After the second bench trial in this contract-interpretation dispute between two agricultural-chemical distributors, the trial court lamented that “much of the evidence submitted at trial does not aid the court in determining the parties’ intent.” The parties clarified nothing on appeal, so we decline to depart from the trial court’s interpretation of the parties’ ambiguous contract. We AFFIRM.
I.
In 1958, Novartis Crop Protection’s corporate predecessor registered atrazine as a pesticide under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), allowing it to sell and distribute atrazine in the United States. Before registering atrazine under FIFRA, the Environmental Protection Agency (EPA) reviewed health, safety, and environmental studies generated and submitted by Novartis’s predecessor to confirm that atrazine “will not generally cause unreasonable adverse effects on- the- environment.” 7 U.S.C. § 136a(c)(5)(D).
In the 1970s, plaintiff Drexel Chemical Company obtained its own registration under FIFRA, allowing Drexel to sell and distribute atrazine. To obtain the registration, Drexel cited Novartis’s health, safety, and environmental data and, as FI-FRA mandates, offered to pay Novartis for producing that data. 7 U.S.C. § 136a(c)(l)(F)(iii). The pesticide industry calls such payments “data compensation” and initial registrants like Novartis “original data submitters.” 7 U.S.C. § 136a(c)(l)(F). Through FIFRA-mandat-ed binding arbitration, 7 U.S.C. § 136a(c)(l)(F)(ni), (c)(2)(B)(iii), Drexel and Novartis split data-compensation costs on a market-share basis. Because the EPA often requests additional pesticide data, whose costs all registrants must share, 7 U.S.C. § 136a(c)(2)(B)(ii), Drexel and Novartis agreed to split future data-compensation costs on the same market-share basis to avoid further disputes.
In the late 1990s, defendant Albaugh purchased a majority share in a foreign company that formulated atrazine. Before Albaugh could import and distribute atra-zine in the United States, it too had to register with the EPA. 7 U.S.C. § 136a(a). Albaugh first offered to pay Novartis to cite its health, safety, and environmental studies. Dissatisfied with the data compensation Albaugh offered, Novartis threatened to oppose Albaugh’s registration with the EPA.
Drexel hatched a solution: a supplemental distributorship. This relationship, blessed by FIFRA, allows a distributor— *469Albaugh — to sell pesticides under its own brand name but another entity’s — Drex-el’s — FIFRA registration, provided that the participants notify the EPA. See generally 40 C.F.R. § 152.132. In return for Albaugh’s use of Drexel’s atrazine registration, Albaugh would pay Drexel $750,000 in installments and “fifty percent (50%) of the costs of all future payments that Drexel makes to Novartis in the future as data compensation in order to maintain the Registration.” Drexel and Albaugh signed the contract in December 1998.
Meanwhile, the EPA was reviewing all atrazine registrations, concerned about the pesticide’s carcinogenicity. Novartis, and later Syngenta Crop Protection,1 submitted 190 studies to the EPA to prove otherwise. Concluding its review in 2003, the EPA cancelled all existing atrazine registrations, including Drexel’s. But it permitted reregistration, provided that the registrants offered to pay Syngenta for 100 selected studies that proved atrazine’s safety (“the Appendix B studies”) and also agreed to conduct more safety studies the EPA would identify in the future (data call-ins or “DCIs”). Those who refused the EPA’s reregistration terms could no longer import or sell atrazine. Drexel, then in the cost-splitting contract with Al-baugh, accepted the EPA’s terms, reregis-tered, and offered to pay Syngenta for the Appendix B studies. In 2004 and 2005 DCIs, the EPA asked atrazine registrants for additional safety studies. Syngenta undertook these studies alone, electing to recoup the costs from other atrazine registrants later. Both years Drexel timely offered to compensate Syngenta for the DCIs. In short, Drexel thrice offered to pay Syngenta during the contract period— for the Appendix B studies, the 2004 DCI, and the 2005 DCI — but made no actual payments.
In summer 2006, Drexel learned that Albaugh “was thinking about getting out of the [a]trazine business.” Drexel began to worry about paying Syngenta without Al-baugh’s help. Accordingly, Drexel asked Syngenta to invoice it $1.5 million in data-compensation costs. Drexel paid Syngenta and invoiced Albaugh for $750,000 in September 2006. Albaugh immediately terminated the contract. It ceased importing atrazine but continued to sell existing inventory until April 2007.
