yourselves." Tr. (Redacted) . The witness's response was instructive and tends to undermine Defendants' claim that agencies can simply provide programmatic services themselves: "We are not a technology company. We are a media agency. So, our concern is primarily about reaching audiences and we will utilize whichever platforms, networks, publishing partners allow us to get that reach. So, that is not what our clients hire us for." Id.; see also Tr. (Redacted) (another ad agency executive identifying different internal capabilities that DSPs have but agencies lack and agreeing that the agency is not "set up to do the same thing as a third-party DSP like Deeplntent").
All told, there is sufficient factual evidence in the record to demonstrate that programmatic advertising has distinct characteristics to differentiate it from other channels, such as social media and endemic, that Defendants have proposed as reasonably interchangeable alternatives.
ii. Industry Recognition
Next, the FTC argues that industry participants recognize HCP programmatic advertising as distinct. Doc. 287 (FOF) at 15. As was partially discussed in the previous section, many industry participants view programmatic as distinct from social, endemic, and other channels of digital advertising. To begin, DSPs confirmed that they see those channels as separate. One witness testified that PulsePoint does not view social media platforms and endemic sites as competitors because they are planned as part of a separate budget when agencies are planning advertising campaigns. Tr. 153-55 (Gerszke) (social media platforms do not appear in RFPs, and PulsePoint does not view them as competitors); Tr. 155-56 (Gerszke) (PulsePoint does not view direct buys on endemic sites as competing with its programmatic offering). A witness from AdTheorent, likewise, explained that "[t]here are budgets and teams within ad agencies and brands who are responsible for social media campaigns, and then there are teams that are responsible for programmatic or open-web, open-Internet budgets and campaigns." Tr. 803 (Lawson).
Similarly, one advertising agency executive characterized social media sites as "different types of partners" and testified that the agency does not include those sites when they are sending out RFIs or RFPs to evaluate programmatic advertising partners. Tr. (Redacted) Separate teams within agencies manage purchases for social media.Tr (Redacted) (. Agencies primarily purchase advertising on social media platforms through those specific platforms, and that process is managed by the agencies' social media team. Tr. (Readcted). The agency executive explained that Facebook and
Instagram "are just different owners internally" and that they were not included in its RFI and RFP. Id. In the same vein, the agency did not include Sermo because "we saw them as a partner we primarily engaged through our general media team where we do direct RFPs and not as a programmatic platform." Tr. (Redacted)
The FTC also points to ordinary course documents in which Defendants referred to HCP programmatic advertising as a distinct market and acknowledged that there are three leading healthcare-focused DSPs. Doc. 287 (FOF) at 15. Courts regularly consider ordinary course documents when denning the relevant market. See, e.g., United States v. H&RBlock, Inc., 833 F.Supp. 2d 36, 52 (D.D.C. 2011) ("When determining the relevant product market, courts often pay close attention to the defendants' ordinary course of business documents."); see also Peabody, 492 F.Supp.3d at 894-95 (ordinary course documents supported defendants' testimony that coal producers viewed other fuels as competitors).
There are several internal documents in which Defendants recognized HCP programmatic advertising as a distinct market. In one notable example, an IQVIA executive wrote that "[Deeplntent] has the #1 position for Healthcare platform[s] - [PulsePoint] 2nd, and [Lasso] 3rd." PX1026-01. The next day, another IQVIA executive sent an email framing the "positives" of both Deeplntent and Lasso in connection with the potential acquisitions, and he included a chart stating that, with respect to "Market Penetration," "[Deeplntent] is #1 and [PulsePoint] #2," while "[Lasso] is #3 position." PX1296-01. It is hard to understand what these characterizations could be referring to if, for instance, Doximity and Medscape, along with their substantial revenues, are included as competitors in the relevant market. Cf. Tr. 1048 (IQVIA Closing) ("Either one of [Doximity or Medscape] alone is far larger than the entire market that the FTC has proposed here."); IQVIA Closing Slides at 22 (pie chart indicating that the combined revenue of Doximity and Medscape would more than triple that of all the DSPs in the FTC's market). In its internal projections, moreover, Deeplntent estimated that it would have 51% of "HCP Market Share" in 2022 and 60% in 2023. PX2502-05. Again, those numbers are not at all consistent with Defendants' intimation of a market in which Doximity and Medscape have such an enormous presence.
The author of the email testified that he was not "trying to represent the full market in this email" but rather was "comparing their revenues with each other." Tr. 282 (Lin). The suggestion that this email did not have anything to do with the market generally strikes the Court as implausible given that the beginning of the sentence states: "I would go stronger with the market penetration .. . ." PX1026-01.
As the Court details more fully below in the section on competitive effects, there is also substantial evidence in the record showing that Deeplntent and Lasso have focused intensely on one another, along with PulsePoint, as key competitors. That evidence provides additional support for the market proposed by the FTC, which includes those three firms as the clear leaders. See Merger Guidelines § 4 ("Evidence of competitive effects can inform market definition, just as market definition can be informative regarding competitive effects.").
Defendants take issue with the FTC's reliance on ordinary course documents. They argue that many of the documents are outdated and were created before IQVIAs planned acquisitions of Deeplntent and Lasso. Doc. 288 at 60. In Defendants' view, courts should be skeptical of the "anecdotal speculation" present in these "informal chat transcripts and emails." Id. (citation omitted). On this point, Defendants cite two cases that do not appear to engage in the evaluation of ordinary course documents. See Aerotec Int'l, Inc. v. HoneywellInt7, Inc., 836 F.3d 1171, 1175 (9th Cir. 2016) (stating in general terms that "anecdotal speculation and supposition are not a substitute for evidence"); California v. Sutter Health Sys., 84 F.Supp.2d 1057, 1080 (N.D. Cal. 2000) (declining to credit testimony of witnesses who seemed to be "testifying based on pure conjecture"). No one would disagree with the proposition that "speculation and supposition" cannot substitute for evidence, but that does nothing to refute the well-established reliance on ordinary course documents to inform market definition.
Defendants also argue that informal comments made in a business environment "carry little value in denning an antitrust market." Doc. 288 at 62. They cite AD/SAT v. Associated Press, 920 F.Supp. 1287 (S.D.N.Y.1996), a Sherman Act monopolization case in which the court noted that one party's internal documents referenced officers' hope to "capture" the market. Id. at 1297 n.7. The court explained that "[f]or antitrust purposes, ... the relevant market is determined by reasonable interchangeability, as evidenced by cross-elasticity of demand and supply, not by laymen's comments made in a competitive business environment." Id.; see also Nobel Sci. Indus, v. Beckman Instruments, 670 F.Supp. 1313, 1318-19 (D. Md. 1986) ("[T]he fact that a company may refer to a 'market' does not necessarily mean that its reference will be to a market for purposes of the Sherman Act.").
The Court agrees that references to a "market" in internal documents are not themselves dispositive in delineating the boundaries of the relevant antitrust market. Again, however, courts in the section 13(b) context and elsewhere have repeatedly held that ordinary course documents can and should play a role in analyzing competitive dynamics and evaluating whether certain products qualify as reasonable substitutes that must be included in the market. See, e.g., Peabody, 492 F.Supp.3d at 894 (testimony about competition from other fuel sources was "reflected pervasively in Defendants' ordinary course documents"); United States v. Aetna Inc., 240 F.Supp.3d 1, 26 (D.D.C. 2017) (ordinary course documents demonstrated where the merging parties focused "most of their competitive efforts"); H&R Block, 833 F.Supp.2d at 53 (ordinary course documents provided "strong evidence" of the relevant product market); cf. United States v. Waste Mgmt, Inc., 743 E2d 976, 980 (2d Cir. 1984) (internal "Budget Questionnaire" that required company to list major competitors informed conclusions as to market definition). And in this case, the FTC has not relied solely on ordinary course documents to establish the relevant market. Defendants' internal references to their understanding of the market for HCP programmatic advertising simply provide additional evidence to support the Court's overall conclusion that the FTC's proposed market is not unduly narrow. See also Whole Foods, 548 F.3d at 1045 (Tatel, J., concurring in the judgment) (noting that evidence of'"industry or public recognition of the submarket as a separate economic' unit matters because we assume that economic actors usually have accurate perceptions of economic realities" (quoting Rothery Storage, 792 F.2d at 218 n.4)).
Defendants further contend that RFPs are not a reliable source of evidence here because they do not reflect market realities. Doc. 288 at 61. Defendants point to United States v. Energy Solutions, 265 F.Supp.3d 415 (D. Del. 2017), where an expert's market share calculations relied in part on RFPs awarded. Id. at 440 n. 16. The court observed that RFPs "can indicate future competition." Id. But in that particular case, the RFP metric produced a result-specifically, a post-merger monopoly-that did "not best reflect market realities." Id. Here, the RFP evidence on which the FTC relies tends to confirm that HCP programmatic advertising is categorized separately from other channels of advertising. Tr. 154 (Gerszke) (social media platforms "do not show up in RFPs"); Tr. 805 (Lawson) (RFPs are directed to "programmatic open-Internet advertising and not social or search"); see also Doc. 287 (FOF) at 27-28, 43, 45 (summarizing responses to RFIs and RFPs in which Deeplntent and Lasso identified each other as primary competitors). That is fully consistent with the picture of the market presented by numerous industry participants, such as healthcare companies and ad agencies, who view HCP programmatic advertising as distinct.
Next, Defendants maintain that although internal documents may be helpful in determining which firms should be included in the relevant market, they are less useful in determining which firms should be excluded from that market. Doc. 288 at 61. They rely on FTC v. CCC Holdings Inc., 605 F.Supp.2d 26 (D.D.C. 2009), but that case tends to support the FTC's position. The court considered two products used by automobile insurers to calculate "total loss." Id. at 33. One option, "the Books," provided reports based on local and regional values. Id. An alternative option, "total loss software systems"-known as "TLV" products-included comprehensive databases of vehicle sales information from numerous sources. Id. In its market definition analysis, the court noted that "[m]ost insurance companies do not view the Books as an adequate substitute for TLV products." Id. at 41. And the defendants' internal documents made little reference to the Books when describing the competitive landscape. Id. The court concluded, therefore, that TLV software products represented a relevant product market and that the Books were not a viable substitute. Id. at 43. In doing so, the court noted
the apt warning that "separate markets are not indicated by documents within A firms that are preoccupied with other A firms. After all, a given producer of A cannot charge more than other A firms and thus may focus entirely on them even though a hypothetical monopolist of product A firms would focus entirely on the price of a close substitute B."Id. at 43 n.18 (citation omitted). But the court went on to say: "If the FTC were relying solely on Defendants' documents that evidence might be insufficient. However, various other industry participants from all sides share the view that TLV and the Books are not interchangeable." Id. So too here. The FTC is not relying exclusively on Defendants' internal documents to show that HCP programmatic advertising is not reasonably interchangeable with channels like social media and endemic. Instead, the documents provide another data point to reinforce the testimony of various industry participants that the products are not readily substitutable. See Peabody, 492 F.Supp.3d at 894-95 (crediting ordinary course documents where they supported witness testimony about whom defendants took to be their competitors).
Lastly, while criticizing the FTC's reliance on ordinary course documents, Defendants simultaneously argue that internal IQVIA documents reveal that it has "long understood the market to be much broader than the three healthcare-specific DSPs on which the FTC has focused, with many players all competing for the same dollars." Doc. 288 at 61 & n.376. Viewed against the balance of the record, however, these documents are far too general to support Defendants' assessment of the relevant antitrust market. See Aetna, 240 F.Supp.3d at 26 ("limited number of documents" cited by defendants were "too few and too general to carry much weight"). No one would dispute, for instance, that there are many companies "vying for Healthcare & Pharma Brand budget." DX0083-17 (displaying logos of several companies in categories such as "All Industry DSPs," "Healthcare Specific DSPs," "Healthcare Agencies," "Social & Walled Gardens," and "Search"). But the fact that many companies are competing in a broad market for advertising dollars does not prove that the FTC's proposed market here is unduly narrow. Cf. United States v. Bazaarvoice, Inc., No. 13 Civ. 00133 (WHO), 2014 WL 203966, at *71 (N.D. Cal. Jan. 8, 2014) ("[T]he mere fact that Bazaarvoice, Amazon, Google and others all operate in the broader eCommerce space, does not mean that they are all competitors for antitrust purposes and should be placed in the same product market."). Instead, the Court must focus on "the narrowest market within which the defendant companies compete that qualifies as a relevant product market." Peabody, 492 F.Supp.3d at 886. Industry recognition of programmatic's distinct features supports the conclusion that the narrowest relevant market in this case need not-and does not- include social media and endemic sites.
iii. Distinct Pricing
On the next Brown Shoe factor, the FTC contends that HCP programmatic advertising features distinct pricing. Doc. 287 (FOF) at 16. Customers purchase advertisements based on the "CPM," or "cost per mille," which refers to the cost to purchase 1,000 impressions. Doc. 284 at 2; Tr. (Redacted) ("We buy a thousand impressions for an agreed upon dollar figure."). There is evidence indicating that HCP programmatic advertising is generally more expensive than DTC programmatic advertising. Doc. 287 (FOF) at 16; Tr. (Redacted) (cost of placing an HCP advertisement is more expensive than cost of placing a DTC advertisement); (Redacted) at 83 ("Typically, healthcare practitioner programmatic is more expensive than patient or direct-to-consumer-related programmatic").
More relevant to the Court's analysis, however, is whether there are significant pricing differences for programmatic as compared to social media and endemic. While the record is not as fully developed on this point, there is some evidence tending to show that programmatic's pricing is at least somewhat distinct. Again, testimony from advertising agencies is instructive. One advertising agency executive explained that, on endemic websites, programmatic ad buys are more expensive than direct buys (which involve negotiating directly with the publisher to place an advertisement). Tr. (Redacted). That agency encourages its customers to buy HCP programmatic advertising, even though it tends to be more expensive than direct buys, due to the benefits that programmatic offers. Tr. (Redacted)? See also Tr. (Redacted) (discussing benefits programmatic provides such as the flexibility to increase or decrease budgets).
Another ad agency witness testified that HCP programmatic advertising provides the opportunity to target HCPs "as efficiently as possible, the lowest cost that we would like to pay and get the highest return." Tr. (Redacted). To some degree, all channels provide efficiencies as agencies attempt to get the best possible results for the lowest possible cost. Tr. (Redacted). But "there is a great amount of control on the pricing through programmatic means." Tr. (Redacted); see also Tr. (Redacted) (programmatic offers "fluidity in what it allows you to do where you can fluctuate your price based on what you are bidding"). Accordingly, this factor provides at least moderate support for the FTC's proposed product market.
iv. Distinct Customers
The FTC further asserts that HCP programmatic advertising serves distinct customers. Doc. 287 (FOF) at 17. HCP programmatic advertising, the FTC says, is distinct from traditional advertising because it "requires the precise delivery of advertisements to the targeted professionals on an individualized, one-to-one basis," along with "being able to measure a campaign's effectiveness by evaluating whether a specifically targeted HCP has changed his/her prescribing behavior subsequent to viewing the ads." Id. The FTC thus emphasizes the importance of the activation and measurement stages to the pharmaceutical companies who serve as customers in this industry. Id.
This factor does not provide much support for the FTC's proposed market. At best, the FTC's arguments tend to show that certain customers are particularly interested in HCP programmatic advertising as compared to more traditional forms of advertising- print or email campaigns, for example. But this does not show that programmatic is distinct from alternatives such as social media or endemic. And in fact, the record tends to reflect that the same customers who purchase programmatic advertising regularly purchase advertising through those other channels as well. Tr. (Redacted) (ad agency witness discussing use of various channels to advertise to HCPs).
