Eni and Transocean are both companies in the oil-drilling business. They formed a contract about drilling for oil. Their relationship soured. Both sued for breaches of that contract. After a bench trial, Eni suffered a resounding loss on all issues: The district court rejected its claims surrounding Transocean's maintenance of its equipment, found that Eni had wrongly repudiated the contract, and awarded damages to Transocean. Eni asks us to undo all of this. Save for the repudiation ruling, we find Eni's arguments meritorious. The district court's judgment is vacated in part and affirmed in part.
I.
In 2008, when oil prices were high, Eni-a multinational oil company-sought to conduct new exploratory drilling in the Gulf of Mexico. It circulated a request for a drilling vessel. Transocean-the largest offshore driller in the Gulf-responded. The parties soon entered into a contract. Transocean agreed to provide Deepwater Pathfinder to Eni for its drilling operations and to operate it on Eni's behalf. In exchange, Eni agreed to pay for the right to use the rig for five years.
Before commencing drilling operations in 2010, the Pathfinder went into a shipyard for maintenance. During this time, Transocean installed a previously used blowout preventer1 salvaged from the sea floor. Though old and initially bent, Transocean successfully refurbished it, and eventually it was certified for use. By February 2011, the Pathfinder 's maintenance was complete, and it set out to drill Eni's first well.
*934Almost immediately thereafter, problems began to bubble up on the Pathfinder . Most notably, the blowout preventer kept malfunctioning-requiring Transocean to suspend drilling on numerous occasions. These problems persisted throughout the contract's duration. From 2011 to 2014, maintenance and repairs resulted in the Pathfinder being inoperable for 19% of the time. Nonetheless, during that same time, the Pathfinder successfully operated on numerous wells for Eni, including one in excess of 10,000 feet underwater.
In early September 2014, the Pathfinder 's drawworks malfunctioned, but were soon fixed. Also in September, a loop current damaged one of the two MUX cables that connect the blowout preventer to its redundant yellow and blue control pods. It is against federal regulation to drill wells unless both control pods are fully operational. See 30 C.F.R. § 250.442(b). Thankfully, when the cable was damaged, the Pathfinder was not drilling. Eni had long before grown tired of the Pathfinder 's equipment problems, so it seized upon these two occurrences to sever the contract. On October 13, Eni sent a letter terminating the contract immediately, and by October 25, Eni's men had fully evacuated the Pathfinder .
About a year before this point, Eni had sued Transocean arguing both breach of contract and breach of warranty. Both claims were anchored in what Eni believed was Transocean's failure to upkeep its equipment in accordance with the contract and with industry standards. Once Eni sent the October 13 letter in 2014, Transocean filed a counterclaim, arguing that the October 13 letter was not a valid termination because it did not comply with the contract's specific procedures. In Transocean's view, the letter was a repudiation and should be treated as a total breach.
After a 10-day bench trial, the district court issued findings of fact and conclusions of law. It rejected Eni's breach-of-contract and breach-of-warranty claims. It held that Transocean had not breached any contractual requirements related to the maintenance of its equipment and that the warranty claims were barred by the contract's indemnity provision. As to Transocean's counterclaim, the district court held that Eni did not follow the contract's specific termination procedures. It therefore agreed with Transocean that the October 13 letter was indeed a repudiation, not a termination. It awarded damages to Transocean on its counterclaim. Eni timely appealed, challenging all the district court's conclusions.
II.
After a bench trial, findings of fact are reviewed for clear error while conclusions of law and mixed questions of law and fact are reviewed de novo. In Re Luhr Bros., Inc. , 325 F.3d 681, 684 (5th Cir. 2003). A finding of fact is clearly erroneous if, after reviewing all the evidence, the court "is left with the definitive and firm conviction that a mistake has been committed." Flint Hills Res. LP v. Jag Energy, Inc. , 559 F.3d 373, 375 (5th Cir. 2009) (quotation omitted).
III.
In Eni's eyes, the district court got nothing right. It wrongly rejected its breach-of-contract and breach-of-warranty claims, incorrectly found it liable on Transocean's breach-of-contract claim, and used an improper methodology to calculate Transocean's damages. We will address each argument in turn, beginning with Eni's breach-of-contract claim.
*935A.
Under § 501(b)(1) of the contract, Transocean needed to "employ commercially reasonable efforts to perform all work and operations ... in conformity with the requirements of the Contract and good oilfield practice." Eni claimed that Transocean materially breached this contractual obligation by not adequately maintaining and repairing important equipment on the Pathfinder , most importantly the blowout preventer. The district court rejected this claim, finding that Transocean complied with § 501(b)(1).
