NEWSOM, Circuit Judge

Christopher Todd Giovinazzo, Patrick Christopher Fagan, Tiana S. Mykkeltvedt, Bondurant Mixson & Elmore, LLP, ATLANTA, GA, for Plaintiff - Appellee.

Thomas E. Reilly, William Alexander Smith, Troutman Pepper Hamilton Sanders, LLP, ATLANTA, GA, for Defendant - Appellant.

Before MARTIN, NEWSOM, and O'SCANNLAIN, Circuit Judges.

NEWSOM, Circuit Judge: This is the story of a deal gone south. PruittHealth, Inc., a healthcare provider, was looking for a new way to organize its patient data; Caradigm USA LLC is in the business of delivering software solutions to healthcare providers. The companies executed a contract pursuant to which Pruitt would pay Caradigm to pull patient information from Pruitt's various systems and organize it so that each patient's medical and billing records would be available in one place.

But alas. A little more than six months in, Pruitt grew dissatisfied with Caradigm's progress and announced that it was walking away from the project. Caradigm informed Pruitt that it stood ready to perform and warned Pruitt that it couldn't just unilaterally ditch the contract. Pruitt didn't respond, and Caradigm eventually sued. The district court decided on summary judgment that Pruitt had anticipatorily breached the contract and that Caradigm was therefore entitled to the full value of the deal. Because that value was uncertain, the district court left the issue of damages for trial. After a four-day trial, a jury awarded Caradigm $11 million, which included contract damages, compound interest on those damages, and attorney's fees and expenses.

On appeal, Pruitt doesn't contest the district court's determination that it breached, but it does contend that the district court erred in numerous ways leading up to the jury's verdict, which, it says, resulted in an overstated damages award and erroneous awards of interest and fees. We conclude that the district court did not reversibly err in most of the ways that Pruitt claims, and we will therefore affirm the awards of contract damages and fees, as well as the determination that Caradigm is entitled to recover interest on the damages award. Because we hold that it was error to compound the interest, however, we will vacate that award and remand for the district court to calculate interest in simple terms.

I

A

PruittHealth, Inc. is a private, for-profit healthcare provider dealing mostly in home healthcare, skilled-nursing facilities, and hospices. Pruitt was looking for a new way to store its patient records, which were located across multiple systems and difficult to access for billing and treatment purposes. Caradigm USA LLC is a software company that services healthcare providers. As relevant here, Pruitt was particularly interested in one service that Caradigm offered—the "Amalga Platform"—which could aggregate patients’ clinical and financial information. Pruitt and Caradigm executed an agreement pursuant to which Caradigm would pull patient information from Pruitt's separate data systems and aggregate it so that each patient's medical and billing records would be accessible in a single platform. The contract provided the legal framework for the agreement, defined important terms, included an order form showing the goods and services that Pruitt purchased, and explained the details of how the parties would implement Caradigm's software. The contract also contemplated a crucial turning point—"First Productive Use"—when Pruitt would initially employ Caradigm's software "to process actual patient data in a live production environment."

After First Productive Use—and this is important to the damages dispute before us—the contract provided for a significant increase in monthly payments. The order form specified that Pruitt's subscription to the Amalga Platform would cost $64,649 per month, based on "[m]edium" usage, with payments "starting on First Productive Use." Pruitt's medium subscription size was calculated based on three metrics: the number of data feeds (i.e. , inbound electronic data from Pruitt's information systems), the number of messages (i.e. , transmission of certain files) per day, and the number of permitted users. So, the more feeds, messages, and users, the bigger the platform size that Pruitt would need (and vice versa). The agreement stated that each quarter, "Caradigm w[ould] review [Pruitt's] actual usage" against the capacity level for its then-current platform size, and "[i]f each metric f[ell] below its threshold for a lower [platform] size, then [Pruitt] w[ould] be moved down" a size and charged fees for that smaller size the next quarter. The agreement also provided for the opposite: If any metric was exceeded in a quarter, Pruitt would be moved up a platform size. By its terms, the agreement was set to "run for an Initial Term ending five years after the date of First Productive Use."

But the project never reached First Productive Use. An early step in the implementation of Caradigm's software called for Pruitt to send Caradigm a data sample so that Caradigm could test its ability to match patient files coming from disparate sources (i.e. , the "patient matching" process). Pruitt was apparently disappointed with the results of an initial patient-matching test in December 2014. Caradigm's internal communications show that, as of January 23, 2015, it planned to offer Pruitt two options with regard to patient matching: Pruitt could either (1) continue with Caradigm's "basic patient matching as described in the proposal," which "include[d] basic support (in house, no third party)," with any "[e]nhancements" and additional services to be billed separately, or (2) purchase a third-party matching system, which had more "bells and whistles."

Caradigm was never able to present those options to Pruitt. On January 29, Pruitt's Chief Information Officer, Dan Martin, telephoned Caradigm account executive Tina Mirkheshti to notify her that Pruitt was likely to abandon the project. Less than a month later—on February 17—Pruitt sent Caradigm a confirmatory email stating, in relevant part:

[W]e have suspended all efforts dedicated to this project and that includes the utilization of your services. Therefore the contract you have with PruittHealth is no longer valid. In addition regarding payments, we will make our final payment for January date(s) of service, and this will fulfill our economic commitment under the agreement.

About a month after Pruitt's email—on March 16—Caradigm sent Pruitt a letter stating, among other things, that Pruitt didn't "have a unilateral right to terminate the Agreement during the Agreement's five-year initial term" and that Pruitt's "purported termination would therefore constitute a breach ... result[ing] in the acceleration of payments owed to Caradigm for the contracted five-year period." Caradigm's letter also stated that it stood "ready, willing, and able to continue to fulfill [its] obligations under the Agreement."

Caradigm says—and Pruitt doesn't dispute—that Pruitt never responded to the March 16 letter. Caradigm sent Pruitt seven invoices (for a total of $287,000) that Pruitt neither disputed nor paid. Pruitt paid Caradigm for fees that accrued through March 2015, but no more.

B

In July 2015, Caradigm sued Pruitt for breach of contract in the U.S. District Court for the Northern District of Georgia. Caradigm alleged that Pruitt had anticipatorily breached, entitling Caradigm to the contract's full value, plus interest and attorney's fees. After discovery, the parties filed dueling summary judgment motions. The district court decided that Pruitt had anticipatorily repudiated the contract before Caradigm's performance was required and that Pruitt was therefore liable for breach. Because the value of the contract wasn't clear, though, the district court left the issue of damages for trial.

