KANNE, Circuit Judge.

James E. Dahl, Dahl & Bonadies, LLC, Chicago, IL, for Plaintiff–Appellant.Anthony Hopp, David T. Van Der Laan, Edwards Wildman Palmer LLP, Chicago, IL, for Defendant–Appellee.

Before WOOD, Chief Judge, and KANNE and TINDER, Circuit Judges.

Opinion

KANNE, Circuit Judge.

After losing his job at CentiMark Corporation, Appellant James Turnell went to work for one of its competitors. That created a problem, as Turnell's employment contract contained restrictive covenants barring him from competing with CentiMark or soliciting its customers for two years after termination. The district court issued a preliminary injunction partially enforcing the restrictive covenants, but only to the extent the court deemed necessary to protect CentiMark's legitimate interests. Turnell challenges the preliminary injunction on appeal. We affirm.

I. Background

CentiMark is a large, nationwide roofing company that primarily sells and installs flat, single-ply roofing for commercial and industrial customers. CentiMark hired Turnell as a laborer in its Chicago office in 1978, when he was eighteen years old. Over the course of thirty-five years with the company, Turnell worked his way up the ranks. In 1988 CentiMark promoted him to Chicago District Operations Manager and increased his salary from $45,000 to $50,000 per year. In return, CentiMark required him to sign an employment agreement (the “Agreement”).

Turnell's new role would give him access to CentiMark's proprietary information and trade secrets, including sales methods and materials, customer accounts, and pricing data. So, in return for his promotion and salary increase, Turnell agreed to a non-disclosure provision (§ 4.01 of the Agreement) requiring him to maintain the confidentiality of CentiMark's information.

He also agreed to two restrictive covenants that CentiMark regularly requires its management-level employees to execute. The first is a non-compete clause (§ 4.05) prohibiting Turnell from “engag[ing] ... in any Competing Business” during the period of his employment and for two years afterward in any of the “regions and/or divisions and/or territories” in which he “operated” as a CentiMark employee. “Competing Business” includes any company that “sells or attempts to sell any products or services” that are “the same as, or similar to,” CentiMark's.

The second restrictive covenant is a non-solicitation provision (§ 4.06) providing that Turnell “shall not ... solicit the trade of, or trade with,” any of CentiMark's “customers or suppliers, or prospective customers or suppliers” during his employment and for two years afterward. “Prospective customer” is defined broadly to include anyone “contacted by [CentiMark] during the restrictive periods mentioned in this Agreement.” But the Agreement does not prohibit all contact with customers or prospective customers—only those that “would constitute [Turnell's] engaging in a competitive business, as described in [the non-compete clause].”

After his promotion in 1988, Turnell's ascent within the company continued. He rose from operations manager to branch manager to regional manager and, finally, to Senior Vice President and Regional Manager for the Midwest Region, which encompasses western Michigan, northwest Indiana, central and northern Illinois, Wisconsin, Minnesota, and North and South Dakota. This region is one of CentiMark's largest and most productive, generating over $25 million in annual revenue. Turnell was responsible for all regional sales (including pricing, marketing, and customer relationships) and operations (including staffing, personnel, and financial performance). He had regular contact with customers, reviewed all major proposals, and had access to a password-protected intranet portal containing information on proposals, leads, quotes, financial performance, and other data. He also participated in monthly conference calls with senior executives on company-wide financial projections and strategy. As regional manager, Turnell earned over $250,000 per year in base salary and bonuses; in addition, CentiMark compensated him with shares in the company worth more than $3 million.

The relationship between Turnell and CentiMark came to an end when the company fired him on January 8, 2013. The reason, according to CentiMark, is that Turnell had misappropriated company resources and covered up fraudulent billing by Vacuum Resources, a subcontracting company owned by his wife. According to Turnell, this was a pretext; the real reason for his termination had to do with his age, his health (he has diabetes and high blood pressure), and his high level of compensation.

Within a week after being fired, Turnell interviewed with Windward Roofing and Construction, Inc., a smaller, more local roofing company in Chicago. Commercial roofing accounts for about half of Windward's business and makes it a competitor of CentiMark. But Windward also does other work that CentiMark does not, such as shingled residential roofing and masonry. CentiMark caught wind of Turnell's discussions with Windward and demanded that he cut them off. Turnell testified that he was aware of his contractual obligations but did not care whether he was competing with CentiMark because he “needed a job.” Turnell made little effort to find a job outside commercial roofing.