Eni argues that the district court's ultimate conclusion must be set aside because the court only conducted half of the relevant analysis necessary for that conclusion. Section 501(b)(1) requires that Transocean use commercially reasonable efforts to do two things: (1) conform with the contract's requirements; and (2) conform with good oilfield practice. These are separate inquiries. The district court, however, only meaningfully addressed the first one. On the good-oilfield-practice requirement, the court simply noted that the term was undefined. It never specifically examined Transocean's practices to determine if they met that standard. Despite its failure to engage with the relevant evidence, the court still concluded that Transocean used good oilfield practice. This, Eni says, was legal error. On its account, the district court needed to lay out the many subsidiary factual findings that would have been necessary to reach the ultimate factual conclusion that Transocean used commercially reasonable efforts to comply with good oilfield practice throughout the contract's life.2
We start with the basics. After a bench trial, a district court must make factual findings. Federal Rule of Civil Procedure 52(a) makes that clear. The question before us is how detailed those factual findings need to be. Is the district court required to make subsidiary findings? Or can it announce only its ultimate factual conclusion? We long ago answered that question: Rule 52(a) compels a district court to lay out enough subsidiary findings to allow us to understand "the basis of the trial court's decision." Gulf King Shrimp Co. v. Wirtz , 407 F.2d 508, 515 (5th Cir. 1969). Put differently, "the findings ... must be sufficiently detailed to give us a clear understanding of the analytical process by which [the] ultimate findings were reached and to assure us that the trial court took care in ascertaining the facts." Golf City, Inc. v. Wilson Sporting Goods, Co. , 555 F.2d 426, 433 (5th Cir. 1977).3 When the district court fails to do this, remand for additional fact finding is proper. See, e.g. , Redditt v. Miss. Extended Care Ctrs., Inc. , 718 F.2d 1381, 1386-87 (5th Cir. 1983) (remanding for additional fact finding on the issue of pretext in a discrimination case).
This basis-of-decision approach encourages district courts to lay out their factual findings as skillfully as possible. See Golf City, Inc. , 555 F.2d at 433 (noting that "every effort should be made to render *936[fact finding] as adequate as it humanly can be" (quotation omitted)); Hydrospace-Challenger, Inc. v. Tracor/MAS, Inc. , 520 F.2d 1030, 1033 (5th Cir. 1975) ("Statements conclusory in nature are to be eschewed in favor of statements of the preliminary and basic facts on which the District Court relied." (quotation omitted)). The more complex the case, the more important the task of articulating detailed factual findings becomes. Chandler v. City of Dall. , 958 F.2d 85, 90 (5th Cir. 1992) (per curiam).
Regardless of the basis-of-decision approach's many benefits, an alternative rule has crept into our precedent-a rule that Transocean would like us to adopt. We have, at times, said that if "a trial judge fails to make a specific finding on a particular fact, the reviewing court may assume that the court impliedly made a finding consistent with its general holding so long as the implied finding is supported by the evidence." Becker v. Tidewater, Inc. , 586 F.3d 358, 371 n.9 (5th Cir. 2009) (quotation omitted). This implicit-finding rule seems to have originated-over the dissent of Judge Godbold-in Gilbert v. Sterrett , 509 F.2d 1389, 1393 (5th Cir. 1975). In that case, Judge Godbold astutely recognized that the majority's implicit-finding rule was inconsistent with Gulf King Shrimp Co. 's basis-of-decision approach. Id. at 1397-98 (Godbold, J., dissenting). We agree. The former transforms this court into the factfinder by supplying the subsidiary facts necessary to support the general holding; the latter cabins this court to its appellate role of reviewing the district court's factual reasoning.
In any event, as Gilbert was decided after Gulf King Shrimp Co. , the rule of orderliness mandates that the basis-of-decision approach be retained, and the implicit-finding rule jettisoned.4 See Mercado v. Lynch , 823 F.3d 276, 279 (5th Cir. 2016) ("Under our rule of orderliness, one panel of our court may not overturn another panel's decision, absent an intervening change in the law, such as by a statutory amendment, or the Supreme Court, or our en banc court." (quotation omitted)). We now vindicate Judge Godbold and set our jurisprudence back on course: Under Rule 52(a), implicit findings will not automatically be inferred to support a conclusory ultimate finding. The district court must lay out enough subsidiary findings to allow us to glean "a clear understanding of the analytical process by which [the] ultimate findings were reached and to assure us that the trial court took care in ascertaining the facts." Golf City, Inc. , 555 F.2d at 433.