Before trial, Pruitt sought clarification regarding Caradigm's burden to prove damages—specifically, whether Caradigm had to "prove to a reasonable certainty that it would have satisfactorily performed its obligations under the Agreement" and that "First Productive Use would have been achieved." Defendant's Motion for Clarification at 1–2, 4. The district court held that, under Georgia law, the entire value of the agreement became due at the moment Pruitt breached, and that the value hinged on the point at which the parties moved from the pre-First Productive Use phase to the "going live" phase. Dist. Ct. Order on Motion for Clarification at 2–3. Thus, the court concluded that Caradigm would have "to show a jury, with reasonable certainty, the time frame in which the pre-First Productive Use phase would likely have concluded." Id. at 2. In so holding, the district court made two things clear concerning Caradigm's burden: (1) Caradigm didn't have to show Pruitt's subjective satisfaction in order to prove that the project would have moved forward to First Productive Use because "Pruitt's performance under the contract was not conditioned on its satisfaction" and "the contract lacked any ‘subjective satisfaction’ requirement"; and (2) by breaching, Pruitt "absolved Caradigm of remedying any issues that Pruitt had with the performance of the software platform." Id. at 2–3, 4.

Over the course of a four-day trial, Caradigm called as witnesses (1) Tina Mirkheshti, who had managed the Pruitt contract; (2) a damages expert; and (3) its counsel (to testify regarding attorney's fees). Mirkheshti testified that Caradigm would have performed the agreement successfully and on time if Pruitt hadn't breached and that Pruitt ignored her calls and emails after Martin had told her that Pruitt was likely backing out. Caradigm's expert explained his damages calculations and said that they were supported by Caradigm's confidence that it would have performed the contract on time. When Caradigm rested, Pruitt made a motion for judgment as a matter of law, which the district court denied. Pruitt didn't call any witnesses of its own.

The jury awarded Caradigm $11,110,580.11, which comprised $5,105,582 in contract damages, $3,686,016 in compound interest, and $2,318,982.11 in attorney's fees and expenses under Georgia Code § 13-6-11, which provides for fee awards against "stubbornly litigious" parties. Pruitt filed a renewed motion for judgment as a matter of law and, in the alternative, a motion for a new trial or to alter or amend the judgment—both of which the district court denied.

This is Pruitt's appeal.

II

Pruitt argues that the district court erred in several ways in the run-up to and during the damages trial, which it alleges led to an overstated contract-damages award and to erroneous awards of interest and attorney's fees. Pruitt's challenges require us to examine a whole host of issues, including the proper interpretation of the parties’ contract, exclusions of evidence, and jury instructions. As we will explain, after carefully considering each of Pruitt's contentions, we conclude that, in the main, the district court did not reversibly err, and we will therefore affirm the awards of contract damages and fees, as well as the determination that Caradigm is entitled to recover interest on the damages award. Because we further conclude, however, that it was error to compound the interest, we will vacate that award and remand so that the district court can calculate simple interest.

A

Pruitt first contends that the jury's award of more than $5 million in contract damages is exaggerated for three reasons: The district court, Pruitt says, (1) erred in its construction of the parties’ obligations under the contract, (2) held Caradigm to a lower burden of proof than it should have, and (3) erred in excluding evidence suggesting that Caradigm's future revenues might have decreased. We take each argument in turn.

1

Pruitt asserts that the parties’ contract did two particular things: (a) it guaranteed Pruitt's right to test Caradigm's software to its satisfaction as a precondition to the project reaching First Productive Use; and (b) it obligated Caradigm to provide a certain level of accuracy in patient matching. Had the district court interpreted the contract to account for these requirements, Pruitt says, the jury could have found that the project wouldn't have reached First Productive Use and, accordingly, that Pruitt wouldn't have owed Caradigm damages for fees that would have been incurred only after that milestone. As we read the contract, it didn't do either of the two things Pruitt thinks it did. Before we address the particular contractual provisions at issue and explain why we reject Pruitt's reading, we start with a primer on Georgia contract law, which governs the parties’ agreement here. The Georgia Code states that "[t]he cardinal rule of construction is to ascertain the intention of the parties" and that, "[i]f that intention is clear and it contravenes no rule of law and sufficient words are used to arrive at the intention, it shall be enforced irrespective of all technical or arbitrary rules of construction." Ga. Code Ann. § 13-2-3. Accordingly, "[w]henever the language of a contract is plain, unambiguous, and capable of only one reasonable interpretation, no construction is required or even permissible, and the contractual language used by the parties must be afforded its literal meaning." First Data POS, Inc. v. Willis , 273 Ga. 792, 546 S.E.2d 781, 784 (2001) ; see also Atlanta Multispecialty Surgical Assocs., LLC v. DeKalb Med. Ctr., Inc. , 273 Ga.App. 355, 615 S.E.2d 166, 167 (2005) ("Where the terms of a contract are unambiguous, the court must enforce it literally, even if one party contends that he understood the contract to mean something else.").

When interpreting a particular contractual provision, we must examine it in the context of other provisions and the contract as a whole. See Ga. Code Ann. § 13-2-2(4) ("[T]he whole contract should be looked to in arriving at the construction of any part."); Archer W. Contractors, Ltd. v. Estate of Pitts , 292 Ga. 219, 735 S.E.2d 772, 777 (2012) ("[T]he context in which a contractual term appears always must be considered in determining the meaning of the term."). The meaning and reach of one contract provision can thus be limited by others. See, e.g. , Sheridan v. Crown Capital Corp. , 251 Ga.App. 314, 554 S.E.2d 296, 299 (2001) (explaining that one section of a contract "should be construed in pari materia" with another to define and limit the meaning of the first section). With these principles in mind, we turn to the contractual provisions at issue.

a

We first address the contract's "testing" clause, which Pruitt says gave it a right to demand subjective satisfaction with its testing of Caradigm's software before the project could proceed to First Productive Use. In pertinent part, that clause provides:

Following Installation, [Pruitt] may test the Cloud Services to determine whether they perform in material conformance with the Documentation. ... With regard to Customer Use Cases, [Pruitt] will notify Caradigm when testing has been satisfactorily completed and [Pruitt] is ready for Caradigm to move the Customer Use Case to the production environment.

Presumably relying on the language that it will notify Caradigm "when testing has been satisfactorily completed ," Pruitt construes the testing clause as a "[s]atisfaction [c]lause." See Br. of Appellant at 30. We disagree. In general, "a provision in a contract requiring the performance of one party to be satisfactory to the other is valid and enforceable." Commercial Mortg. & Fin. Corp. v. Greenwich Sav. Bank , 112 Ga.App. 388, 145 S.E.2d 249, 251 (1965). Pruitt cites Atlanta Realty Co. v. Campion for the proposition that "[w]here one contracting party agrees to perform services ‘to the satisfaction of’ or ‘satisfactory to’ the other party, compliance with the contract is not shown unless it appears that the thing done or the article furnished does in fact satisfy the other party." 94 Ga.App. 136, 93 S.E.2d 781, 782 (1956).