Instead, he accepted an offer from Windward and began selling commercial roofing for his new employer on or about March 1, 2013. He was hired not to service existing accounts but to develop new business. To that end, Turnell contacted numerous customers or former customers of CentiMark, some of whom he had personally worked with during his tenure there. He sold services to at least one of those customers. And he submitted two bids in head-to-head competition with CentiMark, winning one of those jobs for Windward.

CentiMark demanded that Turnell stop working for its competitor. After an unfruitful exchange of letters, the parties took their dispute to court. Turnell and Windward sued first, on March 11, 2013, seeking a declaration from the Circuit Court of Cook County, Illinois that the Agreement's restrictive covenants were unenforceable. CentiMark removed that lawsuit to the federal district court for the Northern District of Illinois on April 9. The following day, CentiMark filed its own separate action in the Northern District of Illinois against Turnell, Windward, and Vacuum Resources to enforce the restrictive covenants and to obtain other relief. The district court consolidated the two actions.

CentiMark moved for a preliminary injunction against Turnell under Fed.R.Civ.P. 65. After expedited discovery and an evidentiary hearing, the district court granted CentiMark's motion on July 10, 2014. But the court found Turnell's restrictive covenants too broad, so it enforced them only to the extent it judged “reasonably necessary for the protection of CentiMark.” The preliminary injunction reads:

James Turnell shall not sell, attempt to sell, or help sell any products or services, or any combination thereof, related to commercial roofing to any person or entity who was a customer of Centimark Corporation as of January 8, 2013

and who is located in Illinois, Indiana, Michigan, Minnesota, North Dakota, South Dakota, or Wisconsin.

The court provided that “[t]his injunction shall remain in effect until the earlier of a decision on the merits ... or two years from the date of this order.” The court counted from the date of the order rather than the date of Turnell's termination because Turnell's breach tolled the running of the restrictive periods under the Agreement. Finally, the court required CentiMark to post a $250,000 bond.

The preliminary injunction is narrower than the contractual covenants in three respects. First, whereas Turnell's non-compete clause bans all competing business (i.e., sales of products or services similar to CentiMark's), the injunction does not. It allows Turnell to remain in commercial roofing, subject to restrictions on his ability to sell to CentiMark's customers.

Second, the injunction only forbids commercial roofing sales to actual customers of CentiMark as of Turnell's termination date. Unlike the non-solicitation provision, the injunction does not extend to prospective CentiMark customers or to persons who were customers in the past.

Third, the injunction is geographically narrower. There is no geographical limitation in the non-solicitation provision, and the non-compete covers any region where Turnell “operated” as a CentiMark employee. But the district court limited the injunction's reach to the Midwest—the region where Turnell primarily worked throughout his career and which he managed at the time of his departure.

The district court rejected CentiMark's request to enjoin Turnell from working at Windward altogether. So long as he abided by the terms of the preliminary injunction, he was free to remain in his new job. The court also rejected CentiMark's separate claim that Turnell had violated the nondisclosure provision, as there was no evidence of breach. Nevertheless, Turnell took exception to the court's order and filed a timely notice of appeal.

We have subject matter jurisdiction based on diversity of citizenship under 28 U.S.C. §§ 1441(a) -(b) and 1332. And although this is an interlocutory appeal, we have authority to decide it under 28 U.S.C. § 1292(a)(1).

II. Analysis

Federal courts sitting in diversity “apply state substantive law and federal procedural law” under the eponymous Erie doctrine. Hanna v. Plumer, 380 U.S. 460, 465, 85 S.Ct. 1136, 14 L.Ed.2d 8 (1965) (citing Erie R.R. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938) ). The district court here applied Pennsylvania law, pursuant to a choice-of-law clause in the parties' Agreement, to determine the enforceability of the restrictive covenants; and it looked to federal law for the standard governing the availability of a preliminary injunction. Both parties agree with the first decision, and neither party contests the second. (They simply do not address the Erie issue). Choice-of-law arguments are normally waivable, see Vukadinovich v. McCarthy, 59 F.3d 58, 62 (7th Cir.1995), so we need not analyze these issues here; we will instead follow the same course as the district court.