The FTC's reliance on testimony from The Trade Desk is not helpful for similar reasons. Doc. 287 (FOF) at 17. The Trade Desk's witness testified that the programmatic advertising needs of pharmaceutical clients are generally distinct from the needs of other clients of The Trade Desk. Tr. 735 (Harper). That assertion suggests that customers advertising in the healthcare industry have different needs from clients advertising in other industries, such as automotive or consumer products. But it does not help the FTC demonstrate that the customers advertising through programmatic are distinct from the customers advertising through social media or endemic websites.
v. Specialized Vendors
Finally, the FTC suggests that HCP programmatic advertising is performed by specialized vendors. Doc. 287 (FOF) at 17. It points to testimony from industry participants suggesting that "generalist DSPs lack the experience, technology, and/or data necessary to compete robustly with healthcare-specific DSPs." Id. at 18.
It is not clear that this factor offers much support for the FTC either. To be sure, there is plenty of evidence in the record showing that healthcare-focused DSPs have certain advantages as compared to generalist DSPs. Tr. 157-58 (Gerszke) (generalist DSPs lack capabilities that are "important for executing HCP digital marketing at a competitive price and scale"); PX2582-01 (Deeplntent blog post explaining that companies built healthcare-specific DSPs "because generalist DSPs, which serve a wide range of advertising verticals, don't necessarily meet a healthcare marketer's complex needs"). But the FTC fails to explain how this differentiation between generalist and healthcare DSPs supports its proposed market, which includes some generalist DSPs. The FTC has not argued that only healthcare-focused DSPs, as opposed to generalist DSPs, provide HCP programmatic advertising. So while healthcare DSPs may have an edge in the marketplace, the FTC has not offered evidence showing that only specialized vendors provide programmatic offerings.
vi. The Court's Conclusions
For the reasons discussed above, the Brown Shoe indicia, on balance, support the FTC's position that HCP programmatic advertising is a relevant product market for antitrust purposes. Most critical to this conclusion is the Court's finding that social media and endemic websites-and Google-are not reasonably interchangeable alternatives for the programmatic offerings in the FTC's candidate market. While the testimony on this subject was not uniform, the Court ultimately finds credible the testimony of several witnesses who explained that programmatic has distinct characteristics as compared to social and endemic and that industry participants recognize the differences.
Two elements of the legal standard are worth emphasizing here. First, as the Court has already discussed, the burden on the FTC in this litigation is lower than the burden it will face in the administrative proceeding. The FTC does not need to prove the merits of its case to obtain a preliminary injunction under section 13(b). As that standard pertains to the market definition inquiry, then, it is "not necessary at this point" for the FTC to prove the existence of the HCP programmatic advertising market. See Whole Foods, 548 F.3d at 1041 (opinion of Brown, J.). The FTC need only raise serious and substantial questions that are suitable for consideration by the Commission itself in the first instance. The FTC has cleared that bar.
Second, it is important to reiterate that market definition requires an attempt to identify the narrowest possible market. In evaluating reasonable interchangeability, "the mere fact that a firm may be termed a competitor in the overall marketplace does not necessarily require that it be included in the relevant product market for antitrust purposes." Sysco, 113 F.Supp.3d at 26 (quoting Staples, 970 F.Supp. at 1075). And as courts have recognized, a "broad product market. . . may contain smaller markets . . . which themselves 'constitute [relevant] product markets for antitrust purposes.'" Peabody, 492 F.Supp.3d at 885 (alteration in original) (quoting Brown Shoe, 370 U.S. at 325). In other words, the existence of a larger market within which two products compete does not necessarily mean that they are reasonably interchangeable substitutes for one another. In this case, there is undeniably a broader market for digital healthcare advertising in which programmatic, social media, and endemic websites all participate. But "the viability of such additional markets does not render the one identified by the government unusable." United States v. Bertelsmann SE & Co., 646 F.Supp.3d 1, 28 (D.D.C. 2022). Traditional methods of reaching HCPs such as email or print advertising might also conceivably be within some broader market, yet no one has suggested that those options should be included in the relevant market here.
For all these reasons, the Brown Shoe factors and the reasonable interchangeability analysis support the FTC's definition of the relevant product market.
b. Hypothetical Monopolist Test
The Court turns next to the quantitative evidence supporting the FTC's proposed market. In particular, the FTC argues that the hypothetical monopolist test (FDVIT) confirms its assertion that HCP programmatic advertising is a relevant market for antitrust purposes. Doc. 287 (FOF) at 22-26.
The FDVIT is "commonly used in antitrust actions to define the relevant market." Peabody, 492 F.Supp.3d at 886 (quoting FTC v. SanfordHealth, 926 F.3d 959, 963 (8th Cir. 2019)). The 2010 Horizontal Merger Guidelines, used by the Department of Justice (DO J) and the FTC to evaluate potential mergers, set out the methodology of the FDVIT.The Guidelines explain:
The Merger Guidelines are not binding on this Court but have been described as "a helpful tool, in view of the many years of thoughtful analysis they represent, for analyzing proposed mergers." United States v. Anthem, Inc., 855 F.3d 345, 349 (D.C. Cir. 2017); see also, e.g., AMR Corp., 2023 WL 2563897, at *3 n.2. The Court also notes that on December 18, 2023, after briefing and argument in this case had concluded, the DOJ and the FTC jointly issued the revised 2023 Merger Guidelines. Because the parties did not have the opportunity to address whether the final version of the new guidelines would have any material effect on the analysis here, the Court considers only the 2010 Merger Guidelines.
The hypothetical monopolist test requires that a product market contain enough substitute products so that it could be subject to post-merger exercise of market power significantly exceeding that existing absent the merger. Specifically, the test requires that a hypothetical profit-maximizing firm, not subject to price regulation, that was the only present and future seller of those products ("hypothetical monopolist") likely would impose at least a small but significant and non-transitory increase in price ("SSNIP") on at least one product in the market, including at least one product sold by one of the merging firms.
Merger Guidelines §4.1.1; see also Peabody, 492 F.Supp.3d at 886; Sysco, 113 F.Supp.3d at 33. If the monopolist could profitably impose such a price increase, then a relevant product market exists for antitrust purposes. Peabody, 492 F.Supp.3d at 886 (citing Merger Guidelines §4.1.1). (Federal agencies typically use a price increase-or SSNIP-of 5% when analyzing prospective mergers. Id. (citing Merger Guidelines § 4.1.2).) But if enough consumers would be able to substitute away from the hypothetical monopolist's product to another product, thereby making the price increase unprofitable, then the relevant market cannot be limited to the hypothetical monopolist's product and must also include the substitute products. Sysco, 113 F.Supp.3d at 33; see also H&R Block, 833 F.Supp.2d at 52.
In this case, then, the HMT inquiry asks: If a single firm controlled the entire HCP programmatic advertising market, as denned by the FTC, could it profitably impose a price increase? Or would that price increase result in customers moving enough of their business to other alternatives to make the price increase unprofitable? If the price increase would be profitable, then the FTC has denned the relevant product market accurately. If it would not be profitable, then the market is broader than the FTC has contended. Cf. Sysco, 113 F.Supp.3d at 34.
i. Dr. Hatzitaskos 's Analysis
The FTC presented expert testimony from Dr. Hatzitaskos, who has a Ph.D. in economics and serves as Senior Vice President at Cornerstone Research, an economics litigation consulting firm. PX6500 (Hatzitaskos Report) at 7. On the issue of market definition, Dr. Hatzitaskos opined that HCP programmatic advertising constitutes a relevant antitrust market. Id. at 11-12. He included in his candidate market "all market participants that shared revenue through discovery," as well as "anybody that IQVIA reported as [a] top customer of their HCP data." Tr. 832 (Hatzitaskos). The proposed market in his initial report included Deeplntent, Lasso, PulsePoint, and other DSPs providing HCP programmatic advertising. PX6500 at 123. Dr. Hatzitaskos did not include social media platforms in the market he originally tested. Tr. 845 (Hatzitaskos). Considering all the available evidence, he testified, these platforms did not "seem to be a reasonable substitute, a meaningful constraint." Id.; see also PX6504 at 20-25. Nor did he include endemic websites such as Medscape, concluding again that the evidence did not indicate that they were "meaningful competitive constraints." Tr. 847 (Hatzitaskos); see also PX6504 at 25-26.
The firms Dr. Hatzitaskos included in his initial report were: Deeplntent, Lasso, PulsePoint, Proclivity, AdTheorent, TI Health, eHealthcare, Amobee, Nativo, The Trade Desk, Viant, Doceree, and Medicx. PX6500at 123, Exhibit 13. In his reply report, he added Healio. PX6504 at 36-37 & Exhibit R-l. Dr. Hatzitaskos also amended several figures in his reply report to reflect updated discovery. Id. at 35-37. In a "conservative thought experiment," moreover, he included additional firms proposed by one of Defendants' experts, even though he found the evidence to be "inconsistent with the additional entities having competitive significance that is proportional to their estimated revenues." Id. at 38. The additional firms were: Vldeoamp, Acuity Ads, Stackadapt, Basis, Vericast, Facebook, Sermo, X (formerly known as Twitter), VDX.tv, and Linkedln. Id. at 43, Exhibit R-2. These additions, Dr. Hatzitaskos concluded, did not "meaningfully change the broader picture of industry revenues." Id. at 42.
Consistent with the case law discussed above, Dr. Hatzitaskos explained that market definition is "not an exercise in just listing every potential competitor." Tr. 837 (Hatzitaskos). Rather, the goal is to determine "who are the most important competitors, who are the most important for competitive constraints." Id. Denning the market as narrowly as possible allows for "a better prediction about competition." Tr. 844-45 (Hatzitaskos); see also Merger Guidelines § 4 ("Denning a market broadly to include relatively distant product or geographic substitutes can lead to misleading market shares. This is because the competitive significance of distant substitutes is unlikely to be commensurate with their shares in a broad market.").
Based on the qualitative evidence he evaluated, Dr. Hatzitaskos concluded that "industry participants recognize HCP programmatic advertising as a distinct product." PX6500 at 100-01; see also Id. at 100-115 (summarizing qualitative evidence indicating that HCP programmatic advertising constitutes a relevant product market). He then used the HMT to confirm that HCP programmatic advertising is a relevant antitrust market. Id. at 115. Dr. Hatzitaskos relied on two methods to apply the HMT: (1) critical loss analysis and (2) merger simulation. Id.; Tr. 841 (Hatzitaskos). Under both approaches, he concluded that the HMT was satisfied for the market he tested.
First, critical loss analysis asks how many customers the hypothetical monopolist would have to lose to alternatives outside the market for the price increase to be unprofitable. Tr. 841-42 (Hatzitaskos); see also FTC v. Swedish Match, 131 F.Supp.2d 151, 160 (D.D.C. 2000) (describing critical loss as "the largest amount of sales that a monopolist can lose before a price increase becomes unprofitable"). Dr. Hatzitaskos estimated that a 5% price increase for Deeplntent would result in a critical loss of 10.6%, meaning that the hypothetical monopolist would need to regain 10.6% of the customers switching away from Deeplntent. PX6500 at 117 & Exhibit 10. He also estimated that a 10% price increase for Deeplntent would result in a critical loss of 21.2%. Id. For both calculations, he relied on a margin estimate of 47.3% for Deeplntent. Id.; see also Id. at 87 & Exhibit 6 (discussing margin calculations).
Dr. Hatzitaskos then compared the critical loss figures to an estimate of the aggregate diversion ratio. Id. at 117. "The aggregate diversion ratio for any given product represents the proportion of lost sales that are recaptured by all other firms in the proposed market as the result of a price increase." H&R Block, 833 F.Supp.2d at 63. These sales remain within the proposed market and thus are not lost to the hypothetical monopolist. Id; PX6500 at 116. If the aggregate diversion ratio to products within the proposed market exceeds the critical loss threshold, then a price increase would be profitable for the hypothetical monopolist. H&R Block, 833 F.Supp.2d at 63; PX6500 at 116.
Here, Dr. Hatzitaskos used 79.4% as an estimate of the aggregate diversion ratio. PX6500 at 118 & Exhibit 11. That figure was based on his analysis of actual customer choices, which relied on Deeplntent's internal "win/loss data" and campaign data from Lasso and PulsePoint. Id. at 118; see also Id. at 78-84 (analyzing customer choices). He found that 79.4% of customers who considered but did not choose Deeplntent ended up choosing Lasso or PulsePoint. Id. at 82. Dr. Hatzitaskos characterized this figure as a conservative estimate of the aggregate diversion ratio because the hypothetical monopolist would control all providers of HCP programmatic advertising in the candidate market rather than just those three firms. Id. at 118.
To reiterate, the ultimate test is whether the aggregate diversion ratio is higher than the critical loss; if it is, then the candidate market passes the HMT. Id; see also H&R Block, 833 F.Supp.2d at 63. The aggregate diversion ratio was 79.4%, while the critical loss was either 10.6% (based on a 5% price increase) or 21.2% (based on a 10% price increase). PX6500 at 118 & Exhibit 11. In both cases, then, the aggregate diversion ratio exceeded the critical loss by a wide margin and thus the HMT was satisfied. Id.
Second, Dr. Hatzitaskos conducted a merger simulation analysis. A merger simulation evaluates how the hypothetical monopolist would change prices after the merger and whether any price increase imposed on the merging firms' products would be more than 5%. Tr. 842 (Hatzitaskos). Dr. Hatzitaskos's analysis considered a hypothetical merger of Deeplntent, Lasso, and PulsePoint. PX6500 at 118. He asked whether, in those circumstances, the hypothetical monopolist would raise the price of either Deeplntent or Lasso by at least 5%. Id. at 118-19. Dr. Hatzitaskos found that the hypothetical monopolist would increase prices by more than 43%-well above the 5% threshold. Id. at 119-120 & Exhibit 12; see also Tr. 843 (Hatzitaskos).
According to Dr. Hatzitaskos, this scenario again represented a conservative approach because the hypothetical monopolist would control additional providers of HCP programmatic advertising beyond those three firms. PX6500 at 118 n.402.
In sum, Dr. Hatzitaskos found that "all of the evidence strongly confirms that HCP programmatic advertising constitutes a relevant antitrust market." PX6500 at 119.
ii. Dr. Israel's Analysis
Defendants challenged Dr. Hatzitaskos's conclusions through the expert testimony of Dr. Israel, who holds a Ph.D. in economics and serves as Senior Managing Director at Compass Lexecon, an economic consulting firm. DX0076 (Israel Report) at 10. Dr. Israel opined that the FTC's proposed market "is overly narrow and defines away important competition." Tr. 1035 (Israel).
Dr. Israel criticized both of Dr. Hatzitaskos's implementations of the HMT. First, with respect to the critical loss analysis, Dr. Israel stated that Dr. Hatzitaskos's aggregate diversion ratios were based on "fundamentally flawed" win/loss data. DX0076 at 91. That data, Dr. Israel asserted, was limited to Deeplntent RFPs and thus did not include any substitution from agencies that decided to spend on social or other alternatives rather than sending an RFP to Deeplntent. Tr. 1061 (Israel); DX0076 at 92-93. Furthermore, the win/loss data failed to account for the tendency of agencies to work with multiple advertising platforms at the same time. Tr. 1061 (Israel); DX0076 at 93-94. If Deeplntent were to lose an RFP, Dr. Israel reasoned, "it's very likely, from what we have heard, that five or six different platforms get some of that money." Tr. 1061 (Israel). But Dr. Hatzitaskos's analysis treated any match in the win/loss data as a 100% substitution to either Lasso or PulsePoint. Tr. 1061-62 (Israel). Dr. Israel cited one $600,000 campaign that Deeplntent lost to Lasso and that Dr. Hatzitaskos treated as a 100% match with Lasso-when in fact, according to Dr. Israel, Lasso received only $12,000 of that total budget. Tr. 1062 (Israel); DX0076 at 101.