The testing clause's language, however, convinces us that it is not a satisfaction clause, as Pruitt contends. It doesn't read like a promise on Caradigm's part to "perform services ‘to the satisfaction of’ or ‘satisfactory to’ " Pruitt, id. , and it looks nothing like the satisfaction clauses that Georgia courts have enforced. The testing clause, for instance, doesn't set up an "if ... then" proposition—say, that "if testing is not satisfactory to Pruitt, then Pruitt shall have the right to cancel." Cf. Yon v. City of Atlanta , 201 Ga. 800, 41 S.E.2d 516, 519 (1947) (enforcing a provision in a lease stating that "if said premises and the merchandise sold therein are not up to a reasonable standard ... [a designated] committee shall have the right and option, on behalf of the lessor, to declare this contract void"). Nor does it state that the project would reach First Productive Use "if testing proved satisfactory" or "if Caradigm performed to Pruitt's satisfaction." Cf. Mackenzie v. Minis , 132 Ga. 323, 63 S.E. 900, 902–03 (1909) (enforcing a satisfaction clause in an employment contract stating that the employment period would be at least three years if the employee "prove[d] himself competent and satisfactory" and performed according "to [his employer's] satisfaction"). To the contrary, the testing clause states that Pruitt "will notify Caradigm when testing has been satisfactorily completed" and Pruitt "is ready" for next steps. If anything, the clause reads like a promise on Pruitt's part—i.e. , an "expression of an intention that some future performance will be rendered." First Nat'l Bank of DeKalb Cty. v. Nat'l Bank of Ga. , 249 Ga. 216, 290 S.E.2d 55, 57 (1982).

Even if the testing clause's language was ambiguous, reading the clause in the larger context of the contract confirms that it didn't give Pruitt a right to demand subjective satisfaction. Take the "Term & Termination" clause, which stated, as relevant here:

This Agreement will run for an Initial Term ending five years after the date of First Productive Use of the Amalga Platform, which will automatically renew for successive 12-month Renewal Terms unless either party provides written notice of non-renewal at least 90 days before the end of the then-current Term. ... Either party may terminate this Agreement or any Order if the other party commits a material breach of this Agreement or the applicable Order that remains uncured for 90 days after written notice of such breach.

This clause provides two important insights into the parties’ intent. First, the contract was anticipated to run through First Productive Use. The language stating that the contract "will run" for a specified term "after ... First Productive Use" evinces the parties’ shared expectation that the project would in fact reach First Productive Use and makes clear that reaching First Productive Use was not conditioned on satisfactory testing. Second, a party could terminate the contract only under particular circumstances and by following particular procedures—one party must have "commit[ted] a material breach," then there must have been "written notice" of that breach, and finally the breach must have "remain[ed] uncured for 90 days." There is no suggestion that Pruitt's dissatisfaction provided a separate basis for termination. Even if there were, there is no evidence—so far as we can tell—that Pruitt provided Caradigm "written notice" of its dissatisfaction.

Pruitt's reading of the testing clause also wouldn't make sense as a practical matter because its satisfaction depended, at least in part, on its own work. This is so for two reasons. First, Pruitt was ultimately responsible for the quality of the data fed into Caradigm's software, which, in turn, would impact the accuracy of patient matching. Another clause of the contract, which we'll explain in detail later, made clear that Pruitt was "solely responsible ... for verifying the accuracy and adequacy of information and data furnished for processing." As Caradigm's account executive explained at trial: "[A]n algorithm is based on mathematics and the data that it receives. Some customers have data that we call dirtier than other[s]. Therefore ... you get sort of garbage in, garbage out ...." Trial Transcript Vol. II at 18. Second, the testing of "Use Cases" involved testing not only cases developed by Caradigm, but also those developed by Pruitt. Because Pruitt purchased a development license from Caradigm, the contract defined "Use Case[s]" to include "use cases developed by [Pruitt] under the [development license]," and defined "Customer Use Case[s]" to include "any Use Case developed by or for [Pruitt] or customized for [Pruitt]."

There is one final point: It seems to us clear that testing was permissive, rather than mandatory, under the testing clause. That provision begins by stating that Pruitt "may test" Caradigm's software, rather than that Pruitt "shall test" or "must test" the software. To determine whether the word "may" denotes permission or a command in a contract, we must consider it "in connection with its context and subject matter." Holden v. Smith , 236 Ga.App. 205, 511 S.E.2d 569, 571 (1999) (quotation omitted). As we have already explained, the testing clause is best understood as a promise on Pruitt's part to notify Caradigm when it had completed testing. Because Pruitt was responsible for the quality of the data and for creating at least some of its own use cases, it makes sense that it had the onus not only to notify Caradigm when testing was completed but also to decide whether to test in the first place.

Accordingly, the testing clause did not give Pruitt a right to demand subjective satisfaction with testing before the project proceeded to First Productive Use. The plain language of the clause and other provisions in the contract—not to mention practical consequences—combine to foreclose Pruitt's interpretation. b

Pruitt separately asserts that the contract obligated Caradigm to deliver a certain level of accuracy in patient matching, and it contends that the district court erred in reading what Pruitt calls "two disclaimers" to provide otherwise. See Br. of Appellant at 33. Specifically, Pruitt contends that the district court erred in relying on these provisions to hold that Caradigm wasn't obligated to deliver " ‘robust’ patient matching." See id.

We hold that the district court correctly read these provisions as disclaiming any responsibility on Caradigm's part to provide a certain level of accuracy in patient matching. Let's start with their text. The contract's "Disclaimer of Warranties" provision states, in relevant part:

OTHER THAN THE LIMITED WARRANTIES SET FORTH IN SECTION 17, CARADIGM PROVIDES NO OTHER EXPRESS OR IMPLIED WARRANTIES OR CONDITIONS WITH RESPECT TO SOFTWARE OR SERVICES. CARADIGM DISCLAIMS ANY IMPLIED REPRESENTATIONS OR WARRANTIES, INCLUDING ... CORRECTNESS OR RELIABILITY OF PATIENT INFORMATION (INCLUDING PATIENT RECORD MATCHING) ....

The other provision, titled "Medical Diagnosis, Treatment, & Billing," states that:

[Pruitt] is solely responsible ... for verifying the accuracy and adequacy of information and data furnished for processing .... Because patient information is maintained in multiple places ... not all patient information is kept in a standard fashion or is regularly updated and, because of the statistical nature of the algorithm used to match patient identities with their clinical information, it is possible that false matches may occur or that there may be errors or omissions in the clinical information.

Because the "plain" and "unambiguous" language of these provisions is "capable of only one reasonable interpretation," it "must be afforded its literal meaning." First Data , 546 S.E.2d at 784. The "Disclaimer of Warranties" provision was written in all caps, clearly labeled as a "disclaimer," and specifically identified "patient record matching" as one of its objects. It clearly demonstrates that Caradigm did not promise to deliver—and in fact, expressly disclaimed—a certain level of "correctness or reliability" in "patient record matching." The "Medical Diagnoses, Treatment, & Billing" provision is likewise unambiguous. It warns that "false matches may occur" and that "there may be errors or omissions in the clinical information." This evinces the parties’ understanding that errors could occur during the patient-matching process, not that Caradigm would deliver a certain level of accuracy. It makes sense that the contract included these disclaimer provisions because, as we have already explained, the accuracy of patient matching depended at least in part on the quality of the data that Pruitt submitted. See supra at 1269–70. Indeed, the "Medical Diagnosis, Treatment, & Billing" provision confirms as much, in making clear that Pruitt "is solely responsible ... for verifying the accuracy and adequacy" of the data it provided.