A preliminary injunction is an extraordinary equitable remedy that is available only when the movant shows clear need. Goodman v. Ill. Dep't of Fin. and Prof'l Regulation, 430 F.3d 432, 437 (7th Cir.2005). In our circuit, a district court engages in a two-step analysis to decide whether such relief is warranted. Girl Scouts of Manitou Council, Inc. v. Girl Scouts of USA, Inc., 549 F.3d 1079, 1085–86 (7th Cir.2008). In the first phase, the party seeking a preliminary injunction must make a threshold showing that: (1) absent preliminary injunctive relief, he will suffer irreparable harm in the interim prior to a final resolution; (2) there is no adequate remedy at law; and (3) he has a reasonable likelihood of success on the merits. Id. at 1086 ; see also Cooper v. Salazar, 196 F.3d 809, 813 (7th Cir.1999).

If the movant makes the required threshold showing, then the court proceeds to the second phase, in which it considers: (4) the irreparable harm the moving party will endure if the preliminary injunction is wrongfully denied versus the irreparable harm to the nonmoving party if it is wrongfully granted; and (5) the effects, if any, that the grant or denial of the preliminary injunction would have on nonparties (the “public interest”). Girl Scouts, 549 F.3d at 1086 ; Cooper, 196 F.3d at 813 ; Roland Mach. Co. v. Dresser Indus., Inc., 749 F.2d 380, 386–88 (7th Cir.1984). The court weighs the balance of potential harms on a “sliding scale” against the movant's likelihood of success: the more likely he is to win, the less the balance of harms must weigh in his favor; the less likely he is to win, the more it must weigh in his favor. Girl Scouts, 549 F.3d at 1086 ; Roland, 749 F.2d at 387–88.

On appeal, we review the issuance of a preliminary injunction for abuse of discretion. Cooper, 196 F.3d at 813. “[B]ecause preliminary injunctions are an unusual remedy requiring the application of a definite set of standards, we subject them to ‘effective, and not merely perfunctory, appellate review.’ ” Id. (quoting Roland, 749 F.2d at 389 ). A district court may abuse its discretion by making a clear factual error or a mistake of law. Id. But we give substantial deference to the court's weighing of evidence and balancing of the various equitable factors. Id.; Girl Scouts, 549 F.3d at 1086.

Only two of the preliminary injunction factors are at issue in this appeal: CentiMark's likelihood of prevailing in its effort to enforce the restrictive covenants (factor 3), and the balance of potential harms to the parties (factor 4).

A. CentiMark's Likelihood of Success on the Merits

The district court concluded that CentiMark would likely prevail in its action to enforce Turnell's non-compete and non-solicitation covenants, but only to the extent reasonably necessary for CentiMark's protection. Pennsylvania law disfavors restrictive covenants because it views them as restraints on trade and impediments to an employee's ability to earn a living. Hess v. Gebhard & Co., 570 Pa. 148, 808 A.2d 912, 917 (2002). But Pennsylvania also recognizes restrictive covenants as “important business tools.” See id. at 917–18 ; cf. Outsource Int'l, Inc. v. Barton, 192 F.3d 662, 669–70 (7th Cir.1999) (Posner, J., dissenting) (discussing the economic benefits of covenants not to compete). Pennsylvania courts will therefore enforce a restrictive covenant so long as it is “incident to an employment relationship between the parties; the restrictions imposed by the covenant are reasonably necessary for the protection of the employer; and the restrictions imposed are reasonably limited in duration and geographic extent.”Hess, 808 A.2d at 917.

If the covenant's restrictions are too broad, “a court of equity may grant enforcement limited to those portions of the restrictions that are reasonably necessary for the protection of the employer.” Id. at 920. Partial enforcement in these circumstances is sometimes called “blue penciling” (apparently a throwback to the days when lawyers edited written work with a blue pencil). The Supreme Court of Pennsylvania has “repeatedly held” that courts have discretion to use the blue pencil. Id. “The reason for this policy is a refusal to allow the employee to profit, at the expense of his former employer, from his wrongful and inequitable conduct.” Sidco Paper Co. v. Aaron, 465 Pa. 586, 351 A.2d 250, 255 (1976).