The critical loss analysis was also flawed, in Dr. Israel's view, because Dr. Hatzitaskos assumed that the profit margin for Deeplntent and Lasso was "roughly 50 percent." Tr. 1062-63 (Israel). That figure was too high, Dr. Israel asserted, because it accounted only for data and media costs while omitting the costs of running the platform. Tr. 1063 (Israel); see also Id. ("So the costs you have to build in have to be enough costs that somebody would actually set a price at that level. There is no indication in any industry like this that people would set a price where all they cover is their data and their media, they don't cover their platform or anything.").
As for the merger simulation, Dr. Israel concluded that Dr. Hatzitaskos's model required market shares as an input and thus was based on "entirely circular logic" that assumed the proposed market was correct. DX0076 at 89-90. This share-based model, Dr. Israel testified, essentially "says substitution is determined by your market share." Tr. 1059 (Israel). In other words, the analysis "doesn't measure substitution. It assumes it is equal to share." Id. According to Dr. Israel, Dr. Hatzitaskos assumed that the market share for any firm outside his candidate market-that is, any firm other than the DSPs he included-was zero. Id. In Dr. Israel's view, this approach failed to establish the proposed market because "[i]t assumes the market by saying if you are not in the market, your share is zero and therefore there is no substitution." Tr. 1060 (Israel).
iii. The Court's Conclusions
Having carefully weighed the testimony of both experts, the Court finds that Dr. Hatzitaskos's conclusions further support the FTC's assertion that HCP programmatic advertising is the relevant product market.
At the outset, Defendants argue that Dr. Hatzitaskos's HMT analyses are legally defective because he asked whether the hypothetical monopolist could profitably increase the price of one product rather than all products in the market. Doc. 288 at 56. According to Defendants, "[i]t makes no sense to evaluate substitutability within the candidate market, because the whole point of the HMT is to determine whether firms outside the candidate market competitively constrain pricing." Id. By this logic, instead of asking whether the hypothetical monopolist could profitably increase Deeplntent's prices, Dr. Hatzitaskos should have asked whether the hypothetical monopolist could profitably increase the prices of all the firms in the proposed market.
The FTC's position, by contrast, is that the test asks whether the hypothetical monopolist "could raise prices by 5 percent on one product." Tr. 1382 (FTC Closing) (emphasis added); see also Doc. 287 (COL) at 12 (framing the inquiry as whether a hypothetical monopolist could impose a price increase "on at least one product of the merging parties in the candidate market"). The question then becomes whether the hypothetical monopolist would lose sales to options outside the candidate market. Tr. 1382-83 (FTC Closing). Potential alternatives such as social media and endemic, therefore, should be included only if, after a price increase, Deeplntent would lose so much business to those channels-rather than to Lasso, PulsePoint, or other DSPs in the candidate market-that the price increase would be unprofitable. Tr. 1383 (FTC Closing). Counsel for the FTC asserted that there is no evidence to suggest "that such a significant shift outside of the pool within the FTC's defined market would occur at such a low change in price." Id.
Courts are not entirely consistent in how they frame the HMT, particularly with respect to whether it tests a price increase on one or all of the hypothetical monopolist's products. Compare, e.g., Peabody, 492 F.Supp.3d at 886 ("The FDVIT is an analytical method that asks 'whether a hypothetical monopolist who has control over the products in an alleged market could profitably raise prices on those products.'" (citation omitted)), with Aetna, 240 F.Supp.3d at 20 ("To determine whether a group of products could be an antitrust market, the hypothetical monopolist test asks whether a hypothetical monopolist of all the products within a proposed market would likely impose a [SSNIP]-typically of five or ten percent-on at least one product in the market, including one sold by the merging firms.").
Both the Merger Guidelines and Dr. Israel's own report, however, suggest that it is permissible to test the effects of a price increase on a single product. Dr. Israel described the HMT as asking whether the hypothetical monopolist would impose a price increase "on at least one product in the market." DX0076 at 51. On cross examination, moreover, Dr. Israel was asked if the relevant question for the HMT is "whether the monopolist could profitably impose a SSNIP on one firm's offering." Tr. 1116-17 (Israel). He answered: "I think that one is debated as a matter of law, so I'm not going to weigh in to the law there. . . . [A]s an economist, you could apply it that way. I think I have in cases where I have been instructed to." Tr. 1117 (Israel). The Merger Guidelines, likewise, characterize the HMT as evaluating whether the hypothetical monopolist would impose a price increase "on at least one product in the market, including at least one product sold by one of the merging firms." Merger Guidelines §4.1.1. In other words, the Guidelines indicate that the HMT does not require testing whether the hypothetical monopolist could profitably impose a price increase on all products in the candidate market.
In addition, Dr. Hatzitaskos's report suggests that the aggregate diversion ratio threshold would be even lower in a scenario where the hypothetical monopolist imposes a price increase on all products rather than a single product. PX6500 at 116 n.399 ("Intuitively, if the hypothetical monopolist is increasing its margins on all products in the candidate market, it can afford to lose more sales outside the candidate market before it breaks even").
Defendants do not identify any authority to sustain their position that the HMT requires analysis of whether the hypothetical monopolist would raise prices on all products in the candidate market. They cite FTC v. Advocate Health Care Network, 841 E3d 460 (7th Cir. 2016), but the language they rely on fails to prove their argument. In discussing the relevant geographic market, the Seventh Circuit described the HMT as asking "what would happen if a single firm became the only seller in a candidate geographic region." Id. at 468. If the hypothetical monopolist can profitably raise prices, then the region is a relevant geographic market. Id. On the other hand, "if customers would defeat the attempted price increase by buying from outside the region, it is not a relevant market." Id. This language hardly demonstrates that it is improper to run the HMT by testing whether the hypothetical monopolist could raise prices on one product. If prices increase on one product, customer substitution to products outside the candidate market might make the price increase unprofitable. See Tr. 843 (Hatzitaskos) (explaining that a "key part" of the critical loss analysis is "estimating how much is actually going to go outside of the candidate market"). To use the Seventh Circuit's language, those customers might "defeat the attempted price increase by buying from outside" the relevant market. And again, several courts have adopted the same framing of the HMT that Dr. Hatzitaskos employed. See, e.g., Aetna, 240 F.Supp.3d at 20; Sysco, 113 F.Supp.3d at 33; H&R Block, 833 F.Supp. 2d at 60.
Defendants also make a passing argument in both their pre-hearing and post-hearing briefs that Dr. Hatzitaskos failed to account for "the two-sided nature of HCP programmatic advertising." Doc. 288 at 57; see also Doc. 201 at 17-18. A two-sided platform is "a business that 'offers different products or services to two different groups who both depend on the platform to intermediate between them.'" US Airways, Inc. v. Sabre Holdings Corp., 938 F.3d 43, 56 (2d Cir. 2019) (quoting Ohio v. Am. Express Co. (Amex), 138 S.Ct. 2274, 2280 (2018)). A credit card platform, for instance, includes merchants on one side and cardholders on the other. Id. The "value of the services that a two-sided platform provides increases as the number of participants on both sides of the platform increases." Id. (quoting Amex, 138 S.Ct. at 2281). These two-sided platforms '"must be sensitive to the prices that they charge each side' of the platform." Id. (quoting Amex, 138 S.Ct. at 2281). They often "cannot raise prices on one side without risking a feedback loop of declining demand." Id. (quoting Amex, 138 S.Ct. at 2285). Economists refer to this phenomenon as "indirect network effects." Id. (quoting Amex, 138 S.Ct. at 2280-81). Furthermore, there is a subset of two-sided platforms that always receive two-sided treatment: transaction platforms. Id. "A transaction platform is a two-sided platform where the business 'cannot make a sale to one side of the platform without simultaneously making a sale to the other.'" Id. (quoting Amex, 138 S.Ct. at 2280). These platforms "exhibit more pronounced indirect network effects" because they require that "both sides of the platform simultaneously agree to use their services." Id. (quoting Amex, 138 S.Ct. at 2286). Accordingly, the relevant market must include both sides of the platform. Id. Defendants argue that DSPs are two-sided transaction platforms that experience indirect network effects. Doc. 288 at 57. As they put it: "If a DSP were to raise prices to advertisers and lose customer volume in hopes of increasing profit, that departure of customers would affect that [DSP's] ability to attract publishers, thereby reducing the value of the DSP and causing a further departure of advertiser customers." Id. As the FTC pointed out in its pre-hearing reply brief, however, Defendants do not cite any IQVIA, Deeplntent, or Lasso documents "supporting the claim that their pricing is affected by publisher relationships." Doc. 232 at 15 n.63. Dr. Israel's report, likewise, does not cite any record materials to support his contention that Deeplntent and Lasso are two-sided platforms. See generally DX0076 at 87-89. Nor did Defendants spend time developing this theory at the evidentiary hearing. The sole hearing testimony Defendants rely on is the PulsePoint witness's statement that PulsePoint's parent company, Internet Brands, "provides digital marketing service in two sided marketplaces where professionals and consumers have complicated journeys." Tr. 169 (Gerszke). That is plainly insufficient for the Court to conclude that DSPs qualify as two-sided transaction platforms or that Dr. Hatzitaskos's analysis overlooked indirect network effects. See also PX6504 at 57 (asserting that "Dr. Israel's use of the term 'buy-side tool' is consistent with HCP programmatic advertising services being on only one side").
Nor is the Court persuaded by Defendants' argument that Dr. Hatzitaskos made "economic errors" rendering his analysis unreliable. Doc. 288 at 58. Dr. Hatzitaskos relied on two key inputs for his critical loss analysis: margins and aggregate diversion ratios. PX6500 at 117. As discussed above, he found that the HMT was satisfied for either a 5% or 10% price increase. Dr. Hatzitaskos also explained, however, that the HMT was satisfied even using Dr. Israel's proposed inputs. Tr. 844 (Hatzitaskos). More specifically, in his reply report, he ran the HMT using Dr. Israel's estimates of both Deeplntent's margins and the aggregate diversion ratio. PX6504 at 88 & Exhibit R-12. Using a conservative approach, Dr. Israel estimated Deeplntent's margin to be (Redacted) for 2022. DX0076 at 174-75. Using that margin, along with Dr. Israel's adjusted diversion ratios, Dr. Hatzitaskos still concluded that the HMT would be satisfied for a merger of just Deeplntent, Lasso, and PulsePoint. PX6504 at 88 & Exhibit R-12.
To be sure, Dr. Israel maintained that his proposed adjustments did not address "the core limitations" of the win/loss data or resolve "all of the shortcomings in Dr. Hatzitaskos' analysis." DX0076 at 104. But the Court is not convinced that the win/loss data is as flawed as Dr. Israel asserted. As Dr. Hatzitaskos observed in his reply report, Deeplntent itself relies on win/loss data to assess its own competitive environment. PX6504 at 62; see PX0505 at 191 (Deeplntent CEO stating that win/loss data was used to help "understand what the level of competition was looking like in the marketplace"); id. at 192 (win/loss data offered "the best available comprehensive idea of what's happening among our clients"). Other DSPs also rely on win/loss data: PulsePoint, for instance, tracks wins and losses, and in "the vast majority of cases," the companies it loses business to are Deeplntent and Lasso. Tr. 151 (Gerszke). Furthermore, Dr. Hatzitaskos explained that he did not limit his consideration of win/loss opportunities to any particular channel; rather, his analysis "focused on opportunities that Deeplntent self-identified as losses to competition." PX6504 at 63. He also pointed out that Dr. Israel's claims of campaign revenue that could have been won by a third party other than Lasso or PulsePoint were "inconsistent with the evidence that providers other than Deeplntent, Lasso, and PulsePoint have small revenues." Id. at 64. Finally, Dr. Hatzitaskos testified that there was a wide margin of error in the results he measured: for the test to fail, he would need to have "overestimated the diversion to Lasso and PulsePoint not just by a little bit but by three times as much." Tr. 934 (Hatzitaskos); see also Tr. 932-33 (Hatzitaskos) (stating that questions about "a handful of matches" would not affect his confidence in the overall picture). In the end, while the Court acknowledges that the win/loss data may not be a flawless metric, the Court largely finds persuasive Dr. Hatzitaskos's responses-explained in part during his testimony and detailed more fully in his reply report-to Dr. Israel's criticisms. And though the Court recognizes that Dr. Israel did not concede the correctness of relying on the win/loss data, the Court nonetheless finds it compelling that the HMT was satisfied even using Dr. Israel's adjusted figures.
Taking a step back from the experts' debates about specific inputs, moreover, the Court ultimately finds Dr. Hatzitaskos's analysis of the market to be more consistent with the weight of the testimony and documentary evidence. Courts in similar proceedings have routinely recognized that, even where some questions have been raised about the precise methods used by expert economists, those experts' conclusions may still support a proposed market definition where they are broadly consistent with the rest of the evidence in the record. See Sysco, 113 F.Supp.3d at 36-37 (in case where Dr. Israel served as the FTC's expert, court noted some issues with aggregate diversion calculations but concluded that, "when evaluated against the record as a whole, Dr. Israel's conclusions are more consistent with the business realities of the food distribution market than [the defense expert's]"); H&R Block, 833 F.Supp.2d at 65 (court declined to treat expert's hypothetical monopolist analysis as "conclusive" but found that it was "another data point" supporting the relevant product market); see also CCC Holdings, 605 F.Supp.2d at 41 (despite gap in FTC expert's analysis, "the real-world evidence shows that [the two products] are not part of the same product market").
Conversely, the Court does not find Dr. Israel's conception of the market persuasive. On cross examination, Dr. Israel was asked whether, in his opinion, a hypothetical monopolist controlling all the DSPs in the candidate market could profitably impose a price increase on one of those firms'offerings. Tr. 1117 (Israel). Dr. Israel testified that if a hypothetical monopolist of all those firms increased Deeplntent's prices by 5%, there would be so much substitution to other channels-that is, firms other than those DSPs-that it would make the price increase unprofitable. Tr. 1118 (Israel). He asserted that there would be "ample substitution elsewhere," particularly given the size of Doximity and Medscape, which provide "really the same service." Id.
Having considered the record as a whole, the Court cannot agree with Dr. Israel's conclusion. As discussed above in the section concerning the Brown Shoe factors, the Court has found that it is not accurate to characterize Doximity and Medscape as providing "the same service" as the programmatic offerings of the DSPs in the FTC's proposed market. Substantial evidence throughout the record demonstrates that programmatic is distinct from social and endemic channels and is viewed as such by industry participants. As a result, the Court does not find it plausible that, in the event the hypothetical monopolist increased Deeplntent's prices, there would be so much substitution outside of the proposed market that the price increase would become unprofitable. Cf. Sysco, 113 F.Supp.3d at 37 (rejecting expert testimony on market definition that was "inconsistent with business reality" and contradicted by "evidence from industry leaders").
To reiterate, the Court need not and does not conclude that Dr. Hatzitaskos's methods and conclusions are beyond any criticism. See Aetna, 240 F.Supp.3d at 42 ("Although the Court does not (and does not need to) adopt his analysis in every detail, [the government's expert] has performed a battery of tests that all point to the same conclusion . . . ."); H&R Block, 833 F.Supp.2d at 71 (expert's analysis "tends to confirm ... the relevant market" even though "the Court would not rely on his analysis exclusively"); cf. Whole Foods, 548 F.3d at 1048 (Tatel, J., concurring in the judgment) ("Although courts certainly must evaluate the evidence in section 13(b) proceedings and may safely reject expert testimony they find unsupported, they trench on the FTC's role when they choose between plausible, well-supported expert studies."). The Court simply finds that, all things considered, his testimony-including the adjustments he made in his reply report in response to Dr. Israel's critiques-tends to reinforce the conclusion that HCP programmatic advertising is a relevant product market.