Pruitt attempts to evade these provisions’ plain meaning by asking us to consider the contract as a whole and the project's "overall vision." See Br. of Appellant at 33, 35. But Pruitt points to no other provision in the contract that obligates Caradigm to provide a particular level of accuracy in patient matching. We reject Pruitt's contention that the contract's warranty provision, which vaguely states that Caradigm "warrants that ... all Services will be performed in a professional and workmanlike manner," required Caradigm to deliver some particular (but undefined) level of accuracy in patient matching. Pruitt relies on a Georgia case holding that "it [wa]s unreasonable to allow an express warranty contained in a contract ... to be negated by a disclaimer of warranty in the same contract" because "the two provisions [we]re not consistent with each other." See Century Dodge, Inc. v. Mobley , 155 Ga.App. 712, 272 S.E.2d 502, 504 (1980). But the specific disclaimer and general warranty provisions here aren't inconsistent. Caradigm could perform "in a professional and workmanlike manner" and yet still disclaim a particular level of accuracy in patient matching. Moreover, for reasons already explained, we likewise reject Pruitt's argument that the testing clause impacts our interpretation of these provisions. See supra at 1268–71. Accordingly, we hold that the disclaimer provisions confirm that the contract did not obligate Caradigm to deliver a certain level of patient matching.

2

Having dealt with Pruitt's contract-interpretation arguments, we turn to its contention that the district court improperly reduced Caradigm's burden of proof at trial. Leading up to trial, the district court held that, because Pruitt had repudiated the contract, Caradigm didn't have to prove "that it actually completed its performance under the contract" but rather had to show only "that at the time of repudiation, it had stood ready and able to perform its obligations." Dist. Ct. Order on Motions in Limine at 10–12. Pruitt contends that the district court wrongly "relieved Caradigm from proving that contractual conditions precedent to certain payments would have occurred but for the breach." Br. of Appellant at 19. More specifically, it contends that "Caradigm cannot recover damages for [payments conditioned upon First Productive Use] unless and until Caradigm proves that First Productive Use would have occurred but for the breach." Id. at 26. And, according to Pruitt, First Productive Use could have "occur[red] only when Pruitt had satisfactorily completed testing of a Customer Use Case." Id. at 25.

Pruitt's burden argument—at its core—amounts to little more than a disagreement with the way the district court interpreted the parties’ contract. As we have explained, the testing clause can't be read as a "satisfaction clause," and satisfactory testing was not a condition precedent to First Productive Use or to further performance under the contract. See supra at 1268–71 & n.6 & n.6. In fact, the contract explicitly disclaimed Caradigm's obligation to achieve a certain level of accuracy in patient matching. See supra at 1270–72. The district court didn't improperly "relieve[ ] Caradigm from proving that contractual conditions precedent" would have occurred—because no such conditions existed. Once Pruitt anticipatorily repudiated the contract, Caradigm could "sue and recover [its] entire damages." Jinright v. Russell , 123 Ga.App. 706, 182 S.E.2d 328, 330 (1971) ; see also John Kimpflen, 7 Georgia Jurisprudence, Business and Commercial Law: Contracts § 5:98 (2020) ("Where there has been an anticipatory breach of a contract, the injured party is not limited to a recovery of the mere value of the services which such party has performed; the injured party may recover the entire present value to the injured party of the contract."). In general, "[t]he object of giving damages to the plaintiff is to put him in as good a position as if the defendant had fully performed the contract." PMS Constr. Co. v. DeKalb Cty. , 243 Ga. 870, 257 S.E.2d 285, 288 (1979). This rule applies equally to cases of anticipatory breach. See Redman Dev. Corp. v. Piedmont Heating & Air Conditioning, Inc. , 128 Ga.App. 447, 197 S.E.2d 167, 170–71 (1973) (stating that in an "anticipatory breach" case, "the doctrine of contractual compensatory damages seeks as its goal to make the injured party whole," which means that "the person injured, is, so far as it is possible to do so by a monetary award, to be placed in the position he would have been in had the contract been performed" (quotation omitted)). Accordingly, Caradigm was entitled to damages that would put it "in as good a position as if [Pruitt] had fully performed the contract," PMS Constr. Co. , 257 S.E.2d at 288, meaning that Caradigm was entitled to fees that it would have been paid after First Productive Use.

Even if completed testing or First Productive Use could be interpreted as a condition precedent to further performance—as the district court (we think erroneously) seemed to assume—Pruitt wouldn't fare any better. A party who anticipatorily breaches cannot use the nonoccurrence of a future contractual event to its benefit. See Jinright , 182 S.E.2d at 330 ("A party may not repudiate a contract and at the same time seek the advantage of a stipulation in the very contract he has repudiated."). While Pruitt argues that this principle applies only to the determination of liability, we read Georgia law as embracing a general rule that applies in anticipatory repudiation cases—whether in determining liability or awarding damages. See, e.g. , John K. Larkins, Georgia Contracts: Law and Litigation § 11:5 (2d ed. 2019) ("Where the contract is contingent on the occurrence of a future event, the repudiating party cannot claim the benefit of the contingency so as to avoid any claim of damages ." (emphasis added) (citing Textile Rubber & Chemical Co. v. Thermo-Flex Techs., Inc. , 301 Ga.App. 491, 687 S.E.2d 919 (2009) )). Accordingly, Pruitt couldn't claim that the project would never have reached First Productive Use as a shield to avoid damages because it repudiated the contract before Caradigm had a chance to perform. Jinright , 182 S.E.2d at 330.

3

On, then, to Pruitt's final challenge to the contract-damages award. Pruitt argues that the district court erred in excluding evidence regarding the variable nature of Caradigm's expected revenues under the agreement—specifically, evidence that Pruitt's use of the software might have decreased, and that Pruitt might have been downsized from the more expensive "medium" platform to the less expensive "small" platform. The district court excluded Pruitt's proposed evidence on the basis of its pretrial order on the parties’ motions in limine, which had held that because Pruitt anticipatorily repudiated, it couldn't "invoke conditions precedent in the contract to avoid damages." Dist. Ct. Order on Motions in Limine at 25.

We disagree that the agreement regarding variable pricing constituted a "condition precedent" that fell under the district court's prior order. The contract did provide that "[i]f" certain metrics of Pruitt's use (i.e. , the number of data feeds, the number of users, and the number of messages) changed, that Pruitt would "be moved up" or "be moved down" a platform size. Cf. Larkins, supra , § 5:4 (stating that conditions precedent are often indicated by words of condition, such as "if"). Even so, we don't think that Pruitt's metrics changing could be construed as a condition precedent—that is, a condition that "must be performed before the contract becomes absolute and obligatory upon the other party." Ga. Code Ann. § 13-3-4 (defining "condition precedent"). Rather than creating a condition that had to occur for further performance to take place, the variable-pricing agreement is better understood as an accommodation between the parties about the details of payment. Cf. Callaway v. Garner , 327 Ga.App. 67, 755 S.E.2d 526, 532 (2014) ("Georgia courts have long recognized a material distinction between a condition precedent and a mere accommodation between the parties over the timing of payment."), vacated in part on other grounds , 333 Ga.App. 747, 776 S.E.2d 829 (2015). Therefore, by repudiating the contract, Pruitt didn't forgo its declining-fees argument, and the district court shouldn't have excluded the evidence on the basis it did.