There is a limit, however, to Pennsylvania courts' willingness to reform a restrictive covenant. If the restrictions are so “gratuitous[ly]” overbroad that they “indicate[ ] an intent to oppress the employee and/or to foster a monopoly,” a court of equity may refuse to enforce the covenant at all. Id. at 257. The leading case is Reading Aviation Service, Inc. v. Bertolet, 454 Pa. 488, 311 A.2d 628 (1973). Bertolet left his job at Reading Aviation and went to work for a competitor located 600 feet away from his former employer. He was subject to a non-compete agreement lacking both a temporal limitation and a geographical one—even though Reading Aviation operated entirely within Pennsylvania. Id. at 629. The lower court refused to enforce the non-compete even in part. Id. at 630. The Pennsylvania Supreme Court affirmed because it found the “open-ended restrictions” to be “far greater than ... reasonably necessary” and “unconscionable.” Id.

The court in Reading was concerned about the “possible adverse consequences” of partially enforcing an oppressive covenant: such enforcement tends to encourage employers with superior bargaining power to overreach. Id. at 630–31. That is a valid concern. If such a practice became widespread, it would chill lawful activity. Although judges could pare down the restrictions in the cases that make it to court, in the many cases that never go to court, or where the employee is deterred from even trying to leave, the employer would unfairly benefit from “the in terrorem effects of the oppressive and overly broad covenants.” Tradesman Int'l, Inc. v. Black, 724 F.3d 1004, 1018 (7th Cir.2013) (Hamilton, J., concurring).

Not every overbroad covenant is oppressive, though. In many cases, overbreadth has a more benign cause (such as poor drafting), and to some extent overbreadth is unavoidable given the imprecision of our language. Thus, “absent bad faith, Pennsylvania courts do attempt to blue pencil covenants before refusing enforcement altogether.” Victaulic Co. v. Tieman, 499 F.3d 227, 238 n. 7 (3d Cir.2007) ; see also Sidco, 351 A.2d at 260 (Pomeroy, J., concurring) (characterizing “striking down such covenants in their entirety” as an “extraordinary sanction”).

In the case before us, the district court wielded the blue pencil. In our view, the court could have narrowed Turnell's restrictive covenants even further than it did. The preliminary injunction's prohibition on certain sales of “commercial roofing” is too broad, as CentiMark sells a more specific offering—commercial, flat, single-ply roofing. And the injunction need not cover all of Michigan, Indiana, and Illinois because only portions of those states were within the region that Turnell managed at CentiMark.

But Turnell does not challenge the scope of the preliminary injunction on appeal. Instead, he argues that the district court should not have enforced the restrictive covenants at all—even in blue-penciled form—because, like the covenants in Reading, they are overbroad and oppressive.We disagree. The covenants are incident to an employment relationship, and two years post-employment is a reasonable duration. See Centimark Corp. v. Vitek, No. 08 C 7323, 2010 WL 5490662, at *7 (N.D.Ill. Dec. 29, 2010). The covenants further CentiMark's legitimate interest in preventing a former regional manager from using its customer relationships and proprietary business information, including its pricing models, for a competitor's benefit. See Sidco, 351 A.2d at 254, 257 (holding that “Sidco clearly has a protectible interest in customer goodwill” and “properly used a restrictive covenant to protect its customer relationships”); see also Bimbo Bakeries USA, Inc. v. Botticella, 613 F.3d 102, 112–14 (3d Cir.2010) (noting that Pennsylvania law protects non-technical trade secrets); CertainTeed Corp. v. Williams, 481 F.3d 528, 530 (7th Cir.2007) (recognizing that under Pennsylvania law employers have a legitimate interest in “[k]eeping a business executive with a wealth of information from taking an equivalent position at a rival”). And while the district court found the scope of prohibited activity and the geographic reach of the restrictions broader than necessary, the covenants are not gratuitously or oppressively overbroad.

1. The Non–Compete Covenant

Turnell's non-compete clause prohibits him from engaging in competing business, defined as selling products or services the same as or “similar” to CentiMark's. This language might benefit from more precision (what counts as similar?) or a narrower focus (similar products do not necessarily compete, see, e.g., Am. Med. Sys., Inc. v. Med. Eng'g Corp., 794 F.Supp. 1370, 1388–90 (E.D.Wisc.1992) (finding that two different medical devices for treating impotence were not serious competitors), rev'd in part on other grounds, 6 F.3d 1523 (Fed.Cir.1993) ). But that is beside the point. The provision reflects a good-faith attempt to tie the scope of the prohibitions to CentiMark's protectable interests.