With the proper test in hand, we must vacate the district court's judgment and remand the case for more factual findings. The district court's order contains two sentences in two separate sections that make the ultimate factual conclusion on the good-oilfield-practice issue. Yet these conclusory statements are unsupported by the factual analysis they follow, leaving us to speculate as to whether the district court sufficiently grappled with the facts relevant to the good-oilfield-practice issue.
The first relevant section of the order is entitled "Eni's Breach of Contract Claim." The only factual analysis in that section *937states that the parties' contract clearly contemplates equipment failures, time for repairs and replacements, and potential downtime. The court also found that Transocean initiated corrective actions within 30 days of Eni's written notices of equipment problems. For these reasons, the district court concluded that "Transocean employed commercially reasonable efforts to perform under the Contract."5 But it does not follow that simply because the contract contemplates equipment failures and time for repairs, Transocean must have followed good oilfield practice. To make that determination, the court would have needed to ask why the equipment failed and whether the repairs were done properly. It did not do so. Neither did it grapple with the extensive expert testimony concerning whether the Pathfinder 's non-productive time is a good indication of whether Transocean followed good oilfield practice. And the fact that Transocean responded to Eni's written notices is only one small piece of the good-oilfield-practice puzzle.
The district court's other conclusory statement is equally unsupported by the discussion it follows. In the section entitled "Eni's Sole Remedy is Termination," the district court held that the sole-termination provisions are enforceable, noted that the parties are sophisticated, and then simply proclaimed that "Eni did not prove that ... Transocean failed to 'employ commercially reasonable efforts to perform all work and operations [ ] in conformity with the requirements of the Contract and good oilfield practice.' " Again, there is no indication that the district court considered the relevant evidence necessary to make this conclusion.
Of course, had the district court elsewhere in the order made factual findings related to this issue, those could buttress the conclusory statements. Alas, no other factual findings can be found. Not only does the district court's order not discuss any other relevant facts; it omits a critical time period in its entirety. The findings of fact do not discuss anything from the day the Pathfinder left the shipyard in 2011 until the blowout preventer's annulars failed testing in early 2013, even though many problems that Eni complains of occurred during that missing time period.
It is the district court-which heard all the evidence and is most familiar with the case-that should make the subsidiary factual findings in the first instance. See Redditt , 718 F.2d at 1386 ("It is not the function of this Court to make credibility choices and findings of fact."). But for the reasons above, it is impossible for us to ascertain whether the district court performed this important function. We would simply be guessing as to the factual basis for the district court's conclusion. See Chandler , 958 F.2d at 89 (noting that the court "cannot review bare conclusions" and that the court "cannot be left to second-guess the factual basis for the district court's conclusion" (quotation omitted)).
Accordingly, we vacate the district court's judgment as to Eni's breach-of-contract claim for failing to comply with Rule 52(a) and remand the claim to the district court to make additional factual findings on the good-oilfield-practice issue. See Redditt , 718 F.2d at 1387. Importantly, we do not hold that the district court's ultimate conclusion that Transocean used good oilfield practice was clearly erroneous.
*938The district court is free to make that conclusion again (or not) on remand. We hold only that the district court must give us a clear indication of the basis of its decision on this issue.6
B.
The district court held that Eni's breach-of-warranty claims were barred as a matter of law by § 910(a) of the contract. That provision states that the parties shall "indemnify, hold harmless and defend" each other "from and against any and all claims, demands, causes of action, damages, judgments and awards of any kind or character, without limit and without regard to the cause or causes thereof, including but not limited to claims, demands, and causes of action arising out of ... breach of representation or warranty (express or implied) ... or any other theory of legal liability."
Eni argues that § 910(a) is not the general indemnification clause the district court made it out to be-it is a definitional provision. We agree. The section clearly states that it is defining the term "be responsible for and hold harmless and indemnify" found in §§ 901-09. It then defines that term using the above language. So the district court needed to turn to §§ 901-09 to see if any of those provisions created a relevant indemnification because § 910(a), by itself, could not defeat Eni's claims.
Transocean does not quarrel with Eni's interpretation of § 910(a). Instead, it directs our attention to § 909, which provides that the parties will indemnify each other for "special, indirect, or consequential damages resulting from or arising out of" the contract. This provision, Transocean claims, encompasses the damages that Eni seeks on its warranty claims.
Transocean's argument is unpersuasive. Section 909 is a limitation on the type of damages allowed. It says nothing about what type of claims can be brought. While § 909 could limit the type of damages Eni would receive if it successfully proves its breach-of-warranty claims, it does not stop Eni from bringing breach-of-warranty claims in the first place. Transocean points to no other contractual provision (and we could not find one) that would preclude as a matter of law Eni's breach-of-warranty claims.