Syngenta initiated binding arbitration with Drexel in May 2007 to fix Drexel’s data-compensation costs for the Appendix B studies, 2004 DCI, and 2005 DCI. The arbitrators awarded Syngenta $131,3402 for the Appendix B studies and $2,393,662 for the 2004 and 2005 DCIs combined. The arbitrators again used the market-share approach to allocate costs. Drexel’s market share included both Drexel’s (5.1% for the Appendix B studies-1 and 6.2% for the DCIs) and Albaugh’s (1.2% for the Appendix B studies and 1.8% for the DCIs) atrazine market shares. Drexel and Syngenta split the arbitration costs. Some of the 2004 and 2005 DCI studies were still in progress, so the arbitrators directed the parties to split future costs under the same market-share approach. As of 2012, Drexel has paid $5,578,518.97 to Syngenta..
In 2008, Drexel sued Albaugh for breach of contract, seeking half its payments to Syngenta. Initially, the court held Al-baugh liable for fifty percent of actual *470payments that Drexel made to Syngenta as data compensation before Albaugh’s termination on September 29, 2006. It therefore entered a $750,000 judgment for Drexel, plus prejudgment interest, representing half of the $1.5 million Drexel paid to Syngenta before Albaugh terminated.
The parties cross-appealed. We affirmed the $750,000 judgment but remanded for further proceedings. Drexel Chem. Co. v. Albaugh, Inc., 489 Fed.Appx, 63, 69 (6th Cir.2012) (Drexel I), We found the contract ambiguous “as to whether Al-baugh is responsible for payments that Drexel makes after the termination of the Agreement but relating to the period that the Agreement was in effect,” ie., the post-termination payments Drexel has made and will make to Syngenta for the Appendix B studies and the 2004 and 2005 DCIs. Id. at 67. We remanded for factual development, suggesting that the lower, court consider: how the agricultural-chemical business understands the contract terms, the FÍFRA regulatory regime, the nature of Drexel’s payments to Syngenta, correspondency between Drexel and Al-baugh, and Albaugh’s post-termination reliance on Drexel’s registration. Id.
On remand, the court held Albaugh liable for half of Drexel’s payments for the studies Syngenta either generated or submitted to the EPA before Albaugh’s termination on September 29, 2006. To set Drexel’s damages, the court relied on ex--Mbits listing Syngenta’s Appendix B studies and the 2004 and 2005 DCI studies and their costs. These lists dated Syngenta’s studies by year only, making a precise damages award to September 29 impossible, The court therefore included all 2006 studies in Drexel’s damages, reasoning “that it is more likely than not that the studies dated 2006 were generated or submitted by Syngenta prior to Albaugh’s termination.” It awarded Drexel $841,726, plus prejudgment interest, representing half the payments Drexel made to Syngen-ta for studies dated 2006 and earlier, less the $750,000 affirmed in the first appeal. The court denied Drexel’s demand that Albaugh pay for half its arbitration costs, including attorney’s fees.
The parties once again cross-appeal. Drexel demands half its $5.5 million total payments to Syngenta, while Albaugh seeks reversal of the $841,726 awarded in the second bench trial.
II.
We must determine whether the trial court correctly interpreted two warring provisions from Drexel and Albaugh’s 1998 contract: (1) “Albaugh shall pay fifty percent (50%) of the costs of all future payments that Drexel makes to Novartis in the future as data compensation in order to maintain the Registration,” (the future-payment provision) and (2) “Albaugh may also terminate this agreement at any time after final payment of the $750,000 called for [under the contract]” (the termination provision).
Tennessee contract law governs our inquiry and requires us “to ascertain and give effect to the intent of the parties.” Allstate Ins. Co. v. Watson, 195 S.W.3d 609, 611 (Tenn.2006) (citing Christenberry v. Tipton, 160 S.W.3d 487, 494 (Tenn.2005)). Because we already deemed the contract’s plain language ambiguous, Drexel I, 489 Fed.Appx. at 67, we apply “established rules of construction to determine the intent of the parties.” Allstate, 195 S.W.3d at 611 (citing Planters Gin Co. v. Fed. Compress & Warehouse Co., 78 S.W.3d 885, 890 (Tenn.2002)). We review the trial court’s conclusions of law de novo and accept its findings of fact unless they are clearly erroneous. Smith v. Jefferson Cty. Bd. of Sch. Comm’rs, 788 F.3d 580, 585-86 (6th Cir.2015) (citation omitted).