But of course "[n]ot every evidentiary error ... requires reversal." Peat, Inc. v. Vanguard Research, Inc. , 378 F.3d 1154, 1162 (11th Cir. 2004). We have held that "a new trial is warranted only" where an error in admitting or excluding evidence "has caused substantial prejudice to the affected party (or, stated somewhat differently, affected the party's ‘substantial rights’ or resulted in ‘substantial injustice’)." Id. ; see also Fed. R. Civ. P. 61 ("Unless justice requires otherwise, no error in admitting or excluding evidence—or any other error by the court or a party—is ground for granting a new trial, for setting aside a verdict, or for vacating, modifying, or otherwise disturbing a judgment or order. At every stage of the proceeding, the court must disregard all errors and defects that do not affect any party's substantial rights.").

Pruitt contends that it was prejudiced by the district court's exclusion of the declining-fees evidence because, had that evidence been admitted, "the jury would have had a compelling basis to award Caradigm hundreds of thousands, if not millions, in damages less than it did." Br. of Appellant at 29. In determining whether Pruitt was substantially prejudiced, we must consider "how much of an effect ... the improperly ... excluded evidence ha[d] on the verdict." Peat , 378 F.3d at 1162 ; see also Furcron v. Mail Ctrs. Plus, LLC , 843 F.3d 1295, 1304 (11th Cir. 2016) ("Substantial rights are affected if one cannot say, with fair assurance ... that the judgment was not substantially swayed by the error." (alteration in original) (quotation omitted)). In answering that question, we may consider, among other factors, "the closeness of the factual disputes (i.e., the strength of the evidence on the issues affected by the error)." Peat , 378 F.3d at 1162.

Our review convinces us that any error that the district court made in excluding Pruitt's proposed declining-fees evidence was harmless. One document Pruitt sought to admit was a Caradigm PowerPoint stating that Pruitt "[a]nticipate[d] 50 users for Concurrent Bill Review Use Case and 20 Users for Clinical Use Case pilot"—numbers, Pruitt says, that reflected a need for only a "small" platform. But that same document—indeed, on the very next page—acknowledged that the contract provided for a "[m]edium" platform for a "5 year term." And even if the number of users specified in the PowerPoint would have qualified Pruitt for the small platform, there's no indication that the other two metrics—the number of data feeds and the number of messages, See supra at 1265—would have decreased to such an extent that Pruitt would have moved down a size. Pruitt cited a second document from the contract's Statement of Work, which estimated that Pruitt would use "10 data feeds." But that estimate appears to pertain to just one of Pruitt's facilities, and the contract's order form confirmed that Pruitt needed a medium platform. Again, even if the "10 data feeds" estimate reflected the total number of data feeds, there's no information about the other two metrics—the number of users and the number of messages—indicating that Pruitt would have been moved down to the small platform.

Caradigm, moreover, introduced evidence to rebut Pruitt's declining-fees argument. In cash-flow predictions that Caradigm sent to Pruitt before it was notified of Pruitt's repudiation, for instance, Caradigm predicted that Pruitt would continue to use the medium-sized platform. Caradigm's account executive also testified that she didn't foresee Pruitt's use decreasing and that, in fact, "a lot of times customers start small" and then move to a larger platform as they bring in additional data or users. Trial Transcript Vol. I at 180.

In sum, we are not convinced that Pruitt's "substantial rights" were affected because "the judgment was not substantially swayed" by the district court's exclusion of Pruitt's declining-fees evidence. See Furcron , 843 F.3d at 1304 (quotation omitted).

* * *

Before proceeding to consider the interest and attorney's-fees issues, we pause briefly to recap our decision with respect to Pruitt's contract-damages arguments. In summary, we hold (1) that the contract didn't give Pruitt a right to demand subjective satisfaction with Caradigm's performance or obligate Caradigm to deliver a particular level of accuracy in patient matching; (2) that the district court didn't improperly reduce Caradigm's burden of proof at trial because satisfactory testing and First Productive Use were not conditions to further performance and, even if they were, Pruitt repudiated and thus couldn't claim that they wouldn't have occurred; and (3) that the district court's error in excluding Pruitt's declining-fees evidence was harmless because Pruitt has not shown that it was prejudiced.

B

We move on, then, to consider Pruitt's challenges regarding the jury's award of interest. The contract provided for interest in the "Payment Terms" provision, which stated, in relevant part:

[Pruitt] will pay all undisputed amounts due under this Agreement in full ... within 30 days after the invoice date. ... If [Pruitt], in good faith, disputes any amounts invoiced, [Pruitt] will report such dispute to Caradigm within 30 days of the invoice date. Failure to make timely payment is a material breach of this Agreement .... Any late amounts will be subject to interest at the lesser of 1.5% per month or the maximum allowed by applicable law.

Pruitt contends (1) that Caradigm's acceptance of Pruitt's breach precluded it from obtaining interest under the contract; (2) that Caradigm can't recover interest because it never invoiced Pruitt the total amount due as the interest provision requires; and (3) that, in any event, compound interest isn't permitted under Georgia law unless it is specifically authorized by the contract. We take up each argument in turn.

1

The gist of Pruitt's first argument is that once Caradigm accepted Pruitt's anticipatory repudiation, the contract—and with it, the contract's interest provision—was no longer operative. If Caradigm wanted to collect interest on the agreement, the argument goes, it should have treated the contract as remaining in force rather than treating the repudiation as a breach and suing. We disagree.

The way we read Georgia law, interest remained available under the contract's interest provision, even after Caradigm accepted Pruitt's repudiation. The cases that Pruitt cites for support simply restate the general principle that, after a repudiation is accepted, the contract at issue "cease[s] to be in effect." E.g. , Legacy Acad., Inc. v. Doles-Smith Enters., Inc. , 344 Ga.App. 805, 812 S.E.2d 72, 80 (2018). Importantly, though, a non-breaching party's acceptance of a repudiation means that the contract "cease[s] to be in effect" for purposes of requiring continued performance —the innocent party is "relieved of its obligation to continue performing under the ... agreement." Id. at 77 ; Larkins, supra , § 11:5 ("The effect of an anticipatory repudiation is to relieve the opposite party of the requirement to perform, or even to tender performance." (footnote omitted)). While accepting Pruitt's breach relieved Caradigm from further performance of its contractual promises, it did not preclude Caradigm from recovering interest under the contract.