2. Relevant Geographic Market
As mentioned above, there is no dispute in this case about the relevant geographic market. The FTC asserts that "while the target[s] of the programmatic advertising are U.S.-based healthcare professionals, which is a factor in the FTC's product market, nothing requires that suppliers or customers of HCP programmatic advertising be located in the United States in order to target U.S.-based HCPs." Doc. 287 (FOF) at 4. During closing arguments, moreover, counsel for the FTC stated that "we denned geographic market conservatively so as not to exclude any providers of HCP programmatic advertising who might be international." Tr. 1367 (FTC Closing).
Defendants, for their part, briefly suggest that the FTC "did not offer any evidence at trial of competition outside of the United States." Doc. 288 at 54. They go on to say, however, that "because there is no evidence of international competition, whether the relevant market is the United States or global is not dispositive of the Court's analysis." Id. Counsel for IQVIA stated during closing arguments that "we don't need to spend too much time on the geographic market" because the FTC conceded that "a supplier of HCP programmatic advertising services can come from anywhere in the world." Tr. 1404 (IQVIA Closing); see also Id. ("A competitive constraint could come from anywhere in the world. We agree with that. So that is probably not a dispositive portion of the proceedings.").
Given the lack of disagreement between the parties on this point, the Court will assume that the relevant geographic market is worldwide.
3. Effects on Competition
Because the FTC has met its burden to define a relevant market, the next step is evaluating the effects of the proposed transaction on competition within that market. E.g., Peabody, 492 F.Supp.3d at 902; Sysco, 113 F.Supp.3d at 52. If the FTC can make out a prima facie case that the acquisition "will result in a significant market share and an undue increase in concentration within [the relevant market], a presumption is established that it will substantially lessen competition." Swedish Match, 131 F.Supp.2d at 166; see also, e.g., Peabody, 492 F.Supp.3d at 902.
The FTC relies on two arguments to show that IQVIA's acquisition of Deeplntent will substantially impair competition in the market for HCP programmatic advertising. First, the FTC looks to post-merger market shares and market concentration. It argues that the merged firm's market share will exceed the 30% threshold that triggers a presumption of anticompetitive effects. The FTC also contends that the Herfindahl-Hirschman Index (HHI), a tool commonly used to measure changes in market concentration, supports a finding that the merger will harm competition. Second, the FTC relies on the elimination of substantial direct competition between Deeplntent and Lasso, pointing to both qualitative and quantitative evidence to support that conclusion. Defendants raise numerous objections to these theories.
It is worth repeating that the Court's task at this stage "is to determine whether the FTC 'has raised questions going to the merits so serious, substantial, difficult and doubtful as to make them fair ground for thorough investigation, study, deliberation and determination by the FTC in the first instance and ultimately by the Court of Appeals.'" Peabody, 492 F.Supp.3d at 907 (quoting Heinz, 246 F.3d at 714-15). The Court finds that the FTC has met its burden. The FTC's market share and HHI calculations-set out in the reports and testimony of Dr. Hatzitaskos-establish a presumption that the proposed acquisition will harm competition in the market for HCP programmatic advertising. And that presumption is reinforced by ample evidence that the transaction would eliminate substantial head-to-head competition between Deeplntent and Lasso.
a. Market Shares and Market Concentration
First, the FTC argues that the post-merger market share of Deeplntent and Lasso would exceed both the 30% threshold first set out in United States v. Philadelphia National Bank, 374 U.S. 321 (1963), and the relevant threshold for market concentration based on the HHI. At the outset, Defendants respond that the 30% threshold set out in Philadelphia National Bank has since been repudiated. Furthermore, they contend that there are significant errors in Dr. Hatzitaskos's market share calculations. The corrected figures, according to Defendants and Dr. Israel, fall short of both the 30% mark and the HHI threshold. The Court is not persuaded by Defendants' arguments.
i. The 30% Threshold
In Philadelphia National Bank, the Supreme Court addressed the standard under section 7 of the Clayton Act for determining whether a merger may substantially lessen competition in the relevant market. See 374 U.S. at 362. The Court explained:
This intense congressional concern with the trend toward concentration warrants dispensing, in certain cases, with elaborate proof of market structure, market behavior, or probable anticompetitive effects. Specifically, we think that a merger which produces a firm controlling an undue percentage share of the relevant market, and results in a significant increase in the concentration of firms in that market is so inherently likely to lessen competition substantially that it must be enjoined in the absence of evidence clearly showing that the merger is not likely to have such anticompetitive effects.Id. at 363. The Court observed that this test "lightens the burden of proving illegality only with respect to mergers whose size makes them inherently suspect in light of Congress' design in [section] 7 to prevent undue concentration." Id. In that case, the merger would have resulted in a single bank's controlling 30% of the relevant market. Id. at 364. The Court held: "Without attempting to specify the smallest market share which would still be considered to threaten undue concentration, we are clear that 30% presents that threat." Id.
Defendants contest the present-day validity of the 30% threshold. Doc. 288 at 65; see also Tr. 1421-22 (IQVIA Closing). They suggest that intervening case law has "cut. . . back sharply" on Philadelphia National Bank. Id. (omission in original) (quoting Baker Hughes, 908 E2d at 990). As a matter of economics, moreover, Dr. Israel testified that "nothing says above or below 30 percent tells you anything in particular." Tr. 1070 (Israel). And Defendants assert that the Merger Guidelines focus solely on HHI without discussing the 30% threshold. Doc. 288 at 65.
Still, the Court is hard-pressed to conclude that the Philadelphia National Bank presumption has been repudiated. Second Circuit precedent appears to directly contradict that conclusion. In United States v. Waste Management, 743 F.2d 976 (2d Cir. 1984), for example, the post-merger market share was 48.8%. Id. at 981. That figure, the court held, was "sufficient to establish prima facie illegality under [Philadelphia National Bank] and its progeny." Id. A few years later, the Second Circuit observed that a post-merger market share of 32.3% was "above the 30% held by the Supreme Court to trigger a presumption of illegality in Philadelphia Nat 7 Bank." Consol. Gold Fields PLC v. Minorco, S.A., 871 E2d 252, 260 (2d Cir. 1989). That market share was "certainly sufficient to satisfy appellees' burden of showing likelihood of success on the merits." Id. More recently, courts have continued to invoke the 30% threshold as sufficient to establish a presumption of anticompetitive effects. See, e.g., Deutsche Telekom, 439 F.Supp.3d at 205 ("By one measure, a merger will be presumptively anticompetitive if the merged firm would have more than a 30 percent market share."); Energy Solutions, 265 F.Supp.3d at 441 ("While there is no bright-line rule as to the minimum percentage that qualifies as undue, the Supreme Court has held that a post-merger market share of 30% triggered the presumption of anticompetitive effects."); see also Bertelsmann, 646 F.Supp.3d at 37 (post-merger market share of 49% was "far above the levels deemed too high in other cases"). In light of the above, the Court cannot agree with Defendants' suggestion that the 30% threshold is no longer valid.
To be sure, market shares alone are not dispositive. See, e.g., Deutsche Telekom, 439 F.Supp.3d at 206 (noting that "market shares and HHIs establish only a presumption, rather than conclusive proof of a transaction's likely competitive impact"). But the case law indicates that objections to the competitive picture provided by market shares are more properly considered at the rebuttal phase of the burden-shifting framework. See Baker Hughes, 908 F.2d at 991 ("In the aftermath of General Dynamics and its progeny, a defendant seeking to rebut a presumption of anticompetitive effect must show that the prima facie case inaccurately predicts the relevant transaction's probable effect on future competition."); see also Waste Mgmt., 743 F.2d at 981 (explaining that under Philadelphia National Bank, "a merger resulting in a large market share is presumptively illegal, rebuttable only by a demonstration that the merger will not have anticompetitive effects").
Here, Dr. Hatzitaskos calculated that the proposed merger would result in IQVIA's controlling 46% of the market. PX6500 at 122-23 & Exhibit 13. According to the FTC, Dr. Israel estimated that the combined firm's post-merger revenue share would be 30.6%. Doc. 287 (COL) at 15 (citing DX0076 at 134-35 & fig.12). As discussed below, Defendants object to Dr. Hatzitaskos's calculations on multiple grounds. But for the reasons just explained, the 30% threshold remains valid as a matter of law. Therefore, the FTC has established a presumption of anticompetitive effects if either expert's figure is accurate.
The Court notes that Dr. Israel qualified this calculation. Specifically, he asserted that Dr. Hatzitaskos erred by miscalculating revenues for several firms within his market and excluding other firms altogether. DX0076 at 128-34. Dr. Israel thus explained that his approach was "necessarily conservative" because it omitted firms such as Doximity that, in his view, should have been included in the market but for whom he did not have any basis to estimate their revenues. Id. at 134. Nevertheless, the Court concludes that Dr. Israel's qualifications are largely irrelevant because the Court has accepted the FTC's proposed market definition. In other words, the Court has found that Doximity, Medscape, and other alternative channels are not reasonably interchangeable substitutes for HCP programmatic advertising, and thus their exclusion from Dr. Israel's estimated revenue shares is not dispositive.
ii. The HHI Threshold
Setting the 30% threshold aside, the FTC argues that the market concentration would also be excessive under the HHI. Id. at 16. "The HHI is calculated by summing the squares of the individual firms' market shares." Perm State, 838 F.3d at 346. For instance, in a market with two firms each controlling 50% of the market, the HHI would be 5,000; by contrast, if one hundred firms each controlled 1% of the market, the HHI would be just 100. Bertelsmann, 646 F.Supp.3d at 49 & n.35. "The HHI takes into account the relative size and distribution of the firms in a market, increasing both as the number of firms in the market decreases and as the disparity in size among those firms increases." Staples, 970 F.Supp. at 1081 n.12.
Consistent with the Merger Guidelines, courts consider "both the post-merger HHI number and the increase in the HHI resulting from the merger." Penn State, 838 E3d at 347. If the post-merger market has an HHI above 2,500, then it is classified as "highly concentrated." Id. A merger that increases the HHI by more than 200 points is "presumed to be likely to enhance market power." Id. (quoting Merger Guidelines § 5.3). "The Government can establish a prima facie case simply by showing a high market concentration based on HHI numbers." Id.; see also Saint Alphonsus, 778 F.3d at 788 ("The extremely high HHI on its own establishes the prima facie case."); Heinz, 246 E3d at 716 ("Sufficiently large HHI figures establish the FTC's prima facie case that a merger is anti-competitive.").
In this case, Dr. Hatzitaskos calculated a post-merger HHI of 3,635 and an increase of 997. PX6500 at 125 & Exhibit 15. Both of those figures, if correct, would be well above the relevant threshold for the FTC to establish its prima facie case. See, e.g., Heinz, 246 F.3d at 716 (HHI increase of 510 points "creates, by a wide margin, a presumption that the merger will lessen competition in the [relevant] market"). So if Dr. Hatzitaskos's calculations are reliable, then the FTC has met its burden to establish a prima facie case based on the HHI as well as the 30% threshold.
iii. Defendants 'Arguments
Defendants object to Dr. Hatzitaskos's market share calculations on various grounds. Doc. 288 at 44-46, 66-67. Ultimately, the Court is not persuaded by these arguments.
First, Defendants argue that Dr. Hatzitaskos's market share figures are erroneous because he excluded social media and endemic sites such as Doximity and Medscape. Id. at 44, 65-66. If the substantial revenues of those firms were accounted for, Defendants contend, then "the merged firm would not meet any pertinent concentration threshold." Id. at 44. For the reasons discussed in the market definition section, however, the Court finds that it was not improper for Dr. Hatzitaskos to exclude Doximity and Medscape-as well as other firms falling into the social media and endemic categories-from his calculation of market shares. And based again on the analysis above, it was also not improper for him to exclude Google. Contra Id. at 45, 66.
Defendants next assert that Dr. Hatzitaskos underestimated the HCP programmatic advertising revenue attributable to (Redacted), a generalist DSP. Id. at 44-45. In his report, Dr. Hatzitaskos attributed (Redacted) for the eighteen-month period he analyzed. PX6500 at 123, Exhibit 13. He used that estimate as a "placeholder" because revenue data from (Redacted) was produced the day before the filing of his report. Id. at 200. The data ultimately produced by (Redacted) indicated that the actual number was (Redacted), but he did not have time to incorporate that figure into his analysis. Id.
Dr. Israel, on the other hand, concluded based on an internal document from (Redacted) that a conservative estimate of its 2023 revenue was (Redacted). DX0076atl06. (Redacted) figure over eighteen months to match Dr. Hatzitaskos's timeframe, Dr. Israel arrived at an estimate of (Redacted) in HCP programmatic revenue for (Redacted) Id. at 107.
In his reply report, Dr. Hatzitaskos amended part of his analysis to use a figure of (Redacted), more than doubling his initial estimate. PX6504 at 84. That figure was based on (Redacted) average monthly revenue from July through September 2023, multiplied by eighteen. Id. Dr. Hatzitaskos also rejected Dr. Israel's suggestion that the correct figure was much higher. Id. at 84, 123-24. The document that Dr. Israel relied on, according to Dr. Hatzitaskos, was based on forecasted potential revenue for (Redacted) rather than realized revenue. Id. at 84, 124. In his testimony, Dr. Israel confirmed that the business document he relied on was a forecast of salespeople projecting "what HCP revenue they think they can obtain in 2023." Tr. 1067 (Israel); see also Id. ("It is salespeople saying what they think they can do.").
During the hearing, the witness from (Redacted) cast substantial doubt on the figure that Dr. Israel used. Specifically, the witness was asked: "If I told you that that (Redacted) annually from HCP programmatic advertising, what would your response be?" Tr. (Redacted). He responded: (Redacted) Id. Counsel for the FTC then asked: "Do you have a rough sense of what the revenues are for HCP programmatic advertising for (Redacted)" Id. The witness answered: "If we are talking specifically about the ability to target HCPs and report out with physician-level data, for this year our actual revenues from those types of campaigns would be, I'm estimating, (Redacted)."
Defendants rely on other evidence in an attempt to contradict the figure provided by (Redacted) witness. Doc. 288 at 17-18. For instance, they point to testimony from one pharmaceutical company whose (Redacted) uses (Redacted) for HCP programmatic advertising. Tr. (Redacted)- The witness testified that the (Redacted) annual spending on HCP programmatic advertising on (Redacted). Tr. (Redacted). Counsel for IQVIA then asked: (Redacted). Counsel then asked: (Redacted) (Redacted)
The Court finds it difficult to conclude from this testimony that Dr. Israel's (Redacted) estimate-over eighteen months-is more accurate than Dr. Hatzitaskos's (Redacted) figure. (Redacted)
Regardless, even assuming Dr. Israel's figure is correct-which, again, requires disregarding testimony from (Redacted) that anybody suggesting such a number would be "very misinformed"-the combined revenue shares he calculated for Deeplntent and Lasso would still be 30.6%. DX0076 at 135 fig. 12. And that total should be even higher, in fact, because Dr. Israel's analysis included revenue from Sermo and Facebook, along with other social media firms, see id., which the Court has determined should not be included in the market for the reasons already explained. Sermo and Facebook alone account for (Redacted) of the revenue shares in Dr. Israel's calculations, so removing them would elevate Deeplntent and Lasso's combined share even further above the 30% threshold.