*471A. Albaugh’s Arguments
Under Albaugh’s proposed interpretation, its liability to Drexel included only actual payments made by Drexel to Syn-genta before Albaugh terminated on September 29, 2006. By paying half of Drex-el’s $1.5 million pre-termination payment to Syngenta after the first bench trial, Albaugh satisfied its sole contractual liability. Albaugh offers two arguments in favor of its interpretation: (1) the trial court failed to construe the future-payment provision against Drexel, the drafter, and (2) the court ignored the parties’ supplemental-distributor relationship.
1. Construe Against Drafter
Drexel drafted the future-payment provision, and it remained materially unchanged in every draft the parties circulated. Albaugh therefore urges the court to construe the provision in its favor, citing Tennessee contract cases construing ambiguous provisions against the provisions’ drafters. See, e.g., Realty Shop, Inc. v. RR Westminster Holding, Inc., 7 S.W.3d 581, 598 (Tenn.Ct.App.1999); Nat’l Garage Co. v. George H. McFadden & Bro., 542 S.W.2d 371, 374 (Tenn.Ct.App.1975).
But Albaugh ignores the trial court’s finding that both the future-payment and termination provisions were ambiguous, particularly when read together, and that Albaugh drafted the termination provision. The court therefore construed each provision against its drafter, refusing to extend the future-payment provision in perpetuity for Drexel or to favor the termination provision over the future-payment provision for Albaugh.
As to the future-payment provision, in Drexel I, we noted that “[t]he language of the [provision] could even be said to have no outer limit for when Albaugh’s payment obligations end, if read strictly literally.” 489 Fed.Appx. at 67. The trial court construed this open-ended language against Drexel, its drafter, cutting off Albaugh’s obligation for future payments when it left the atrazine market.3 And the court construed the termination provision against Albaugh. Although the contract twice referenced “future” payments, Albaugh’s termination provision never accounted for such payments. Accordingly, the court declined to elevate the termination provision’s silence on future payments over the future-payment provision. This argument avails neither Albaugh nor Drexel.
2. Supplemental Distributorship
Next, Albaugh argues that the trial court overlooked Drexel and Albaugh’s supplemental-distributor relationship and its effect on the contract. But we agree with the trial court that the supplemental distributorship provides no interpretive assistance. True, supplemental distributors shoulder limited regulatory responsibilities under FIFRA — they need not pay data compensation to the original data submitter, for example. That responsibility falls to the registrant, who typically prices data-compensation costs into the supplemental-distributor agreement. But Al-baugh contractually expanded its limited regulatory responsibilities through the future-payment provision. As the trial court correctly noted, “the parties’ rights and obligations were [therefore] governed by *472the terms of the Agreement,” not FIFRA’s supplemental-distributor regulations. Al-baugh fails to persuade us that the court erred in rejecting its proposed interpretation.
B. Drexel’s Arguments
Drexel presses an alternate interpretation: Albaugh must pay half of every cost obligation Drexel incurred before termination. Drexel incurred three such cost obligations — the Appendix B studies, the 2004 DCI, and the 2005 DCI. Drexel therefore seeks half the $5.5 million it paid to Syngenta as of 2012 and asks us to interpret the contract accordingly. To support its proposed interpretation, Drexel cites the FIFRA regulatory regime and Albaugh’s reliance on its registration. Drexel also protests the court’s refusal to award as damages half of Drexel’s costs and attorney's fees for the arbitration with Syngenta.
1. FIFRA Regulatory Regime
Drexel claims the FIFRA regulatory regime renders its offers to pay Syngenta “irrevocable” once made. Because those offers immediately triggered Drexel’s irrevocable obligation to pay, the contract immediately imposed on Albaugh a duty to pay Drexel. In other words, when Drexel offered to pay Syngenta, all payments resulting from Drexel’s offers instantly and irreversibly became “future payments” for which Albaugh was liable under the contract.
Drexel’s claim that offers to pay are irrevocable wilts under scrutiny. Drexel fails to cite any FIFRA provision that supports this contention and for good reason. The FIFRA regulatory regime largely cedes payment disputes to the private sector. Once the EPA secures proof that a party has offered to pay, it leaves the parties alone to negotiate, and if necessary arbitrate, over compensation. See 7 U.S.C. § 136a(c)(l)(F)(iii) (“Registration action by the Administrator shall not be delayed pending the fixing of compensation.”). Indeed, Congress amended FI-FRA to relieve the EPA of data-compensation responsibilities. Thomas v. Union Carbide Agric. Prods. Co., 473 U.S. 568, 573, 105 S.Ct. 3325, 87 L.Ed.2d 409 (1985) (noting that FIFRA data compensation “do[es] not require active government involvement” and “should be determined to the fullest extent practicable, within the private sector” (quoting 123 Cong. Rec. 25710 (1977))). Whether Syngenta ever collects on Drexel’s offer to pay depends wholly on Syngenta — if Syngenta chooses not to demand binding arbitration, neither the EPA nor FIFRA regulations would step in.