Pennsylvania courts and others applying Pennsylvania law have enforced comparable restrictions, sometimes in their entirety, e.g., CertainTeed, 481 F.3d at 529 (involving restriction on developing products “competitive with or similar to” the employer's products); Vitek, 2010 WL 5490662, at *2 (enforcing covenants almost identical to Turnell's), and sometimes with recourse to the blue pencil, e.g., CertainTeed Ceilings Corp. v. Aiken, No. 14–cv–3925, 2014 WL 5461546, at *3, 2014 U.S. Dist. LEXIS 152446, at *8 (E.D.Pa. Oct. 27, 2014) (involving restriction on products “competitive with or similar to” the employer's).

Nor does the non-compete cross the line with respect to its geographic reach. It applies only in the regions where Turnell “operated” as a CentiMark employee. This language is vague, at least at the margins: if Turnell placed a telephone call from Chicago to a customer in California, for example, does that mean he “operated” in California? What if he traveled there a few times on business? The answer is unclear. Turnell claims he “worked in” twenty-four different states during his thirty-five years at the company, but CentiMark agrees only that he operated in the seven states covered by the injunction. Their dispute illustrates the vagueness of the term “operate.” Perhaps the parties should have defined it more precisely. But they could not have foreseen in 1988 what positions Turnell would later hold or where he would work. To some extent, the geographic language had to be open-ended given the national scope of CentiMark's business.

None of this suggests deliberate or oppressive overreaching. Where the employer's business was sufficiently widespread, courts have enforced restrictive covenants with even greater geographic reach than Turnell's. Sometimes they have done so without even modifying the covenants, see, e.g., Mohr v. Bank of N.Y. Mellon Corp., 393 Fed.Appx. 639, 644–45 (11th Cir.2010) (per curiam) (applying Georgia law); Quaker Chem. Corp. v. Varga, 509 F.Supp.2d 469, 476–77 (E.D.Pa.2007) ; other times they have blue penciled first, see, e.g., CertainTeed Ceilings, 2014 WL 5461546, at *11–13, 2014 U.S. Dist. LEXIS 152446, at *31–36 (restricting enforcement to the employee's former sales territory).

Turnell invokes Reading, but the employer in that case was a one-state company trying to enforce a restrictive covenant with no geographic limit. 311 A.2d at 629. There is no similar mismatch here. The other case on which Turnell relies, Fres–Co System USA, Inc. v. Bodell, is distinguishable on similar grounds. No. Civ. A. 05–3349, 2005 WL 3071755 (E.D.Pa. Nov. 15, 2005) (applying Pennsylvania law). The employee in Fres–Co sold coffee packaging in the southeastern United States and the Caribbean, but his non-compete forbade him from selling packaging material in at least four industries having little to do with coffee, and on at least three continents. Id. at *4, 7. There were also other indicia of oppressive intent in Fres–Co , including a lack of consideration for the restrictive covenant, and a requirement that all employees, no matter how junior, agree to the same restrictions. Id. at *5. Turnell's non-compete, by contrast, reflects an attempt to tailor his restrictions to the products and services he sold (commercial roofing) and the areas where he sold them, leaving him free to work in other product and geographical markets.

2. The Non–Solicitation Covenant

Turnell's non-solicitation provision presents a closer call. It prohibits marketing not only to CentiMark's actual customers but also to any “prospective customers” it has “contacted” during the restrictive periods. This language is too vague (what kinds of contact count?), too broad (what relationship can CentiMark legitimately seek to safeguard with a prospect it merely called a few years ago?), and impracticable (Turnell cannot know everyone his former employer contacted)—which is why the district court excised it.

The non-solicitation provision also contains no geographic limitation. It applies wherever CentiMark's customers or prospective customers happen to be located. That reaches too widely. Although Turnell was a management-level employee with access to customer data, proposals, and customers themselves, he did not have relationships with all customers across the country. While the parties could not easily have predicted in 1988 where Turnell would operate, they still could have written a narrower restriction—for example, by limiting it to customers contacted by Turnell himself. Cf. Mohr, 393 Fed.Appx. at 642.