Aside from the dispute over (Redacted) revenues, Defendants argue that Dr. Hatzitaskos's figures are inaccurate because "many of the largest healthcare advertising agencies do no business with PulsePoint, Deeplntent, or Lasso at all." Doc. 288 at 66. A chart in Dr. Israel's report lists the top twenty agencies (by revenue) involved in healthcare advertising and their spending with PulsePoint, Deeplntent, and Lasso in 2022. DX0076 at 294 & fig.45. Dr. Israel observed that six of the twenty agencies did not use any of those three firms. Id; Tr. 1065-66 (Israel). It is hard to see how this establishes that the market shares estimated by Dr. Hatzitaskos are erroneous. Dr. Israel's chart does not demonstrate which other firms these agencies are allocating their budgets to or, more critically, how much of their budgets are allocated to those firms. The Court cannot conclude from these figures alone that Dr. Hatzitaskos's calculations are unreliable.
Nor is the Court persuaded by Defendants' argument that Dr. Hatzitaskos's market shares left out some DSPs that are engaged in HCP programmatic advertising. Doc. 288 at 45. For instance, Defendants point to testimony from one company that it has activated HCP audiences on generalist DSPs such as The Trade Desk, Yahoo, and Viant. (Redacted) at 22-23. They assert that Dr. Hatzitaskos did not have proper data for these DSPs. Doc. 288 at 45. Even assuming Dr. Hatzitaskos's original analysis was underinclusive, however, he expanded the scope of the market he tested in his reply report. When Dr. Hatzitaskos calculated revised market shares that incorporated some of the firms included in Dr. Israel's analysis, he still arrived at a post-merger HHI of 3,320 and an HHI increase of 893 points-both, again, well above the thresholds set forth in the Merger Guidelines. PX6504 at 91. And as mentioned above, the 30% threshold was exceeded even according to Dr. Israel's estimated shares, DX0076 at 135 fig. 12, which included social media firms that the Court has concluded are not reasonably interchangeable substitutes.
Additionally, Defendants contend that Dr. Hatzitaskos's market shares are unreliable because they are based on only six quarters of revenue-from 2022 and the first half of 2023-and because he "failed to capture the fact that the industry is dynamic and changing." Doc. 288 at 45. The Court concludes that these objections are more appropriately analyzed as part of Defendants' rebuttal arguments. See, e.g., Energy Solutions, 265 F.Supp.3d at 442 ("Once the government establishes a prima facie case, the defendant must 'show that the market-share statistics [give] an inaccurate account of the acquisitions' probable effects on competition.'" (alteration in original) (quoting United States v. Citizens & S. Nat'l Bank, 422 U.S. 86, 120 (1975))).
In evaluating all these arguments, the Court is cognizant of the fact that "[t]he FTC need not present market shares and HHI estimates with the precision of a NASA scientist." Sysco, 113 F.Supp.3d at 54. "The 'closest available approximation' often will do." Id. (quoting FTC v. PPG Indus., Inc., 798 F.2d 1500, 1505 (D.C. Cir. 1986)); cf. FTC v. TronoxLtd, 332 F.Supp.3d 187, 212 (D.D.C. 2018) (finding expert's conclusions consistent with "business realities"). The Court concludes that under either the 30% threshold or the HHI benchmark, the FTC has established a prima facie case that the proposed transaction will have anticompetitive effects.
b. Head-to-Head Competition
The high post-merger levels of market concentration alone would be sufficient for the FTC to state a prima facie case. See, e.g., Energy Solutions, 265 F.Supp.3d at 442 (post-merger FtHI established "a prima facie case that market concentration will result in anti-competitive effects"). Nevertheless, the FTC has also offered substantial evidence of head-to-head competition between Deeplntent and Lasso that would be eliminated by the proposed transaction.
In several cases, courts have acknowledged that the elimination of competition between the merging parties can strengthen the conclusion that the merger will have anticompetitive effects. See, e.g., Heinz, 246 F.3d at 717 (market concentration statistics were "bolstered by the indisputable fact that the merger will eliminate competition between the two merging parties"); Sysco, 113 F.Supp. at 61 ("Courts have recognized that a merger that eliminates head-to-head competition between close competitors can result in a substantial lessening of competition."); Staples, 970 F.Supp. at 1083 (noting that the merger "would eliminate significant head-to-head competition between the two lowest cost and lowest priced firms in the superstore market"); see also Peabody, 492 F.Supp.3d at 903 (collecting cases). Where the merger "would result in the elimination of a particularly aggressive competitor in a highly concentrated market," that factor "is certainly an important consideration when analyzing possible anti-competitive effects." H&R Block, 833 F.Supp.2d at 79 (quoting Staples, 970 F.Supp. at 1083). Crucially, this principle holds true "even where the merging parties are not the only two, or even the two largest, competitors in the market." United States v. Anthem, Inc., 236 F.Supp.3d 171, 216 (D.D.C. 2017); accord, e.g., Aetna, 240 F.Supp.3d at 43. Courts frequently rely on ordinary course documents and witness testimony illustrating that two merging parties view each other as strong competitors. See, e.g., FTC v. Wilh. Wilhelmsen Holding ASA, 341 F.Supp.3d 27, 63 (D.D.C. 2018) (finding anticompetitive effects where "the record includes multiple instances of serious competition between [the merging parties], specifically on issues such as price and other non-price incentives"); Aetna, 240 F.Supp.3d at 44-45 (documents showed "significant evidence of head-to-head competition between [the merging parties] throughout the country").
Defendants criticize the FTC for failing to provide any "economic measure for how 'close' that competition [between Deeplntent and Lasso] must be before a merger is unlawful." Doc. 288 at 69-70. But Defendants provide no support for the proposition that such an "economic measure" is necessary. And the cases just cited indicate that courts regularly evaluate ordinary course documents and other evidence to provide a qualitative assessment as to whether the merging parties are head-to-head competitors.
Here too, the documents tell the story. Time and again, Defendants' own records reveal evidence of fierce competition between Deeplntent and Lasso. For instance, Deeplntent documents repeatedly refer to Lasso as a significant competitor. PX2571-01 (referring to Lasso as "our #1 competitor" and stating that "[t]he entire Senior Leadership Team feels that we should definitely address Lasso heads on via marketing and sales 1:1s"); PX2506-01 (noting focus on Lasso and PulsePoint "since they're our direct competition"); PX2736-04 (stating that "we are in a dogfight" for an agency's business and that "I would say it's primarily between us and Lasso"); PX2564-04 ("Lasso is on fire" and "[i]t's going to be really important we're blunting them" "and pulling out every stop with the key agencies they're referencing"); PX2554-01 ("we can't let [Lasso] steal share away from us this RFP season"); PX2804-01 (in response to Lasso press release, Deeplntent's CEO stated that "we need to develop a kill sheet on this" and that "gloves are off with Lasso"); PX2764-01 ("we need a few strong bullets as to what makes our integrated planning, activation & real-time optimization, stronger than Lasso" (emphasis omitted)). See generally Tr. 617-18 (Paquette) (Deeplntent CEO agreeing that Deeplntent competes with Lasso on price and product capabilities).
While some of these documents are from 2021 or 2022, there is evidence indicating that this competitive focus persisted into 2023. Deeplntent continued to refer to Lasso internally as a key competitor. PX2771-02 ("let's figure out how low we can go on CPMs as we've lost 5 in a row to Lasso due to rates"); PX2746-01 (identifying Lasso and PulsePoint as "our direct competitors"); PX2747-01 (noting that an ad agency was "aggressively shifting budgets from us to Lasso"). Deeplntent also regularly identified Lasso as one of its main competitors when responding to RFIs from ad agencies. PX2812;PX2816.
The attention paid to the competition was mutual. Internal documents from Lasso display a similar focus on Deeplntent as a significant competitor. PX1056-04 ("We have been very clear that Deep[I]ntent is our largest competitor on the programmatic side of things . . . ."); PX1375-02 (identifying Deeplntent as a direct competitor). Again, this includes documents from 2023. PX1743-01 ("We're very well aware that we have caused our two competitors on[] the programmatic side to drop their costs pretty dramatically. So we've seen [Deeplntent] in particular really kind of drop almost to exactly what we're bringing from a pricing perspective."); PX1628-02 (discussing need to distinguish Lasso from "our two key competitors (PulsePoint and Deeplntent)"); PX1625-02 (referring to Deeplntent, Lasso, and PulsePoint as "the Big 3" and noting that "the feeling is we should be talking about what makes our data unique vs. the competition").
In particular, the documents reflect that Deeplntent and Lasso compete on price. In one email, Deeplntent's CEO wrote: "We will increasingly be bumped up against [Lasso] in pitches. They also make claims that their CPMs are 50% lower than us; let's be cognizant and alert as to any pricing conversations that are happening with our clients. In most cases, I don't want to lose market share to these guys because of pricing." PX2573-01; see also, e.g., PX2508-01 (Deeplntent CEO noting "more and more push back from clients on rates," that "Lasso is driving the market price down," and that "we need [to] address this head on"); PX2509-01 (Deeplntent CEO stating that (Redacted) we've the of our targeting data by (Redacted) to answer Lasso's practice of undercutting our prices"); PX2510-01 (in email prior to IQVIA's acquisition of Lasso, Deeplntent CEO stated that "IQVIA has given us the go-ahead to price match Lasso's HCP clinical data targeting on a case-by-case basis. I believe this should move the needle with some Lasso strongholds . . . ." (emphasis omitted)); PX2772-03 (Deeplntent approving price reduction for a client where Deeplntent believed it was "being evaluated against Lasso rates" and reduction would "go a long way to protect against Lasso"); PX2822-01 (Deeplntent telling an ad agency it could offer rates "that are (Redacted) less than [Lasso's] rates"); PX2843-04 (Deeplntent describing Lasso as its "primary competitor" on pricing); PX1443-01 (Lasso chat speculating as to whether "Deeplntent sent that RFP on purpose to try and trick us/get us to lower our rate").
Deeplntent and Lasso compete to stay ahead of each other with respect to product quality and innovation as well. PX2797-01 (Deeplntent chat stating that Lasso measurement announcement "reinforces the driving need to successfully launch Patient Planner, our latest product innovation, to the marketplace. . . . We need to continue the focus on making our current product suite superior to the competition while accelerating product innovation to maintain our competitive advantage"); PX2512-01 (Deeplntent CEO citing Lasso press release and stating "competition is catching up . . . need to accelerate the innovation"); PX2578-01 ("The recent Lasso measurement announcement is a good reminder that the competitio[n] is relentlessly pursuing Deeplntent. To stay ahead of the competition, Engineering and Product needs to move faster, with more shared urgency and agility to keep our competitive distance from Lasso and PulsePoint."); PX1433-03 (Lasso chat discussing making measurement product free "to (Redacted) [D]eep[I]ntent"); PX1064-02 (Lasso chat that referenced "making sure [D]eep[I]ntent, (Redacted)" and stated that "I want to destroy them so bad" and that "it's annoying (Redacted) to see them doing as well as they are. Even more frustrating to see them copying and now playing dirty").
Along with these ordinary course documents, testimonial evidence indicated that the two firms are close competitors. See Wilhelmsen, 341 F.Supp.3d at 62 (relying on both documents and customer testimony as evidence of head-to-head competition); Sysco, 113 F.Supp.3d at 65 (relying on testimony from industry actors who viewed the merging parties as "close competitors"). Importantly, both customers and other industry participants testified that they view Deeplntent and Lasso as head-to-head competitors for HCP programmatic advertising. Tr. (Redacted) (pharmaceutical company witness stating that (Redacted) considers Deeplntent and Lasso to be direct competitors and explaining that "[t]hey are both DSPs that have pitched to me as a DSP, as an HCP-specific DSP, and so in my mind they are competitors"); Tr. 149, 151 (Gerszke) (PulsePoint witness testifying that Deeplntent and Lasso are among its key competitors, that it typically encounters those two firms when competing for RFPs, and that when tracking wins and losses it primarily loses business to those two firms); Tr. (Redacted) (generalist DSP identifying its biggest competitors for HCP programmatic advertising as Deeplntent, PulsePoint, and Lasso); Tr. 806-07 (Lawson) (witness from another generalist DSP explaining that "[w]hen we compete for programmatic dollars, we typically come up against the same organizations: Deeplntent, PulsePoint, Lasso being the three that are most prevalent").
Throughout these proceedings, Defendants have attempted to minimize the significance of this evidence of head-to-head competition between Deeplntent and Lasso. See, e.g., Tr. 60 (IQVIA Opening) ("[I]t is totally unremarkable that two new entrants into a crowded market were taking a similarly narrow approach to focus on one another."); Tr. 1432 (IQVIA Closing) ("[T]he FTC suggests that the elimination of that [head-to-head] competition is somehow itself nefarious which is why they cherry-picked some documents to suggest that Deeplntent and Lasso only compete with one another."); Doc. 288 at 70 (asserting that the testimony and documentary evidence do not "support a freestanding assumption of competitive harm"). During the hearing, witnesses from the merging parties tried to downplay their focus on Deeplntent and Lasso in these emails, chats, and other documents. On balance, however, the Court is more persuaded by the plain import of their contemporaneous statements as reflected in the documentary record than by Defendants' attempts to diminish the substantial evidence of head-to-head competition. See FTC v. HackensackMeridian Health, Inc., No. 20 Civ. 18140 (JMV), 2021 WL 4145062, at *9 (D.N.J. Aug. 4, 2021) (merging parties "downplayed the fact that they are competitors" at the hearing, but emails "and other ordinary course documents . . . demonstrate otherwise"); cf. Peabody, 492 F.Supp.3d at 895 (recognizing the "risk of relying on such testimony [about significance of competition], particularly when it comes from Defendants' employees").
In addition, Defendants argue that the evidence of head-to-head competition is unimportant because Deeplntent and Lasso also compete with many other firms. See Doc. 288 at 70-72 (summarizing evidence and testimony). There is no doubt that Defendants compete with other firms to some degree. But the FTC is not required to establish that Deeplntent and Lasso are exclusive competitors. See Anthem, 236 F.Supp.3d at 216 ("The acquired firm need not be the other's closest competitor to have an anticompetitive effect; the merging parties only need to be close competitors."). It is sufficient to show, as the FTC has, that Defendants vigorously compete head-to-head and that this competition would be eliminated by the proposed transaction. See Id. at 219 (documentary record "unquestionably" showed that the merging parties were aggressive competitors); see also Merger Guidelines § 6 ("The elimination of competition between two firms that results from their merger may alone constitute a substantial lessening of competition."). And in any event, the examples of competition with other firms cited by Defendants pale in comparison to the overwhelming evidence reflecting that Deeplntent and Lasso have been intensely focused on each other and PulsePoint as direct competitors.
c. Quantitative Evidence
The evidence of anticompetitive effects from the proposed acquisition is further buttressed by Dr. Hatzitaskos's quantitative analysis. Consistent with the documentary record highlighted above, Dr. Hatzitaskos testified that "Deeplntent and Lasso are strong competitive constraints on one another." Tr. 856 (Hatzitaskos). Therefore, "if you remove that head-to-head competition . . . there will be a substantial lessening of competition that will lead to harm to customers that may take several forms, higher prices, lower quality, lower innovation." Id. Dr. Hatzitaskos explained that his quantitative analysis evaluated both actual customer choices and pricing incentives. Tr. 857 (Hatzitaskos). First, he concluded from customer choice data that around 80% of customers who considered but did not choose Deeplntent elected instead to go to Lasso or PulsePoint. Tr. 858 (Hatzitaskos). In other words, "the vast majority of customers who considered] Deeplntent and didn't choose them ended up going to one of the other two of the big three." Id. Dr. Hatzitaskos evaluated 2023 opportunities in his initial report and then expanded the analysis to 2022 in his reply report, but he found that the overall picture "doesn't really change." Tr. 859 (Hatzitaskos).
This customer choice data also formed the basis for Dr. Hatzitaskos's calculation of the aggregate diversion ratio, which the Court discussed above in the context of the hypothetical monopolist test.