Evidence at trial reinforced that the private sector, not FIFRA, dictates the outcome- of an offer to pay. For example, Albaugh’s expert testified that original data submitters like Syngenta tend not to pursue claims for data compensation against companies that abandon their registrations and exit the pesticide market. Drexel’s Vice President of Growth and Development agreed. He admitted that companies sometimes make offers to pay, later abandon the underlying registration, and pay no data-compensation costs arising from that offer to pay. He also testified that the opposite can be true — companies that abandon their registrations after an offer to pay sometimes pay data compensation to particularly zealous original data submitters. And Albaugh’s in-house counsel testified that original data submitters have never pursued Albaugh for costs after Albaugh cancelled registrations for other pesticides but had previously offered to pay.
The FIFRA regulatory regime provides no support for Drexel’s contention that its *473offer to pay immediately and irrevocably triggered either its or Albaugh’s duty to pay data compensation. Mixed real-world experiences on the issue also undermine Drexel’s argument.
2. Albaugh’s Reliance
Drexel next contends that the court allowed Albaugh a free ride on its atrazine registration, with Albaugh importing 6,029,742 gallons of atrazine during the contract period but failing to pay its share of the regulatory costs incurred. Meanwhile, Drexel maintained the registration despite three opportunities to let it lapse. If it had refused to offer to compensate Syngenta for the Appendix B studies, 2004 DCI, or 2005 DCI, the EPA could have cancelled or suspended Drexel’s atrazine registration, which would have halted Al-baugh’s atrazine importation and sales. See 7 U.S.C. §§ 136a(c)(l)(F)(iii) (instructing the EPA to cancel pesticide registrations of applicants who refuse to offer to pay for the data cited in their registration applications), 136a(c)(2)(B)(ii) (instructing the EPA to suspend pesticide registrations for registrants who refuse to share DCI costs). In short, Drexel incurred costs under the contract for Albaugh’s benefit, but received nothing from Albaugh for its trouble.
Albaugh unquestionably benefitted from and relied on Drexel’s registration during the contract period. But Albaugh’s reliance supports the trial court’s interpretation, not Drexel’s. Albaugh relied on Drexel to maintain the atrazine registration with the EPA as provided in the contract. Drexel performed. In turn, the court held Albaugh to its duties under the contract — half of Drexel’s “future payments” to Syngenta. The studies Syngen-ta was generating and submitting to the EPA during the contract period represented Drexel’s “future payments” to Syngen-ta. The trial court ordered Albaugh to cover half those costs. Albaugh’s reliance merits no damages beyond what Drexel has already received.
' 3. Arbitration Costs and Attorney’s Fees
Finally, Drexel argues that the court erred in refusing to award Drexel attorney’s fees and costs from the Syngenta arbitration as part of the “costs of all future payments” to Syngenta. Drexel maintains that Albaugh’s $750,000 initial contract payment represented half of Drexel’s total costs from the first arbitration with Novartis, attorney’s fees and costs alike. Because Albaugh’s $750,000 initial contract payment included arbitration costs, Drexel presses, the “costs of all future payments” must also include such costs. Further, Drexel notes that the contract differentiates between the “payments” to Syngenta and the “costs” of those payments. To give effect to all the contract terms, Drexel claims, Albaugh must pay Drexel’s arbitration costs and attorney’s fees — the “costs” of the data-compensation “payments.”
Drexel fails to convince. To begin, as the court noted, Drexel never told Albaugh that its $750,000 initial contract payment encompassed Drexel’s out-of-pocket costs for the first Novartis arbitration, and Drexel may not unilaterally impose its understanding on Albaugh. Next, Drexel first advanced its demand for fees and costs in the second bench trial. Indeed, its pre-suit demand letter to Albaugh asked for “fifty percent of any future payments Drexel makes to Syngenta (Novartis’ successor) to maintain the atrazine registrations,” omitting costs and fees. Finally, Tennessee contract law demands more specificity to award attorneys fees. “The only way parties to a contract have been able to specifically and expressly create a *474right to recover attorney fees has been by incorporating the phrase ‘including reasonable attorney fees’ or some other similar, yet equally specific, contractual language.” Cracker Barrel Old Country Store, Inc. v. Epperson, 284 S.W.3d 303, 310 (Tenn.2009) (finding that a contract provision for “expenses” lacked the specificity necessary to award attorney’s fees). Drexel offers no reason to question the court’s refusal to award fees and costs.