Caradigm wasn't required to treat the contract as remaining in force and to sue for interest only as it accrued. Notwithstanding a repudiation, an innocent party can sue to "recover the entire present value ... of [the] contract," Parker v. King , 68 Ga.App. 672, 23 S.E.2d 575, 577 (1942) (emphasis added), and Pruitt's own cases confirm that in those circumstances, awards can still be made under the contract's terms, see, e.g. , Legacy , 812 S.E.2d at 74–75, 78 (stating that the nonbreaching party could "immediately sue to recover damages for such breach, including the present value of the future fees [i.e. , monthly advertising and royalty fees provided for in the contract]"). In the same way that an innocent party doesn't have to continue to perform after an anticipatory repudiation and wait to bring suit for damages as they come, Caradigm didn't have to wait for Pruitt's monthly payments to be late in order to recover interest under the contract.

2

Pruitt also contends that Caradigm isn't entitled to contractual interest because the agreement's interest provision applies only to amounts that were invoiced, and, so Pruitt says, Caradigm never sent it an invoice for the contract's entire value. We needn't decide whether the contract required invoicing of the full amount because, even assuming that Pruitt is right that invoicing was required, the jury could nonetheless have awarded damages by finding that it would have been "futile" for Caradigm to invoice Pruitt.

The district court instructed the jury as follows: "[T]he law does not require a futile or other useless act. Thus, if you find that it became futile or useless for Caradigm to invoice Pruitt a certain amount, then you should still award contractual interest on that amount." Trial Transcript Vol. IV at 62. Pruitt argues that this was error because "[f]utility under Georgia law relates to acts required by common law—not by contract." Br. of Appellant at 41 & n.18.

We hold that the district court's futility instruction was proper under Georgia law. For example, in Mayor & Aldermen of City of Savannah v. Batson-Cook Co. , the Georgia Court of Appeals held that a damages award was not precluded on account of a contractor's failure to comply with the contract's conditions for final payment, which required the contractor to, among other things, "submit a final accounting" and "submit a final application for payment." 310 Ga.App. 878, 714 S.E.2d 242, 247 (2011), rev'd on other grounds , 291 Ga. 114, 728 S.E.2d 189 (2012). The court explained that "[t]he law does not require a futile act," and because the parties disputed whether the contract price needed to be adjusted, "it would have been futile for [the contractor] to comply with the[ ] procedural requirements for final payment." Id. ; see also Allen v. Henderson , 48 Ga.App. 74, 172 S.E. 94, 95 (1933) ("The law does not require the doing of an unnecessary thing."); Larkins, supra , § 5:4 ("[U]nder some circumstances, performance of such a condition [i.e. , conditions regarding the time of performance] may be deemed futile and therefore excused."). Because the district court here didn't err as a matter of law in instructing the jury, it had "wide discretion as to the style and wording employed," and it didn't abuse its discretion. See Gowski v. Peake , 682 F.3d 1299, 1310 (11th Cir. 2012) (quotation omitted).

Moreover, there was sufficient evidence for the jury to find that invoicing Pruitt would have been futile. There is no dispute that Pruitt never responded to Caradigm's March 16, 2015 letter informing Pruitt that a breach would result in "the acceleration of payments owed to Caradigm for the contracted five-year period." There's also no evidence that Pruitt responded to or disputed other invoices sent after Caradigm was notified of the breach—from March through July 2015. The jury could have found that there would have been no point in Caradigm sending yet another invoice when Pruitt had failed to dispute or otherwise respond to invoices sent over the course of the previous five months.

3

Finally—with respect to interest, anyway—we consider Pruitt's argument that the district court erred in allowing the jury to compound the interest award. Pruitt contends that Georgia law requires that the interest award be calculated in simple terms because the parties’ contract didn't explicitly provide for compound interest.

Before addressing the merits of Pruitt's compound-interest argument, we must determine whether it was waived, as Caradigm contends. The district court refused to consider Pruitt's argument following the jury's verdict on the ground that Pruitt had failed to sufficiently raise the compound-interest issue. In its order denying Pruitt's motion for judgment as a matter of law, the court acknowledged that the parties’ pretrial order had memorialized Pruitt's position that "normal damages principles would apply," but it held that this statement was "insufficient" to preserve an objection to compounding any interest award. In its order denying Pruitt's motion for a new trial or to alter or amend the judgment, the district court further explained that Pruitt had "waived any argument regarding compound interest by not including the issue in the pretrial order, objecting to Caradigm's evidence on compound interest, or moving for a directed verdict on the issue at the close of evidence."

We ordinarily give district courts wide discretion in enforcing the limits of pretrial orders. See, e.g. , Morro v. City of Birmingham , 117 F.3d 508, 515 (11th Cir. 1997) ("We have not hesitated to back up district courts when they put steel behind the terms of pretrial orders and hold parties to them."). We have recognized that district courts must "firmly (but fairly) enforce[ ]" pretrial orders in order to narrow and define the issues to be litigated at trial. Id. ; see also Lexington Ins. Co. v. Cooke's Seafood , 835 F.2d 1364, 1368 (11th Cir. 1988) ("Given the vast number of details competing for the attention of a federal district judge, reducing all issues to writing before the pretrial conference substantially assists the trial court in its ability to understand the issues and to prepare for trial."). Consequently, we have held that "[t]he decision of the trial court to modify or enforce a pre-trial order will not be disturbed absent a showing of abuse of discretion." Santiago v. Lykes Bros. S.S. Co. , 986 F.2d 423, 427 (11th Cir. 1993).

Based on the unique circumstances presented in this case, however, we conclude that Pruitt didn't waive its compound-interest argument by failing to provide additional detail in the pretrial order. To be sure, questions of law, like other issues, can be considered waived if not preserved in a pretrial order. Cf. Morro , 117 F.3d at 516 n.3 (explaining that "[c]ounsel may waive the right to have an issue decided by failing to identify the issue to the court at the pretrial conference, regardless of whether the issue is a legal or factual one"). The compound-interest award here, though, isn't just a question of law, but also—as we explain below—a routine matter with respect to which the underlying law is very clear. Although a short statement that "normal damages principles will apply" might not be sufficiently detailed to encompass some legal arguments, we are satisfied that it fairly reflects the baseline assumption that under Georgia law any award of interest would be simple, not compound. We can't fault Pruitt for failing to state more explicitly that simple interest should apply given—again, as we will explain—that it clearly appears to be the default (i.e. , "normal") legal rule.

Moreover, while our precedent suggests that pretrial orders are intended to avoid "blindsid[ing]" the district court and opposing parties, there was no "ambush" here. Cf. id. at 512, 516 & n.3 (affirming the district court's refusal to consider an affirmative defense that wasn't raised until the charge conference). Both Caradigm and the district court were on notice prior to trial that Pruitt believed it was entitled to a simple-interest calculation because Pruitt had stated in its trial brief (albeit in a footnote) that "even if Caradigm persuades the jury that it should recover prejudgment interest (whether contractual or statutory), Georgia law requires such interest to be calculated as simple interest, not compound interest." Defendant's Trial Br. at 3 n.1. Pruitt also raised the issue when cross-examining Caradigm's damages expert and in its renewed motion for judgment as a matter of law. We thus conclude that Pruitt sufficiently objected to an award of compound interest in the district court.