Second, Dr. Hatzitaskos used two models to analyze pricing incentives: (1) a merger simulation, and (2) a gross upward pricing pressure index (GUPPI) analysis. Tr. 860 (Hatzitaskos). He ultimately concluded that the merged firm would have an incentive to raise Deeplntent's prices by up to 11% or 13%. Id. Dr. Hatzitaskos noted that he used "many different inputs" and conducted "various what-if analyses to test the results." Id. For example, he took the size of the fourth-largest member of the market- (Redacted), the firm with the highest revenue outside of the Big 3-and added ten firms of that same size into the market. Tr. 861 (Hatzitaskos); PX6504 at 90. While Dr. Hatzitaskos did not believe he omitted any firms from his candidate market, he observed that he still found "a substantial lessening of competition" even with that fourth-largest firm added to the market ten times. Tr. 861 (Hatzitaskos). And he explained that his findings were broadly consistent with the qualitative evidence throughout the record:
Again, in my report I go through a lot of evidence, both in the initial and in [the] reply report, where we just see again and again in the ordinary course of business documents where Deeplntent identifies Lasso as a close competitor. Lasso identifies Deeplntent as a close competitor, where they essentially view other providers to be less competitive and missing some features. Third parties identified them as close competitors. And just direct evidence. We have heard some examples here too where the competition between them seems to have led to lower prices, to accelerated innovation and so on.Tr. 861-62 (Hatzitaskos).
Defendants again take issue with Dr. Hatzitaskos's analysis. During the hearing, Dr. Israel discussed his objections to both models used by Dr. Hatzitaskos. First, he explained that the GUPPI analysis is "generally not used in business-to-business settings with negotiations." Tr. 1079 (Israel). He continued: "It's really an analysis designed for consumer products, like buying Pepsi or Coke, as opposed to negotiated prices." Id. Furthermore, the GUPPI analysis fundamentally depends on win/loss data and margins, which Dr. Israel said were subject to the same flaws he identified in his criticism of the market definition analysis. Id.; see also Tr. 1060-63 (Israel).
With respect to the merger simulation-also referred to as a second score auction model-Dr. Israel explained its logic as follows: "Buyers need at least two good choices to get a good deal. They are going to choose one person they are going to buy from, and they need a stalking horse. . . . The model explicitly says that as long as there [are] two good options post merger, there won't be a competitive harm." Tr. 1079-80 (Israel). Dr. Israel asserted that PulsePoint could be the second choice in this market, and so could Medscape, Doximity, or The Trade Desk. Tr. 1080 (Israel). As a result, "[t]he logic of the model really explains why there shouldn't be a problem, because these powerful agencies have good options left to constrain prices." Id. Furthermore, Dr. Israel opined that the inputs used by Dr. Hatzitaskos-specifically, the market shares and profit margins-were incorrect and caused the model to predict effects that were too large. Id. Using his own market shares and partially adjusted margins, Dr. Israel arrived at a much lower prediction of a mere 2% price increase. Tr. 1080-81 (Israel). And with further corrections to the margins, he concluded that there would be no harm at all. Tr. 1081 (Israel). Finally, Dr. Israel contended that Dr. Hatzitaskos failed to account for benefits from the transaction in his analysis. Tr. 1081-82 (Israel). Accounting for lower data costs to Deeplntent after a merger with IQVIA, he concluded, the model predicts lower prices rather than higher prices post-merger. Id.
On the whole, the Court finds Dr. Hatzitaskos's analysis to be more persuasive. First, with respect to the GUPPI analysis, other courts appear to have concluded that it is an appropriate tool in business-to-business industries characterized by bargaining. See Wilhelmsen, 341 F.Supp.3d at 64-65. In any event, Dr. Hatzitaskos responded to Dr. Israel's critiques by testing a variant of the GUPPI analysis that "specifically accounts for industries with bidding competition." PX6504 at 66. Under that test, Dr. Hatzitaskos found that the diversion ratios for the merging parties were even higher than those he initially calculated-meaning that his original analysis was conservative. Id. As for Dr. Israel's criticism of the inputs used, the Court generally is not swayed by his objections to the win/loss data for the reasons already discussed in the Court's analysis of the hypothetical monopolist test. And with respect to the margins, the Court finds persuasive Dr. Hatzitaskos's explanation in his reply report that his margins reflect the relevant costs of running an additional HCP programmatic advertising campaign and that the costs he accounted for are consistent with the costs reflected in Lasso's campaign data. Id. at 69; see also Merger Guidelines § 4.1.3 (noting that the DOJ and the FTC "often estimate incremental costs . . . using merging parties' documents or data the merging parties use to make business decisions").
Second, while the merger simulation "is 'an imprecise tool,' it 'nonetheless has some probative value in predicting the likelihood of a potential price increase after the merger.'" Aetna, 240 F.Supp.3d at 46-47 (quoting H&RBlock, 833 F.Supp.2d at 88); see also PX6500 at 93 (explaining that the merger simulation "is not meant to be an exact prediction of post-merger pricing"). In this case, Dr. Hatzitaskos's merger simulation, using Deeplntent's margin, estimated a post-merger price increase of 7.4%. PX6500 at 93, Exhibit 9. Dr. Israel ran the same analysis but replaced Dr. Hatzitaskos's market shares with his own estimates, arriving at a lower price increase of 4.5%. DX0076 at 173 & fig.26. And when Dr. Israel adjusted the margins used by Dr. Hatzitaskos, he found an even lower price increase of 2.4%. Id. at 173-75 & fig.26. For the reasons mentioned above, the Court does not find Dr. Israel's criticisms of the inputs used by Dr. Hatzitaskos-in terms of both market shares and margins-to be compelling. Regardless, it is notable that even using Dr. Israel's modified inputs, the merger simulation still suggests a post-merger price increase. Cf. Aetna, 240 F.Supp.3d at 46-47 (noting expert's conclusion that "all iterations of the merger simulation point toward a price increase following the merger" and finding that merger simulation provided additional evidence of anticompetitive effects where results were consistent with other evidence such as high HHI scores and "observed competition" between insurers).
Along with his criticisms of Dr. Hatzitaskos's analysis, Dr. Israel testified that he ran his own test on whether Lasso's entry into the market had any effect on Deeplntent's pricing. Tr. 1082-83 (Israel). The FTC's theory, he explained, is that "if you eliminate competition between Lasso and Deeplntent, prices will go up. The counterpoint to that, an equivalent statement is, if you introduce competition between Lasso and Deeplntent, prices should go down." Tr. 1083 (Israel). Based on his analysis, however, he did not find that Lasso's entry into the market had "any systemic effect" on Deeplntent's prices. Tr. 1083-84 (Israel). Dr. Hatzitaskos raised a number of technical issues with this study-such as insufficient data points-in his reply report. PX6504 at 77-79. He also asserted that Dr. Israel's analysis was conceptually flawed because "[e]ven under perfect competition, the 'test' would not yield the result that Dr. Israel claims it would." Id. at 75. Dr. Hatzitaskos used the example of two gas stations that are in "perfect competition" but might nonetheless see their profits move up or down in tandem based on myriad factors. Id. at 75-76. More fundamentally, however, the Court does not find Dr. Israel's analysis compelling because it is plainly at odds with the overwhelming documentary evidence showing that Deeplntent and Lasso compete on price and have lowered their prices in direct response to each other's offerings. Even Dr. Israel, while noting competition between "lots of players," conceded that there are "lots and lots of documents in this record showing competition between Deeplntent and Lasso." Tr. 1084 (Israel); see also Tr. 1142 (Israel) ("There are specific documents where there is discussion of prices. Yes, again, there is a variety of those with different firms. I think it is correct that there are some with Deeplntent and Lasso."); Tr. 618 (Paquette) (Deeplntent CEO agreeing that "Deeplntent has lowered its rates to match Lasso's pricing").
All told, the Court largely finds Dr. Hatzitaskos's analysis of the merger's anticompetitive effects to be persuasive. That conclusion is reinforced by the fact that his findings are broadly consistent with the documentary evidence and witness testimony. See Sysco, 113 F.Supp.3d at 64 (noting that expert's conclusions were "corroborated by other evidence in the record" indicating that the merging parties were "close competitors"); see also Anthem, 236 F.Supp.3d at 219 ("[Defendant's] ordinary course documents tell a consistent story that contravenes the firm's litigation position.").
The Court is cognizant of the fact that it "need not decisively sift through various models and theories." Peabody, 492 F.Supp.3d at 907 (quoting Tronox, 332 F.Supp.3d at 212). At this stage, the question is not whether the FTC has established that the proposed acquisition violates the Clayton Act; it is whether the FTC has raised questions about the antitrust merits that are substantial enough to warrant consideration by the Commission itself in the first instance. The Court concludes that the FTC has cleared that threshold. The post-merger market shares and market concentration discussed above give rise to a presumption that the merger will impair competition. And that conclusion is strengthened by the considerable evidence of head-to-head competition between Deeplntent and Lasso-which would be eliminated by the merger-as well as the quantitative evidence offered by Dr. Hatzitaskos. The FTC has made out a sufficiently strong prima facie case of anticompetitive effects.
4. Defendants' Rebuttal Evidence
Once the FTC has established a presumption that the proposed merger will substantially lessen competition, Defendants "may rebut that presumption by showing that the traditional economic theories of the competitive effects of market concentration are not an accurate indicator of the merger's probable effect on competition or that the procompetitive effects of the merger are likely to outweigh any potential anticompetitive effects." Sysco, 113 F.Supp.3d at 72. Defendants can do so "by affirmatively showing why a given transaction is unlikely to substantially lessen competition, or by discrediting the data underlying the initial presumption in the government's favor." Baker Hughes, 908 F.2d at 991. "The more compelling the prima facie case, the more evidence the defendant must present to rebut it successfully." Id.; accord, e.g., Wilhelmsen, 341 F.Supp.3d at 66. Because the FTC has established a strong prima facie case, "Defendants have an uphill climb to rebut the resulting presumption that the [proposed acquisition] will harm competition." Peabody, 492 F.Supp.3d at 908.
Defendants have raised rebuttal arguments based on (1) the inability of current market shares to predict future competition; (2) ease of entry into the market; (3) the sophisticated customers in the market; and (4) efficiencies that will result from the transaction. Doc. 288 at 74-87. The Court will evaluate each of these arguments in turn. In the end, the Court concludes that Defendants' rebuttal arguments are not sufficient to overcome the strong prima facie case of anticompetitive effects that the FTC has established.
a. Market Shares and Future Competition
First, Defendants argue that current market shares "are not indicative of future competition in this dynamic and competitive industry." Id. at 74-75. They assert that the field of digital healthcare advertising is "dynamic and fast-moving." Id. at 76. According to Defendants, HCP programmatic advertising "did not even exist ten years ago, and was still in its infancy when COVID-19 upended traditional business practices." Id. Firms are rapidly innovating to provide new offerings for customers, while some agencies are developing tools that bypass DSPs altogether. Id. Defendants suggest that the market presence of the merging parties illustrates this dynamism: Lasso has grown dramatically since entering the market, while Deeplntent's share in HCP programmatic advertising has been trending downward. Id. As a result of these trends, Defendants conclude, "a static snapshot of market shares is not an accurate predictor of what the market will look like even just one year from now." Id.
At the hearing, Defendants offered expert testimony on this subject from Dr. Yael Hochberg, who holds a Ph.D. in business administration and is the Ralph S. O'Connor Professor in Entrepreneurship and Professor of Finance at the Jones Graduate School of Business at Rice University. DX0077 (Hochberg Report) at 3. Dr. Hochberg was asked to evaluate Dr. Hatzitaskos's conclusion that the market for HCP programmatic advertising is a stable market. Tr. 985 (Hochberg). She disagreed with his view, identifying several examples of dynamic change in the industry. First, some generalist DSPs "have looked at the HCP programmatic and the healthcare programmatic advertising industry in general and seen opportunity for entry into that market." Tr. 991 (Hochberg). Second, publishers of advertising have begun to integrate the data from their publishing inventory with various parts of the service of programmatic advertising. Tr. 992 (Hochberg). The publisher Haymarket, for example, launched a product that "effectively skips the DSP and goes straight to the supply-side platform ... to effectively offer programmatic on their inventory." Tr. 992-93 (Hochberg). Third, ad agencies have taken steps to integrate certain parts of the programmatic advertising process directly into their own offerings rather than contracting with a third party. Tr. 993-94 (Hochberg). Finally, there is increasing accessibility in the market insofar as several components that used to be differentiators-such as the ability to identify and target HCPs to build an audience list-are more readily available for purchase. Tr. 996-98 (Hochberg).
In addition to these changes, Dr. Hochberg pointed to investment capital flowing into the industry. Tr. 998 (Hochberg). Based on private equity firms' investments in companies such as Doceree-a healthcare-focused DSP-she opined that these firms "are seeing an opportunity for new entrants to compete in new ways with new technologies." Tr. 999 (Hochberg). Dr. Hochberg also observed that recent acquisitions in the market add "one more data point" showing "how much this industry is changing." Tr. 1002-03 (Hochberg). While Dr. Hochberg noted that she had not evaluated revenues or market shares for various industry participants, she asserted that it was not necessary to do so in order to reach her conclusion that the market is not stable. Tr. 1006 (Hochberg) ("There is so much happening in this industry. We don't need to look at market shares, and I don't think that they would be informative, given the forces that are at play in the market.").
As a general matter, the Court agrees with Defendants that the market for HCP programmatic advertising has experienced many changes since its relatively recent inception. But the Court does not agree that those changes render the market so unstable that the Court cannot rely on recent revenues and current market shares to guide its analysis. As Dr. Hatzitaskos explained, while merger review is a forward-looking exercise, "economists recognize that historical revenues are typically the best available predictor of future competitive significance." PX6504 at 83; see also Tr. 851 (Hatzitaskos) ("[M]ore recent revenues are going to be more informative about what is going to happen in the future."). While there may be new entrants into the market going forward, that does not necessarily compel the conclusion that current market shares are unreliable. See Meta, 654 F.Supp.3d at 924 (explaining that "new entrants do not necessarily result in shifting or deconcentrating market shares").
Furthermore, Dr. Hatzitaskos identified trends in the revenue data indicating that future market shares are unlikely to significantly differ from the current picture. For one, the market leaders in revenue remained the same over the periods Dr. Hatzitaskos analyzed-that is, comparing revenues in 2022 to the first half of 2023. PX6504 at 53-56 & Exhibit R-3. PulsePoint, Deeplntent, and Lasso make up a clear Big 3 for both periods, while other firms in the competitive "fringe"-the firms in the market outside of the Big 3-lag far behind. Id; cf. Stanley Works v. FTC, 469 E2d 498, 504 (2d Cir. 1972) ("Where 'concentration is already great, the importance of preventing even slight increases in concentration and so preserving the possibility of eventual deconcentration is correspondingly great.'" (quoting Phila. Nat'l Bank, 374 U.S. at 365 n.42)).
The Court takes Defendants' point that Dr. Hatzitaskos analyzed only an eighteen-month window and was working with limited data. But the conclusions he reached are consistent with the documentary evidence discussed above demonstrating that the Big 3 stand apart from others in their HCP programmatic offerings. Even over a longer period, moreover, Dr. Hatzitaskos observed that several members of the competitive fringe have seen only modest growth or a decrease in their annual revenues from HCP programmatic advertising. PX6504 at 54. For example, the largest member of the competitive fringe-(Redacted)-saw revenues from (Redacted) Id. That is particularly notable given that HCP programmatic advertising became more prevalent over the same period due to the COVID-19 pandemic. Id. Other firms in the competitive fringe have likewise seen their revenue decrease in recent years. Id. (collecting examples). While (Redacted) has seen recent growth, that growth has been relatively modest-an from (Redacted) in the first half of 2023. Id. at 54, 56 & Exhibit R-3. These trends undermine Defendants' characterization of the market as so dynamic and volatile that relying on current market shares would be misleading. And Defendants have offered no evidence of their own to suggest that historical shifts in market shares are significant enough to render recent revenue data unusable for antitrust analysis. See Meta, 654 F.Supp.3d at 924 (noting that "Defendants have not presented evidence of actual historical shifts in shares for the relevant market here").