But there is no reason to suspect any more sinister cause of these shortcomings than poor drafting. It is reasonable for an employer to seek protection for budding customer relationships as well as established ones. And it is difficult to define in advance which relationships will warrant protection. (You cannot simply list the customers, for example, because the list will change over time.) CentiMark invested time, expertise, and money in developing its customers and potential customers. Turnell was directly involved in those efforts; for instance, he reviewed significant proposals before they went out the door. That is reason enough for the restrictions on solicitation here. There is no evidence that CentiMark was actually trying to oppress Turnell or to somehow foster a monopoly. Cf. Outsource Int'l, 192 F.3d at 670 (Posner, J., dissenting) (noting that a non-compete in in an employment contract is unlikely to impair the vitality of competition).

Other courts have enforced similar agreements barring solicitation of prospective customers. See, e.g., Mohr, 393 Fed.Appx. at 642, 645–46 (reversing district court's denial of preliminary injunction and enforcing covenant under Georgia law); Diodato v. Wells Fargo Ins. Servs., USA, Inc., 44 F.Supp.3d 541, 549, 570 (M.D.Pa.2014) (blue penciling under Pennsylvania law); Pulse Techs., Inc. v. Dodrill, No. CV–07–65–ST, 2007 WL 789434, at *2–3, 6–8, 2007 U.S. Dist. LEXIS 18520, at *6–7, 20–23 (D.Or. Mar. 14, 2007) (same).

We conclude that Turnell's restrictive covenants are overbroad but not oppressively so. The district court therefore properly exercised its equitable discretion under Pennsylvania law to blue pencil the restrictions. And the court properly concluded that CentiMark has a strong chance of enforcing them, as narrowed, in a final resolution of this action.

B. The Balance of Harms

The district court found that “an injunction would weigh more heavily on Turnell than the status quo would [on] Centimark.” Nevertheless, using the sliding scale approach, see Girl Scouts, 549 F.3d at 1086, the court concluded that CentiMark's strong likelihood of success outweighed the harm to Turnell.

We agree. Both parties claim that they will suffer monetary harm: lost income in Turnell's case, and lost customer relationships and proprietary information in CentiMark's. Neither party has quantified those harms. But in neither case do they seem likely to be crippling. Even under the preliminary injunction, Turnell is still able to work: he can remain at Windward and sell commercial roofing, just not to CentiMark customers in seven states; or he can sell other types of roofing without restriction. As a successful salesperson, he may also be able to work in another field. In short, the injunction does not prevent him from earning a living. On the other hand, CentiMark has no proof that it has lost customers or suffered any economic loss because of Turnell's breaches. The harms are pretty evenly balanced, perhaps tipping in Turnell's favor, as the district court thought.

But that is not the end of the analysis. The kind of harm we are concerned about when deciding whether to issue a preliminary injunction is not harm tout court but rather irreparable harm. See Roland, 749 F.2d at 387. Turnell's harm seems largely reparable. If he prevails in a trial on the merits, it should be possible to quantify his losses and compensate him fully with damages. Moreover, the bond posted by CentiMark ensures that any damages would be recoverable (at least up to $250,000). See Fed.R.Civ.P. 65(c). Turnell has not explained why these remedies would be insufficient. There is no reason to think the preliminary injunction will consign him to poverty or bankrupt him: he earned a high salary at CentiMark, has at least $3 million in assets, and has income from his wife's company, in addition to what he can earn on his own.Cf. Roland, 749 F.2d at 386 (discussing various scenarios when damages after-the-fact would be inadequate).

The potential damage to CentiMark, on the other hand, is a canonical form of irreparable harm. The injuries that flow from the violation of a non-compete are difficult to prove and quantify. See CertainTeed, 481 F.3d at 529 (“[I]t may be very difficult to show that the ex-employee has used confidential information.”); John G. Bryant Co. v. Sling Testing and Repair, Inc., 471 Pa. 1, 369 A.2d 1164, 1167 (1977) (characterizing such damage as “incalculable”). That is what makes restrictive covenants prime candidates for injunctive relief. Turnell's breaches, if left unchecked, might cause little harm to CentiMark, or they might cause great harm. But either way, the magnitude and even the existence of the injury will be difficult to discern. Injunctive relief is therefore the best, and probably the only, adequate remedy.

On the whole, then, we think the balance of harms favors Turnell, if at all, only slightly. It is not enough to overcome CentiMark's likelihood of success on the merits.

III. Conclusion

The district court's order granting a preliminary injunction in CentiMark's favor is AFFIRMED.