Thus, we vacate the district court's warranty holding and remand for a new trial on this issue to determine whether Transocean made any expressed or implied warranties and whether Transocean breached those warranties.
C.
We now turn to Transocean's breach-of-contract counterclaim, which centers around Eni's attempt to get out of the contract.
On October 13, 2014, Eni sent a letter to Transocean purporting to terminate the contract effective on that date. The letter stated that the termination was based on §§ 203(c)(1) and 1305(c). Those provisions, respectively, allow Eni to terminate the contract if (1) Transocean failed to "initiate the correction of any material non-conformity *939with [the] Contract within thirty (30) days of having received written notice thereof from [Eni]," or (2) Eni had a good-faith belief that Transocean had taken any action violating a relevant law or regulation.
The letter alleged that the failure of the drawworks constituted a material non-conformity within the meaning of § 203(c)(1).7 As for § 1305(c), it claimed that the Pathfinder had violated 30 C.F.R. § 250.442(b), which mandates that a blowout preventer must have two operable control pods attached to it when drilling. Eni claimed that this requirement had been violated when the MUX cable connecting one control pod to the blowout preventer was damaged.
The district court held that the letter did not establish proper termination under either provision. Because of this, it treated the letter as a repudiation that materially breached the contract since it unequivocally demonstrated a refusal to perform. On appeal, Eni argues that it properly terminated the contract under both provisions.8
i.
The district court held that Eni could not terminate under § 203(c)(1) based on the drawworks' failure because Transocean had initiated corrective action, completed repairs, and successfully used the drawworks by October 10, 2014-three days before Eni sent its letter.
Eni now argues that while the letter clearly singles out the malfunctioning drawworks as the reason for termination, it was really terminating for all the past problems with the blowout preventer. On its account, to "initiate a correction" means that Transocean was required to successfully fix the blowout preventer. Eni points to the blowout preventer's repeated failures as evidence that Transocean never did this. There are two problems with Eni's position.
The first one being that the letter says nothing about the blowout preventer. Indeed, Eni's own Vice President of Operations testified that the reason for the termination was the failure of the drawworks-not the past problems with the blowout preventer. It was perfectly reasonable for the district court to read the letter for what it said.
The second one being that Eni misunderstands what it means to "initiate a correction." Nothing in § 203(c)(1)'s text requires Transocean to fix a problem with finality on the first go; nothing stops it from using more convenient, short-term fixes; and nothing gives any indication that a repair's effectiveness is the measure for when termination is proper.9 Eni's attempt to redefine "to initiate a correction" in a way that would cover the above circumstances is irreconcilable with ordinary English. Section 203(c)(1) is clear: It allows for termination only if Transocean took no action to fix the problem-or in the contractual *940words, "failed to initiate a correction." The end.
It appears that even Eni understands this. It sought to modify the contract in 2014 to allow for termination if Transocean failed to "diligently pursue [a correction] to completion." This revision would not have been needed if Eni really thought that § 203(c)(1) already required such diligence.
In the end, Eni admits that after receiving each written notice about the blowout preventer, Transocean initiated a repair job. That fact-even in a counterfactual world where termination was based on the blowout preventer's problems-is enough to defeat Eni's argument under § 203(c)(1). Moreover, Transocean's repairs on the blowout preventer were not as shoddy has Eni makes them seem. Before each drill, the blowout preventer was certified by third-party experts hired by Eni and found to be fully functional.
ii.
Section 1305(c) allows for termination when Eni has a good-faith belief that Transocean materially violated a relevant law or regulation, provided certain procedures are observed: (1) Eni must advise Transocean in writing of the violation; (2) the parties must meet within 10 days; (3) Transocean must fail to provide Eni with evidence within 30 days that reasonably demonstrates that it did not run afoul of the law; (4) Eni must still believe in good faith that Transocean did violate the law; and (5) Eni must send a written notice of termination.
The district court held that Eni did not correctly terminate under § 1305(c) for three reasons: (1) Eni did not meet § 1305(c)'s notice requirements; (2) Transocean provided Eni with evidence reasonably demonstrating that it did not violate 30 C.F.R. § 250.442(b) ; and (3) Transocean did not actually violate 30 C.F.R. § 250.442(b). We find it unnecessary to address the last two points; the first point is sufficient to uphold the district court's decision.
Section 1305(c) requires two written documents-an initial notice advising Transocean of a violation and a final termination notice. But Eni only sent one written document, namely the October 13 letter. That letter is a final termination notice. It states that termination was effective immediately. No initial violation notice was sent. Thus, Eni did not comply with § 1305(c)'s dual-writing requirement.