Defendants next argue that "evidence that a merging party is unlikely to 'compete effectively in the future may serve to rebut a presumption that the merger would have anticompetitive effects.'" Doc. 288 at 77 (quoting Deutsche Telekom, 439 F.Supp.3d at 217). While Deeplntent's business was originally focused on HCP programmatic advertising, Defendants explain, it has now shifted much of its focus to the DTC side. Id. As a result, "Deeplntent's revenues in HCP programmatic advertising have been 'flatlin[ing],' and Deeplntent's market share therefore is dropping." Id. (alteration in original) (citation omitted). Accordingly, Defendants deem it "unlikely" that the post-merger market shares alleged by the FTC will persist into 2024. Id.
As a preliminary matter, this argument is in at least some tension with Defendants' position that the market for HCP programmatic advertising is dynamic and characterized by constant innovation. If that is so, it is hard to see why the Court should dismiss out of hand Deeplntent's ability to continue innovating and competing in the HCP market. More importantly, the Court is not persuaded by Defendants' attempts to minimize Deeplntent's market position. It is true that the company's share of the market for HCP programmatic advertising declined by approximately three percentage points from 2022 to the first half of 2023, PX6504 at 56, Exhibit R-3, and that its HCP revenues are flattening out to some extent, Tr. 1077-78 (Israel); DX0076 at 140 & fig. 15. Looking at the broader picture of the market, however, Deeplntent remains an extremely significant player, with 25.4% market share in the first half of 2023-more than five times higher than the largest non-Big 3 firm in the market. PX6504 at 56, Exhibit R-3. Furthermore, courts have recognized that this type of "weakened competitor" defense "is among the weakest grounds for rebuttal" and is credited "only in rare cases." Deutsche Telekom, 439 F.Supp.3d at 218. Nothing in the record indicates that Deeplntent's slight decline on the HCP side "cannot be resolved by any competitive means [and] would cause [Deeplntent's] market share to reduce to a level that would undermine the government's prima facie case." Id. (first alteration in original) (quoting ProMedica Health Sys., Inc. v. FTC, 749 E3d 559, 572 (6th Cir. 2014)). And this is not the kind of case where Deeplntent faces a "lack of resources required to compete long-term, financial difficulties that constrain the firm from improving its competitive position, and poor brand image and sales performance." Id. at 217-18 (collecting cases).
Defendants also contend that the Court's predictive analysis must account for "whether there are substitutes for the merging firms' products such that prices are constrained." Doc. 288 at 77. They maintain that Deeplntent and Lasso "are hemmed in on all sides by competition," including from DSPs, publishers, social media platforms, data providers, and advertising agencies themselves. Id. at 78. Citing United States v. General Dynamics Corp., 415 U.S. 486 (1974), Defendants argue that this is true even if those other firms are excluded from the relevant product market. Doc. 288 at 78.
The Court, however, has already accounted for "whether there are substitutes for the merging firms' products." For the reasons previously explained in the section on market definition, the substitutes proposed by Defendants are not reasonably interchangeable for the services offered by Deeplntent and Lasso. The purpose of the reasonable interchangeability analysis is to determine whether products are readily substitutable such that they could provide an alternative for customers in the event of a price increase. To the extent there may be any competitive pressure exerted by Defendants' proposed alternatives, it is not sufficient to rebut the FTC's strong prima facie case. In addition, Defendants' reliance on General Dynamics is misplaced. In that case, the district court concluded that "coal faced strong and direct competition from other sources of energy such as oil, natural gas, nuclear energy, and geothermal power which created a cross-elasticity of demand among those various fuels." Gen. Dynamics, 415 U.S. at 491. As a result, the court found that "coal, by itself, was not a permissible product market." Id. That was the context in which the Supreme Court-in language quoted in Defendants' post-hearing brief, Doc. 288 at 78-referenced the district court's finding that coal companies were facing "increasingly stiffer competition from oil and natural gas as sources of energy for industrial and residential uses." Gen. Dynamics, 415 U.S. at 499. The reference, then, was to competition within the relevant product market. See Id. at 491 (noting district court's finding that "the 'energy market' was the sole 'line of commerce' in which anticompetitive effects could properly be canvassed"). Here, by contrast, the Court has already concluded that the substitutes identified by Defendants are not within the relevant market.
Defendants return to their contention that ad agencies would simply shift their spending if the merged firm sought to raise prices. Doc. 288 at 79. This argument fails for the reasons already discussed-namely, that HCP programmatic advertising provides unique features that customers would not readily abandon. Defendants are also wrong to suggest that "the FTC never actually analyzes how brands and agencies would respond to a price increase." Id.; see, e.g., (Redacted) (declaration from pharmaceutical company executive stating that HCP programmatic advertising is "an important component of [the company's] overall advertising strategy, and it is not something the company would walk away from unless prices increased substantially").
Ultimately, the Court's task "is to assess the alleged antitrust violations presented, irrespective of the dynamism of the market at issue." Bazaarvoice, 2014 WL 203966, at *76. The Court is mindful of the fact that the HCP programmatic advertising market is young and continues to evolve. But even if Defendants are operating in a dynamic industry, they have nonetheless failed to "present evidence that the evolving nature of the market itself precludes the merger's likely anticompetitive effects," id, or a court's ability to analyze the same. Accounting for the "structure, history and probable future" of the industry, Brown Shoe, 370 U.S. at 322 n.38, the Court finds that Defendants have not rebutted the FTC's showing of a reasonable probability that the proposed acquisition will substantially lessen competition in the market for HCP programmatic advertising.
b. Ease of Entry
Defendants' second rebuttal argument focuses on ease of entry into the market. A defendant may attempt to rebut the government's prima facie case by introducing evidence "that entry by new competitors will ameliorate the feared anticompetitive effects of a merger." Aetna, 240 F.Supp.3d at 52. The Merger Guidelines require consideration of whether "entry would be timely, likely, and sufficient in its magnitude, character, and scope to deter or counteract the competitive effects of concern." Merger Guidelines § 9. Defendants take issue with this standard, Doc. 288 at 83, but it has been applied by several courts in similar cases, see, e.g., Deutsche Telekom, 439 F.Supp.3d at 226 ("[T]he Merger Guidelines provide that new market entry may counteract concerns about anticompetitive effects if entry would be 'timely, likely, and sufficient in its magnitude, character, and scope' to address those concerns." (quoting Merger Guidelines § 9)); United States v. Visa USA, Inc., 163 F.Supp.2d 322, 342 (S.D.N.Y.2001) ("The higher the barriers to entry, and the longer the lags before new entry, the less likely it is that potential entrants would be able to enter the market in a timely, likely, and sufficient scale to deter or counteract any anticompetitive restraints."); see also Aetna, 240 F.Supp.3d at 52-53 (collecting cases). That is true with respect to expansion as well as entry. See, e.g., Wilhelmsen, 341 F.Supp.3d at 67 ("The expansion of current competitors is regarded as 'essentially equivalent to new entry,' and is therefore evaluated according to the same criteria." (citation omitted)); H&R Block, 833 F.Supp.2d at 73 (invoking same standard in discussing "the likelihood of expansion by existing competitors rather than new entry into the market").
In this case, internal documents from both Deeplntent and Lasso recognize substantial barriers to entry. PX2581-24 (Deeplntent presentation listing several "primary barriers to entry" including "technical and regulatory complexity of integrating healthcare data within advertising"; "talent scarcity at intersection of healthcare and programmatic"; "deep agency and client integrations and contracts"; and "patents" (capitalization omitted)); PX2504-18 (Deeplntent presentation stating that "barrier to entry remains high for healthcare"); PX1128-12 (Lasso presentation stating that "Lasso's unique and industry-leading healthcare marketing and analytics platform provides significant barriers-to-entry," including that "[infrastructure takes years and millions of dollars to build"). These documents are probative of the significant barriers that a new entrant would face. See CCC Holdings, 605 F.Supp.2d at 49-50 (relying on documents in which defendants had repeatedly touted barriers to entry); see also Bazaarvoice, 2014 WL 203966, at *49 (citing defendant's pre-acquisition statements about barriers to entry).
Industry participants confirmed in their testimony that new entrants face significant challenges. One witness from a generalist DSP-AdTheorent-explained that, since 2020, he was "not aware of any new entrants that we come up against when we are competing for budgets in the market today other than PulsePoint, Deeplntent, and Lasso." Tr. 808 (Lawson). He elaborated:
There is a pretty steep learning curve, data curve, and a number of other factors that make it harder for a new company to get into the space: Access to their correct data, expertise around the proxy laws that are relevant to targeting advertisements to, for example, patients. . . . Generalists don't have the types of expertise and knowledge in products and solutions tailored to help, based on top of the health-specific data relevant to health-specific [key performance indicators] for campaign goals.Tr. 809 (Lawson). Similarly, PulsePoint's testimony highlighted the expertise that is required for firms providing HCP programmatic advertising: "From a capability perspective, the generalist DSPs are lacking certain platform capabilities as they relate to targeting, optimization, that are important for executing HCP digital marketing at a competitive price and scale." Tr. 157-58 (Gerszke); see also Tr. (Redacted) (ad agency witness noting that Deeplntent, Lasso, and PulsePoint all specialize in healthcare and that "[t]he healthcare field is highly regulated so working with somebody that knows that space, whether it was HCP or patient, gave myself, the team a great deal of comfort and confidence in relying on their platforms versus others in the market").
Attempting to downplay these barriers to entry, Defendants point to Lasso's rapid ascent as evidence that new firms can easily enter the market and find success. Doc. 288 at 80. While Lasso's trajectory demonstrates that entry is possible, it fails to establish that future entry will be timely, likely, and sufficient to counteract the anticompetitive effects of the proposed acquisition. The weight of the evidence indicates that barriers to entry are significant and that firms may face unique challenges in attempting to break into the HCP programmatic advertising space. Defendants have not offered any reason to think that Lasso's rise to its prominent market position is likely to be replicated. Cf. Wilhelmsen, 341 F.Supp.3d at 68 (defendants' contention that merging party's business was "simple and capable of replication in a short period of time is at odds with inferences drawn from the state of the current market and with documentary and testimonial evidence from customers and suppliers").
Nor sufficient that generalist DSPs such as (Redacted) have "aspirations" to expand their HCP programmatic advertising business. Doc. 288 at 81. Consistent with the barriers to entry already discussed, customer testimony indicates that generalist DSPs currently lack some of the capabilities needed to succeed in HCP programmatic advertising. Tr. (Redacted) (ad agency witness explaining that the agency did not include (Redacted) in a recent RFP due to "their capability in the health care professional focused marketplace" and that he would not currently consider (Redacted) to be a viable substitute for the agency's business with Lasso and PulsePoint); see Staples, 190 F.Supp.3d at 134 (rejecting defendants' ease of entry argument in part because customers did not view the potential entrant "as a viable to [the parties]"). (Redacted) As Dr. Hatzitaskos put it: "[I]f we see [some of these players] making only a tiny fraction of the revenues of the merging parties, and some of them have been in the market since 2019 or early on, that means that they are lacking in capabilities, sort of the proof is in the pudding." Tr. 890 (Hatzitaskos).
Defendants also suggest that the mere threat of entry into the market or expansion by existing firms is enough to provide a competitive constraint and rebut the FTC's prima facie case. Doc. 288 at 82-83. As discussed above, the standard is that entry must be "timely, likely, and sufficient." But even setting that aside, the Court is not persuaded by Defendants' argument that firms like Google, Yahoo, and Microsoft pose a competitive threat merely because they are "already involved in HCP programmatic advertising to varying degrees." Id. at 82. "The marketplace may be filled with many strong and able companies in adjacent spaces. But that does not mean that entry barriers become irrelevant or are somehow more easily overcome." Bazaarvoice, 2014 WL 203966, at *71.
Some courts have directly questioned whether the mere threat of entry is sufficient. See Chi. Bridge & Iron Co. N. V. v. FTC, 534 F.3d 410, 430 n. 10 (5th Cir. 2008) ("Baker Hughes' conclusion that a mere threat of entry is sufficient to constrain anti-competitive effects has been criticized, and we will not adopt it here.").
Finally, Defendants place significant emphasis on the Second Circuit's decision in Waste Management, but the circumstances in that case are not analogous to those present here. At issue was a proposed acquisition involving two companies in the waste disposal business. See 743 E2d at 977-78. The district court concluded that the relevant product market included all trash collection, except for collection at certain residences. Id. at 978. After affirming that finding, the Second Circuit turned to the defendants' rebuttal arguments. Id. at 980-81. The court held that "entry into the relevant product and geographic market by new firms or by existing firms in the Fort Worth area is so easy that any anti-competitive impact of the merger before us would be eliminated more quickly by such competition than by litigation." Id. at 983. In fact, the district court had found that "individuals operating out of their homes can acquire trucks and some containers and compete successfully 'with any other company.'" Id. There were "examples in the record of such entrepreneurs entering and prospering." Id. Entry by larger companies, likewise, would be "relatively easy" because "Fort Worth haulers could easily establish themselves in Dallas if the price of trash collection rose above the competitive level." Id. Thus, the merged firm would not be able to exercise market power due to "the ease with which new competitors would appear." Id. at 983-84. In this case, by contrast, the record makes clear that market entry is not nearly as simple as individuals "operating out of their homes" being able to compete with established players. That much should be clear from the fact that IQVIA has proposed (Redacted) to complete the consolidation of Deeplntent and Lasso.
c. Sophisticated Customers
Next, Defendants argue that customers in this market-such as healthcare companies and advertising agencies-"are sophisticated customers with extensive knowledge and experience in the industry." Doc. 288 at 85. Courts have recognized that "the existence of power buyers-sophisticated customers who retain strategies post-merger that may constrain the ability of the merging parties to raise prices-is a factor that can serve to rebut a prima facie case of anti-competitiveness." Wilhelmsen, 341 F.Supp.3d at 70 (internal quotation marks and citation omitted). But the ability of these large buyers to keep prices down "depends on the alternatives these large buyers have available to them." Id. (quoting Sysco, 113 F.Supp.3d at 48). Accordingly, "the mere presence of power buyers 'does not necessarily mean that a merger will not result in anticompetitive effects.'" Id. (quoting FTC v. Cardinal Health, Inc., 12 F.Supp.2d 34, 59 (D.D.C. 1998)). Courts assessing such a defense must "examine the choices available to powerful buyers and how those choices likely would change due to the merger." Id. (quoting Merger Guidelines § 8).
Defendants assert that the relevant customers here "have substantial pricing power in negotiations with advertising platforms and testified that they would expect to negotiate rigorously on price regardless of how many DSPs were offering HCP programmatic advertising services." Doc. 288 at 85. This evidence, according to Defendants, indicates that even a hypothetical monopolist DSP could not raise prices without accounting for agencies' buying power. Id.