C. Damages
The parties having failed to undermine the trial court’s contract interpretation, we turn to the damages calculation. Though the court purported to calculate damages based on the costs of studies Syngenta generated or submitted to the EPA before the September 2006 termination, it eventually included all 2006 studies. It justified its award in two ways. First, given that Albaugh terminated well into the latter half of 2006, the court found “that it is more likely than not that the studies dated 2006 were generated or submitted by Syngenta prior to Albaugh’s termination.” Second, the court found that Albaugh relied on Drexel’s registration to sell its atrazine inventory for all of 2006 and into 2007.
We review awards of money damages under an abuse-of-discretion standard. Hance v. Norfolk S. Ry. Co., 571 F.3d 511, 517 (6th Cir.2009) (citing United States v. City of Warren, 138 F.3d 1083, 1094-97 (6th Cir.1998)). “A court abuses its discretion when it relies on clearly erroneous findings of fact, or when it improperly applies the law or uses an erroneous legal standard.” Id. (quoting City of Warren, 138 F.3d at 1096). In Tennessee, “[damages for breach of contract are permissible even when the plaintiff is unable to prove the exact amount of those damages.” BancorpSouth Bank, Inc. v. Hatchel, 223 S.W.3d 223, 230 (Tenn.Ct.App.2006) (citing Provident Life & Accident Ins. Co. v. Globe Indent. Co., 156 Tenn. 571, 3 S.W.2d 1057, 1058 (1928)). The party seeking damages must provide sufficient evidence for the trier of fact, “using its discretion, to make a fair and reasonable assessment of damages ... within a reasonable degree of certainty.” Id. (internal citations and quotation marks omitted).
Citing Syngenta’s imprecise dating, Al-baugh argues that Drexel failed to show that any of Syngenta’s 2006 studies were generated or submitted to the EPA before Albaugh’s September 29 termination. Moreover, Albaugh claims its post-termination atrazine sales prove nothing — it relied on an EPA policy allowing supplemental distributors to sell inventory “up to 18 months after termination of the [supplemental-distributor] Agreement.” Because Drexel failed to carry its burden of proof on damages, Albaugh asks us to subtract the costs of all 2006 studies ($204,869.18) from the damages award.
Albaugh misconstrues the EPA policy. That policy allows “a supplemental distributor 18 months to sell and distribute existing stocks after the basic registrant terminates the supplemental registration.” In other words, the policy protects distributors from unexpected registration cancellations. The policy is silent on the cancellation of the supplemental distributorship by the supplemental distributor. Nothing in the record indicates that Drexel ever terminated the atrazine registration, so Al-baugh was, in fact, relying on Drexel’s registration in selling out its inventory.
The court’s observation that Syngenta likely generated or submitted the 2006 studies before September plus Albaugh’s post-termination reliance on the registration situate the damages award within a reasonable range of damages. The court *475therefore did not abuse its discretion in awarding damages for all 2006 studies.
D. Evidentiary Ruling
The court excluded on relevance grounds a handwritten note by Albaugh’s president, stating of Drexel out of pocket + 50% forward,” finding that no witness could establish when or why the president wrote the note. The note therefore had no tendency to make any relevant fact more probable.
Drexel protests this ruling but fails to show that the trial court abused its discretion in excluding the note and that the court’s error affected Drexel’s substantial rights. See United States v. Al-Din, 631 Fed.Appx. 313, 323 (6th Cir.2015) (citations omitted). The note duplicates evidence already in the record and sheds no light on the contract’s meaning, rendering its exclusion irrelevant.
III.
Neither party convinces us that the court reversibly erred. We accordingly AFFIRM.
. Syngenta is the corporate successor to Novartis. The parties agree that the contract covers Drexel’s payments to Syngenta.
. The arbitrators credited Drexel’s $1.5 million September 2006 payment toward the Appendix B studies, leaving Drexel only $131,340 in unpaid Appendix B costs.
. Although an early draft provided that Al-baugh’s data-compensation obligation would survive until December 31, 2005, whether or not the parties terminated, the final draft omitted this provision, Albaugh claims that the deletion proves that Drexel knew how to (but did not) provide for post-termination payments. This argument sidesteps our inquiry: Is Albaugh liable for costs incurred before termination, but not paid until after termination?