On the merits, we hold that the district court erred in awarding compound interest. Section 13-6-13 of the Georgia Code states that "[i]n all cases where an amount ascertained would be the damages at the time of the breach, it may be increased by the addition of legal interest from that time until the recovery." The Code further defines the "legal rate of interest," in relevant part:

(a)(1)(A) The legal rate of interest shall be 7 percent per annum simple interest where the rate percent is not established by written contract . Notwithstanding the provisions of other laws to the contrary ... the parties may establish by written contract any rate of interest, expressed in simple interest terms ... where the principal amount involved is more than $3,000.00 but less than $250,000.00 ....

(B) Where the principal amount is $250,000.00 or more , ... the parties may establish by written contract any rate of interest, expressed in simple interest terms or otherwise , and charges to be paid by the borrower or debtor.

Ga. Code Ann. § 7-4-2 (emphasis added). Section 7-4-2 suggests that compound contractual interest is permissible where, as here, "the principal amount is $250,000.00 or more" because it states that parties may choose "any rate of interest, expressed in simple interest terms or otherwise ." Importantly for our purposes, though, the parties must "establish" that rate "by written contract." Under Georgia law, then, it's clear that parties must explicitly agree to compound interest in their contract.

In Noons v. Holiday Hospitality Franchising, Inc. , the defendant argued that the trial court had erred in awarding the plaintiff "interest on interest" under the parties’ agreement, which provided for interest of "1.5% per month or the maximum interest permitted by applicable law, whichever is less." 307 Ga.App. 351, 705 S.E.2d 166, 170 (2010) (alteration adopted). The plaintiff relied on § 7-4-16 (which governs the interest rate on "commercial accounts") to support the compound award, but the court held that § 7-4-16 didn't apply to breach-of-contract suits. Id. Accordingly, the court concluded that the plaintiff's entitlement to interest was "a matter governed by the contract at issue" and awarded simple interest of 1.5% on all past-due fees. Id. The language of the contract—providing for interest of "1.5% per month"—wasn't clear enough to justify an award of compound interest. Id.

Just so here. The parties’ agreement states that late amounts under the contract would "be subject to interest at the lesser of 1.5% per month or the maximum allowed by applicable law"—a provision almost identical to the one at issue in Noons . Following the Georgia Code and Noons , we hold that the language in the contract to the effect that interest would be awarded at "1.5% per month" does not establish that the parties agreed to compound interest. We will therefore vacate the compound-interest award and remand for the district court to calculate simple interest.

C

Finally—as in finally finally—attorney's fees. The jury awarded attorney's fees to Caradigm on the ground that Pruitt had been "stubbornly litigious" within the meaning of Georgia Code Annotated § 13-6-11, which states:

The expenses of litigation generally shall not be allowed as a part of the damages; but where the plaintiff has specially pleaded and has made prayer therefor and where the defendant has acted in bad faith, has been stubbornly litigious, or has caused the plaintiff unnecessary trouble and expense, the jury may allow them.

In general, a stubborn-litigiousness fee award under § 13-6-11 will be precluded if there is evidence of a "bona fide controversy" between the parties. David G. Brown, P.E., Inc. v. Kent , 274 Ga. 849, 561 S.E.2d 89, 90–91 (2002). Pruitt asserts that the award of attorney's fees was erroneous because the district court (1) excluded evidence relevant to whether a bona fide controversy existed and (2) improperly instructed the jury on stubborn litigiousness. We consider each argument in turn.

1

Pruitt sought to introduce two categories of evidence that it claims were relevant to the stubborn-litigiousness issue: (a) evidence relating to its reasons for breaching the contract; and (b) communications that took place between the parties after Caradigm filed suit.

a

In the first category, Pruitt wanted to introduce evidence showing that it thought it could cancel the contract because of Caradigm's perceived inability to provide patient matching. The district court held that "[a] bona fide controversy as to Caradigm's patient-matching capabilities is irrelevant to whether or not there was a bona fide controversy as to Pruitt's belief it could terminate the contract in the way that it did." Dist. Ct. Order on Emergency Motion at 13–14. The court further concluded that patient-matching evidence "would cause confusion to the jury that even a curative instruction could not remedy." Dist. Ct. Pretrial Conference Order at 3.

The district court did not abuse its discretion in excluding evidence of Pruitt's breach and more specifically, Pruitt's contention that Caradigm wasn't able to perform patient matching. As we've explained, the contract didn't require that Caradigm provide a certain level of accuracy in patient matching, expressly disclaimed any implied warranties guaranteeing otherwise, and warned that there might be false matches. See supra at 1268–70. Pruitt's subjective satisfaction with Caradigm's patient-matching capabilities was irrelevant to whether a bona fide controversy existed between the parties. Moreover, and as the district court held, any evidence in this regard would have risked confusing the jury or causing prejudice.

Notably, Pruitt wasn't left without recourse. The district court allowed Pruitt to introduce evidence of its understanding of the agreement's termination provision and to "refer to a disagreement over a contract provision as long as it d[id] not directly—or indirectly—allude to the quality of patient matching, its own subjective satisfaction with patient matching, or Caradigm's ability to provide a particular level of patient matching." Dist. Ct. Order on Emergency Motion at 15. We can't say that the district court's handling of this evidence—including the alternative course it left open for Pruitt—constituted an abuse of discretion.

b

Pruitt separately sought to admit four letters that the parties’ counsel exchanged after Caradigm filed suit. In them, each party outlined its theories of why the other had breached and briefly discussed resolving the conflict to prevent further litigation. The district court rejected Pruitt's request to amend the pretrial order to add the letters as trial exhibits, concluding that "attorney correspondence written after the commencement of litigation is not relevant evidence of Pruitt's beliefs at the time it terminated the contract or in the months thereafter," and that post-filing letters "may be post hoc rationalizations for a party's behavior." Dist. Ct. Pretrial Conference Order at 4–5. On the eve of trial, the court refused to consider Pruitt's argument that the letters, although arising in post-filing communications, could be evidence of a bona fide pre-filing controversy. While Pruitt made this argument in a footnote in its trial brief, the court explained that "Pruitt did not make this argument when it sought to have the pretrial order amended at the pretrial conference (despite the Court specifically asking about this issue)," nor did it "cite any of the case law contained in its footnote [of its trial brief]." Dist. Ct. Order on Emergency Motion at 7 n.3.

The district court did not abuse its discretion in excluding the letters. On balance, we read Georgia law as requiring that any evidence of a bona fide dispute relate to "the conduct arising from the transaction underlying the cause of action being litigated, not conduct during the course of the litigation itself." Brown , 561 S.E.2d at 90 ; see also Padgett v. Moran , 167 Ga.App. 244, 306 S.E.2d 96, 97 (1983) ("Allowance of attorney fees pursuant to [Georgia Code Annotated] § 13-6-11... based upon a party's conduct in the course of litigation is wholly improper."). The letters that Pruitt sought to admit were part of the parties’ litigation conduct. They were drafted by the parties’ lawyers and written months after Caradigm filed suit—and after the cure period provided for in the contract had passed.