This defense is not persuasive for many of the same reasons as Defendants' arguments regarding market definition. The key question is what choices are available to buyers and how those choices would change post-merger. See Wilhelmsen, 341 F.Supp.3d at 70. On balance, the evidence tends to show that customers' choices would be substantially reduced by the proposed acquisition because the number of leading firms at the top of the market would go from three to two. The head-to-head competition between Deeplntent and Lasso would also be eliminated, resulting in less competition on price and innovation and leaving customers with fewer choices when selecting a provider of HCP programmatic advertising services. And even if the presence of power buyers were more compelling, "courts have not typically held 'that power buyers alone enable a defendant to overcome the government's presumption of anticompetitiveness."' Id. (quoting Cardinal Health, 12 F.Supp.2d at 58); see also, e.g., Chi. Bridge & Iron Co. N.V. v. FTC, 534 F.3d 410, 440 (5th Cir. 2008) ("[C]ourts have not considered the 'sophisticated customer' defense as itself independently adequate to rebut & prima facie case."). The defense is not sufficient in this case either.
d. Efficiencies
Finally, Defendants point to three purported efficiencies that will be created by the proposed acquisition. Doc. 288 at 85-87. First, the deal will reduce costs for the merged firm. Id. at 86. Second, the deal will benefit customers because it will expand the merged firm's capabilities. Id. at 86-87. And third, the deal will "further IQVIA's objective of enhancing patient care and producing better health outcomes for consumers who can more easily and affordably find relevant information about a needed medication or medical device." Id. at 87.
At the outset, "it is not at all clear that [efficiencies] offer a viable legal defense to illegality under Section 7." United States v. Anthem, Inc., 855 F.3d 345, 353 (D.C. Cir. 2017). In FTC v. Procter & Gamble Co., 386 U.S. 568 (1967), the Supreme Court held that "[p]ossible economies cannot be used as a defense to illegality. Congress was aware that some mergers which lessen competition may also result in economies but it struck the balance in favor of protecting competition." Id. at 580. The D.C. Circuit has noted that Procter & Gamble remains good law. See Anthem, 855 F.3d at 353-54. And other circuits have likewise expressed skepticism about the availability of an efficiencies defense. See Perm State, 838 E3d at 348 ("[W]e are skeptical that such an efficiencies defense even exists."); Saint Alphonsus, 778 E3d at 790 ("We remain skeptical about the efficiencies defense in general and about its scope in particular. It is difficult enough in § 7 cases to predict whether a merger will have future anticompetitive effects without also adding to the judicial balance a prediction of future efficiencies.").
Nevertheless, several courts have analyzed proposed efficiencies to determine whether they might rebut a prima facie case. See Anthem, 855 E3d at 354-55 (collecting cases); see also Deutsche Telekom, 439 F.Supp.3d at 208 ("Despite the skepticism that some courts have expressed and the lack of Second Circuit precedent on point, this Court will consider evidence of efficiencies, given courts' and federal regulators' increasingly consistent practice of doing so, and because Section 7 requires evaluation of a merger's competitive effects under the totality of the circumstances.").
Here too, the Court will assume that an efficiencies defense is available. But Defendants must overcome a high bar. See, e.g., Heinz, 246 F.3d at 720 (noting that high market concentration levels require "proof of extraordinary efficiencies"). To be cognizable, the efficiencies must "offset the anticompetitive concerns in highly concentrated markets." Penn State, 838 F.3d at 348. In addition, the efficiencies must be "merger specific," meaning that they "cannot be achieved by either company alone." Id. (quoting Heinz, 246 F.3d at 722). The efficiencies also must be "verifiable, not speculative." Id. (quoting Saint Alphonsus, 778 F.3d at 791). And the efficiencies "must not arise from anticompetitive reductions in output or service." Id. at 349.
None of Defendants' proposed efficiencies satisfy this rigorous test. With respect to reduced costs, Defendants point to two categories of anticipated savings. First, the acquisition would eliminate the need for Lasso to pay Xandr to provide DSP functionalities. Doc. 288 at 86 ("Combining Lasso's omnichannel advertising platform with Deeplntent's DSP technology would reduce the need for that licensing arrangement [with Xandr] and save costs, which could be passed on to customers."). Second, Deeplntent currently pays fees to license data from IQVIA that could be lessened or eliminated by the merger. Id. Defendants largely rely on testimony from an IQVIA executive about these savings. He stated:
We believe there are considerable efficiencies. On the cost side, there are synergies and efficiencies that are driven from the ability to have an incremental option on the DSP side. Lasso rents its space from Microsoft Xandr today, so we pay a premium on that. So the ability to create an option for our clients that would allow that to be done at a non-marked up cost would create synergy.
And the other place is data. . . . [B]y providing data on our platform, we think that there will be an opportunity to get it without markups that other people, that other participants would put on top of it, so incremental efficiencies.Tr. 1228 (Resnick). In addition, Dr. Israel conducted a merger simulation and concluded-after adjusting Dr. Hatzitaskos's shares and margins-that the synergies from the proposed acquisition would lead to lower prices for Deeplntent's and Lasso's customers. DX0076 at 230-33 & fig.34.
For the reasons explained in the section discussing effects on competition, the Court does not find Dr. Israel's adjustments of the inputs used by Dr. Hatzitaskos to be persuasive, and therefore the Court cannot credit his merger simulation model suggesting that the acquisition would lead to lower prices. According to Dr. Hatzitaskos, moreover, the data cost savings estimated by Dr. Israel were roughly twice as high as the merging parties' own projections. PX6504 at 110-11. Even if Dr. Israel had used Defendants' estimates, the Court would not be able to credit those internal figures as sufficiently verifiable for purposes of the efficiencies analysis. See Wilhelmsen, 341 F.Supp.3d at 73 ("The court cannot substitute Defendants' assessments and projections for independent verification."); H&R Block, 833 F.Supp.2d at 91 ("While reliance on the estimation and judgment of experienced executives about costs may be perfectly sensible as a business matter, the lack of a verifiable method of factual analysis resulting in the cost estimates renders them not cognizable by the Court."). Defendants have not provided any indication that their internal estimates of these savings were independently verified. Cf. Peabody, 492 F.Supp.3d at 916 (while the court noted that it was "concerned about the fundamental unverifiability of efficiencies that are grounded in the business judgments of Defendants' employees," it also observed that the defendants had "supported their claims with evidence from past transactions"); Sysco, 113 F.Supp.3d at 82 (defendants' efficiencies estimate was the product of an eight-month analysis conducted by an independent consulting firm).
As for Defendants' second and third anticipated efficiencies, both are related to the merged firm's improved capabilities and product offerings. Specifically, Defendants argue that the merger would help IQVIA "compete with larger firms" by "combining Lasso's omnichannel marketing platform with Deeplntent's DTC capabilities and DSP technology." Doc. 288 at 87. Defendants further assert that the deal would enhance IQVIA's ability to provide improved, lower-cost services to customers. Id. In turn, "[b]y providing better and more efficient advertising, advertisers can synchronize information to patients and HCPs, driving better patient outcomes." Id. On this point, Defendants presented expert testimony from Dr. Anupam B. Jena, who is a physician at Massachusetts General Hospital and holds a Ph.D. in economics. DX0075 (Jena Report) at 4. Dr. Jena opined that a "more unified approach" to HCP and DTC advertising "would benefit a lot of different healthcare parties," including drug companies, marketing agencies, doctors, and patients. Tr. 1310-11 (Jena). He explained that there are benefits to having "a coordinated message between the HCP and the patient." Tr. 1313 (Jena). For instance, both a doctor and a patient might see targeted advertisements for a new drug-with "some uniform messaging that's happening at the same time"-and then discuss it during a patient visit. Tr. 1314 (Jena).
As the Court noted earlier, the FTC moved in limine to exclude some of Dr. Jena's opinions. Doc. 215. The Court has considered Dr. Jena's testimony but ultimately concludes that Defendants cannot make out an efficiencies defense. Accordingly, the FTC's motion in limine is denied as moot.
As a threshold matter, it is not enough to show-even assuming Defendants had made such a showing-that the merger would allow IQVIA to better serve its customers or even patients and doctors. See Saint Alphonsus, 778 E3d at 791 ("It is not enough to show that the merger would allow St. Luke's to better serve patients."). "The Clayton Act focuses on competition, and the claimed efficiencies therefore must show that the prediction of anticompetitive effects from the prima facie case is inaccurate." Id. Defendants cannot successfully make out an efficiencies defense based on claims that the acquisition will allow them to provide better services. See Id. at 792 (observing that hospital's ability to provide better services post-merger was "a laudable goal, but the Clayton Act does not excuse mergers that lessen competition or create monopolies simply because the merged entity can improve its operations"); see also Anthem, 236 F.Supp.3d at 252 ("[N]o court has held that a potential general benefit to consumers at the end of the day can negate competitive harm; what precedent there is states precisely the opposite."); cf. Penn State, 838 E3d at 351 ("An efficiencies analysis requires more than speculative assurances that a benefit enjoyed by the Hospitals will also be enjoyed by the public").
Aside from that issue, Defendants also fail to demonstrate that these efficiencies are verifiable. It is not clear how an independent party would be able to verify Defendants' claims that the merger will allow them to compete more effectively with larger firms or further their objective to enhance patient care. Defendants do not cite any cases-in either their pre-hearing or post-hearing briefing-recognizing an efficiencies defense based on the kind of subjective, nonquantifiable claims of efficiencies that they have advanced here. Cf. Univ. Health, 938 E2d at 1223 (rejecting efficiencies defense because defendants "simply concluded that the intended acquisition would reduce 'unnecessary duplication' between University Hospital and St. Joseph; they then approximated, in dollars, the savings these efficiencies would produce. They did not specifically explain, however, how these efficiencies would be created and maintained").
The Court does not find persuasive Defendants' argument that the combination of Deeplntent's and Lasso's purportedly complementary capabilities will allow IQVIAto compete more effectively with "larger firms." Doc. 288 at 87. As the extensive documentary record illustrates, Deeplntent and Lasso are most appropriately classified not as complements but as direct head-to-head rivals who routinely compete for customers' business and strive to best one another on price and innovation. Even if the two firms have some features that might be considered complementary, Defendants have not shown that those complementary capabilities will translate into an improved competitive offering that would be sufficient to counteract the increased market concentration and removal of direct competition likely to result from the acquisition. And it is difficult to see why IQVIA, with all its resources, could not on its own improve Lasso's capabilities on the DTC side, meaning that the asserted efficiencies also fail to satisfy the merger-specific requirement. See Heinz, 246 E3d at 722 (noting that efficiencies are sufficient only if they "cannot be achieved by either company alone because, if they can, the merger's asserted benefits can be achieved without the concomitant loss of a competitor"). Ultimately, the Court is not persuaded that any potential benefits provided by the combined firm's offerings will be sufficient to mitigate the high level of market concentration and the removal of head-to-competition that the merger is likely to produce. See generally Aetna, 240 F.Supp.3d at 95 ("On balance, the Court is unpersuaded that the efficiencies generated by the merger will be sufficient to mitigate the transaction's anticompetitive effects for consumers in the challenged markets.").
On the whole, Defendants have not demonstrated that any of their asserted efficiencies will be sufficient to counteract the anticompetitive effects of the transaction.
In the final analysis, Defendants' rebuttal arguments fail to overcome the FTC's strong prima facie case of anticompetitive effects. Accordingly, the FTC has carried its burden to demonstrate that it is likely to succeed on the merits of its horizontal challenge. The Court thus proceeds to consideration of the equities.
B. Equities
The FTC's demonstration of a likelihood of success "creates a presumption in favor of preliminary injunctive relief." Penn State, 838 F.3d at 352 (quoting Heinz, 246 F.3d at 726). Under section 13(b), however, the Court "must still weigh the equities in order to decide whether enjoining the merger would be in the public interest." Id. (quoting Heinz, 246 F.3d at 726). Because the Court has concluded that the transaction is likely to substantially lessen competition, Defendants "face a difficult task in justifying the nonissuance of a preliminary injunction." Id. (quoting Univ. Health, 938 F.2d at 1225); see also Peabody, 492 F.Supp.3d at 918 (noting that '"no court has denied a Section 13(b) motion for a preliminary injunction based on weight of the equities' where the FTC has demonstrated a likelihood of success on the merits" (citation omitted)).
As other courts have recognized, section 13(b) requires consideration of the equities but "is silent as to what specifically those equities are." Penn State, 838 F.3d at 352. "The prevailing view is that, although private equities may be considered, they are not to be afforded great weight." Id. In other words, where the FTC has demonstrated a likelihood of ultimate success, "a countershowing of private equities alone would not suffice to justify denial of a preliminary injunction barring the merger." Id. (quoting FTC v. Weyerhaeuser Co., 665 F.2d 1072, 1083 (D.C. Cir. 1981)).
Here, the public interest in effective enforcement of the antitrust laws weighs heavily in favor of an injunction. See, e.g., Peabody, 492 F.Supp.3d at 918; Sysco, 113 F.Supp.3d at 86. "The purpose of the preliminary injunction is to preserve the FTC's ability to obtain effective relief if the [transaction] is ultimately found to violate Section 7 of the Clayton Act." Peabody, 492 F.Supp.3d at 918. Although Defendants make a passing argument that it would not be difficult to unwind the transaction, Doc. 288 at 99, the only testimony on that point was more equivocal than they suggest. An IQVIA executive testified that it would take time to integrate Deeplntent, that other IQVIA acquisitions have resulted in divestiture years later, and that Deeplntent's intellectual property, client accounts, and employees could all be separated if necessary. Tr. 1232-33 (Resnick). But even where "it may not be impossible to order divestiture, courts have repeatedly recognized that it is difficult to do so, especially considering the practical implications of denying the preliminary injunction request." Penn State, 838 F.3d at 353 n.ll; see, e.g., Heinz, 246 F.3d at 726 ("Section 13(b) itself embodies congressional recognition of the fact that divestiture is an inadequate and unsatisfactory remedy in a merger case."); Peabody, 492 F.Supp.3d at 918 ("Allowing the transaction to proceed and then later unscrambling the eggs upon a finding of illegality by the FTC is a daunting and potentially impossible task, which supports the issuance of an injunction." (internal quotation marks and citation omitted)).
Defendants also contend that the equities weigh against injunctive relief because of "[t]he loss of the procompetitive benefits that will flow from the transaction." Doc. 288 at 100. As explained above, to the extent the transaction will generate any procompetitive benefits-and the Court has already found that Defendants have not made such a showing-they cannot overcome the substantial harm to competition that the FTC has shown is likely to arise from the acquisition. And Defendants' argument that the deal will not go forward if the injunction is granted does not affect the Court's conclusion. See Penn State, 838 F.3d at 353 ("[A]ccepting the Hospitals' assertion that they would abandon the merger following issuance of the injunction, the result-that the public would be denied the procompetitive advantages of the merger-would be the Hospitals' doing."); Wilhelmsen, 341 F.Supp.3d at 74 ("[A]lthough the court recognizes the time, resources, and effort that Defendants have put into planning this transaction, the parties' stated intention to abandon the transaction prior to the merits proceeding is a private equity, and cannot on its own overcome the public equities that favor the FTC").
The equities thus weigh in favor of injunctive relief "to preserve the status quo while the FTC develops its ultimate case." Whole Foods, 548 F.3d at 1036 (opinion of Brown, J.).
V. CONCLUSION
The Court concludes by returning once more to the legal standard. The inquiry in a section 13(b) proceeding is whether the FTC has raised serious and substantial questions going to the antitrust merits that are fair ground for investigation by the Commission itself in the first instance. See, e.g., Heinz, 246 F.3d at 714-15. To succeed, the FTC "need only show that there is a reasonable probability that the challenged transaction will substantially impair competition." Sysco, 113 F.Supp.3d at 22 (quoting Staples, 970 F.Supp. at 1072). The Court does not decide "whether the FTC will ultimately prove its case" or whether Defendants' rebuttal arguments will prevail. Heinz, 246 F.3d at 727. After considering the record in its entirety, the Court finds only that, at this juncture, the FTC has demonstrated a likelihood of success and that the equities weigh in favor of granting preliminary injunctive relief.
Accordingly, the FTC's motion for a preliminary injunction, Doc. 168, is GRANTED. A separate order accompanies this opinion.
It is SO ORDERED.