Despite the letter's plain meaning, Eni now dubs the letter as a notice of "its intention to terminate." But even if it could be classified as the initial notice, termination would still be improper under § 1305(c), because then Eni would not have sent a final written notice of termination after the 30-day review period.
Eni's response to this fact is simple. It claims that it never had to send a second letter. It insists that on November 12-30 days after the first notice was sent-it had a unilateral right to terminate the contract and that the court should treat the initial "notice of termination" also as a premature final termination that became operative on November 12. This argument contorts the dual-writing requirement into a single-writing requirement. The letter cannot operate as an initial notice and a final notice at the same time. The parties bargained for a dual-writing requirement, and Eni cannot now wiggle out of it. Eni did not properly terminate under § 1305(c).
iii.
We affirm the district court's judgment on Transocean's breach-of-contract claim because Eni did not properly terminate the contract under § 203(c)(1) or § 1305(c). Therefore, the October 13 letter *941was rightly treated as a repudiation that materially breached the contract.
D.
Lastly, we turn to damages. The district court awarded $160,526,322.10 to Transocean on its breach-of-contract claim. "A district court's damages award is a finding of fact, which this court reviews for clear error. The conclusions of law underlying the award are reviewed de novo ." Jauch v. Nautical Servs., Inc. , 470 F.3d 207, 213 (5th Cir. 2006) (per curiam). Eni argues that the award must be vacated for two reasons-one factual; one legal.
First, Eni argues that the district court failed to meaningfully assess whether Transocean was ready, willing, and able to perform under the contract-a finding that is necessary before damages can be awarded after a material breach. 15 WILLISTON ON CONTRACTS § 47:1 (4th ed.). While admitting that the court did make an explicit finding on this issue, Eni complains the finding is conclusory, with no meaningful reasoning or explanation to back it up.
This argument must be rejected. The district court did, in fact, make other factual findings supporting its conclusion. As the court noted, just before Eni sent the termination letter, Eni and Transocean attempted to negotiate a new deal for the Pathfinder for four more years of drilling in Ghana. That deal fell through because the Ghanaian Government did not give them permission. And immediately after Eni sent the October 13 letter, Transocean began to remarket the Pathfinder . These factual findings, supported by the record, demonstrate that Transocean was ready to continue drilling operations-or at the very least, would not have totally failed to perform. See RESTATEMENT (SECOND) OF CONTRACTS § 254 cmt. a (stating that a breaching party's "duty to pay damages is discharged if it subsequently appears that there would have been a total failure of performance by the injured party"). Thus, the district court's factual conclusion that Transocean was ready, willing, and able to perform under the contract was not clearly erroneous.
Next, Eni argues that the district court committed legal error in calculating the expectation damages. Expectation damages attempt to give the nonbreaching party the benefit of the bargain by placing him "in as good a position as he would have been in had the contract been performed." Id. § 344 cmt. a; see also Hoffman v. L & M Arts , 838 F.3d 568, 584 (5th Cir. 2016).
Had Eni not breached the contract, Transocean would have been entitled to monthly payments from Eni. The amount of each payment would depend on what the Pathfinder did during each month. For example, if the Pathfinder was drilling, the full Operating Rate would apply; if it was being repaired, the Repair Rate would kick in, and if Eni failed to give any instructions to Transocean, the Standby Rate would take effect. The district court selected the Standby Rate as the applicable rate for the remainder of the contract. It did so because after Eni repudiated the contract and evacuated the Pathfinder on October 25, 2014, Eni never issued any further instructions.
We cannot approve of this analysis. That is because it looks to what Eni actually did after termination, when the operative question is what Eni would have done in a non-breach world. See 24 WILLISTON ON CONTRACTS § 64.3 (4th ed.) (noting that expectation damages are "designed to secure for that party the benefit of the bargain that he or she made"). Had Eni not repudiated the contract, Eni would have likely continued to issue instructions to Transocean. The district court should have attempted to determine, in the hypothetical non-breach world, how many days *942the Pathfinder would have spent at each applicable rate. Possibly relevant to this inquiry is the Pathfinder 's past performance, any testimony on how the Pathfinder 's machinery would have operated going forward, and any evidence suggesting what Eni's plans for the Pathfinder would have been. What is clear, though, is applying the Standby Rate simply because Eni never issued any instructions after repudiation missed the mark.
Accordingly, we vacate the damages award and remand with instructions to recalculate the damages using the correct methodology.
* * *
The district court's judgment is VACATED in part and AFFIRMED in part. This case is REMANDED to the district court for proceedings consistent with this opinion.