It's true that some Georgia authority supports Pruitt's assertion that evidence created after a lawsuit's filing may be relevant to whether a bona fide controversy existed. See, e.g. , Spring Lake Prop. Owners Ass'n v. Peacock , 260 Ga. 80, 390 S.E.2d 31, 32 (1990) (holding that trial testimony showed that, prior to the litigation, there was no bona fide controversy). But Pruitt didn't make this argument when it asked the district court to amend the pretrial order. And in any event, while it seems that post-filing evidence could be relevant to a pre-filing dispute, the letters here so clearly evidence litigation conduct that we are not persuaded that they accurately reflect the parties’ pre-suit positions.

Indeed, the letters are inconsistent with other pre-litigation communications between the parties. In the August 10, 2015 letter, Pruitt—through its counsel—stated that it "declined to move forward with the project" because it "became increasingly dissatisfied" with Caradigm's software and services. In contrast, when notifying Caradigm that it was abandoning the project on February 17, 2015, Pruitt didn't suggest that it was doing so because of its dissatisfaction with Caradigm's capability to perform; rather, it simply told Caradigm that it had "suspended all efforts dedicated to th[e] project" and that the contract was "no longer valid." This inconsistency suggests that the letters—rather than evidencing a bona fide pre-litigation dispute—were, in fact, a post hoc attempt to explain Pruitt's pre-litigation behavior in a way that squared with its litigation strategy.

Thus, the district court did not err in excluding the letters as irrelevant to the stubborn-litigiousness issue. The district court did not "ma[k]e a clear error of judgment or ... appl[y] an incorrect legal standard," Furcron , 843 F.3d at 1304 (quotation omitted), and we therefore affirm its evidentiary rulings.

2

Pruitt separately argues that the district court's jury instructions on "stubborn litigiousness" misstated Georgia law. The jury instructions stated, in relevant part: "Attorney fees are not authorized under [ Georgia Code Annotated §] 13-6-11 if the evidence shows that a genuine dispute existed in regards to liability. Where a bona fide controversy existed, an award of attorney's fees is inappropriate." Pruitt contends that, under Georgia law, attorney's fees could not be awarded if the jury found a bona fide dispute existed "whether of law or fact, on liability or amount of damages , or on any comparable issue." Br. of Appellant at 46 (emphasis added).

On our reading, the Georgia Court of Appeals’ decisions conflict on the question whether the bona fide dispute must be about liability alone, or whether it can be about liability or damages. Compare, e.g. , Daniel v. Smith , 266 Ga.App. 637, 597 S.E.2d 432, 438 (2004) ("Our research indicates that a jury may award attorney fees under ... § 13-6-11 if there is no bona fide controversy as to liability, even if there is a bona fide controversy as to damages."), with, e.g. , Hester Enters., Inc. v. Narvais , 198 Ga.App. 580, 402 S.E.2d 333, 336 (1991) ("Even if the evidence shows no bona fide dispute as to the [defendant's] liability for breach of the construction contract, the evidence nevertheless shows a bona fide dispute as to the amount of damages [the plaintiffs] were authorized to recover."). The district court carefully considered the conflicting cases put forward by the parties and decided that a dictum in one Georgia Supreme Court case was the "tie-breaker." Trial Transcript Vol. III at 209. In that case, the court stated that "statutory recovery for stubborn litigiousness or causing unnecessary trouble and expense is authorized if there exists no bona fide controversy or dispute regarding liability for the underlying cause of action." Brown , 561 S.E.2d at 90–91 (emphasis added).

We conclude that the district court didn't err in relying on Brown to break the tie. True, whether or not a bona fide controversy over damages could defeat a stubborn litigiousness claim wasn't squarely at issue in Brown . See id. at 89 (explaining that the issue was whether § 13-6-11 allows attorney's fees for appellate proceedings). But we simply can't say that the district court reversibly erred by relying on a dictum from the Georgia Supreme Court when faced with diverging Georgia Court of Appeals decisions. After all, we have explained that "[a]lthough dicta [in a state-court decision] is not binding, it can provide federal courts with insight into a state court's thinking." LeFrere v. Quezada , 582 F.3d 1260, 1267 (11th Cir. 2009).

Even if we were to give Pruitt the benefit of the doubt and conclude that the district court erred in its instruction, we wouldn't reverse because Pruitt didn't suffer prejudice. See Oladeinde v. City of Birmingham , 230 F.3d 1275, 1295–96 (11th Cir. 2000) ("We will not overturn a jury verdict because of an erroneous jury instruction unless there is also a showing of prejudice." (quotation omitted)); see also Jennings v. BIC Corp. , 181 F.3d 1250, 1254 (11th Cir. 1999) ("[W]e will find reversible error in the refusal to give a requested instruction only if ... the failure to give the instruction resulted in prejudicial harm to the requesting party." (quotation omitted)). Were we to consider the excluded evidence, it doesn't show a bona fide pre-litigation dispute as to either liability or damages. As explained, any evidence of Pruitt's dissatisfaction with Caradigm's patient matching was irrelevant to the bona fide dispute inquiry because the contract didn't contain a satisfaction clause and expressly disclaimed the accuracy of patient matching. See supra at 1268–70. The post-filing letters exchanged between the parties’ lawyers are also irrelevant with respect to whether a bona fide dispute existed before the filing of the lawsuit. See supra at 1281–83.

So far as we can tell, no other evidence produced at trial showed that there was any bona fide pre-litigation dispute between the parties. In its email notifying Caradigm of its breach, Pruitt said only that it had "suspended all efforts dedicated to [the] project" and that the contract was "no longer valid." When Caradigm's CEO sent Pruitt the March 16, 2015 letter stating that Pruitt didn't have a unilateral right to terminate the contract, Pruitt never even responded. Pruitt likewise failed to dispute invoices sent by Caradigm over a five-month period. In sum, Pruitt hasn't produced any evidence of a bona fide pre-filing dispute regarding either liability or damages, rendering any possible error in the district court's jury instructions harmless. Cf. Rand v. Nat'l Fin. Ins. Co. , 304 F.3d 1049, 1052 (11th Cir. 2002) (holding that "there was no prejudice" stemming from the district court's failure to give a jury instruction on a particular defense where the defendant "failed to produce any evidence" on "an essential element of that defense").

III

In conclusion, we cannot say that the district court reversibly erred in handling the damages trial in most of the ways that Pruitt contends that it did. Moreover, with respect to the award of contract damages and attorney's fees, in any instances where the district court erred or might have erred, the errors did not cause Pruitt prejudice. Finally, however, while we conclude that Caradigm is entitled to recover interest under the parties’ agreement, we hold that the district court erred in awarding compound interest. Accordingly, we affirm the awards of contract damages and attorney's fees, as well as the determination that Caradigm is entitled to recover contractual interest, but we vacate the compound-interest award and remand so that the district court can calculate it in simple terms.

AFFIRMED IN PART, VACATED IN PART, AND REMANDED.