Title Source v. House Canary

OPINION From the 73rd Judicial District Court, Bexar County, Texas Trial Court No. 2016CI06300 Honorable David A. Canales, Judge Presiding Opinion by: Beth Watkins, Justice Sitting: Sandee Bryan Marion, Chief Justice Patricia O. Alvarez, Justice Beth Watkins, Justice Delivered and Filed: August 26, 2020 AFFIRMED IN PART; REVERSED AND REMANDED IN PART On June 18, 2020, appellant filed a motion for rehearing of our June 3, 2020 opinion and judgment in this case. Appellee filed a response to appellant’s motion on July 13, 2020. Having considered appellant’s motion for rehearing and appellee’s response, we grant appellant’s motion, withdraw our June 3, 2020 opinion and judgment, and substitute this opinion and judgment in their stead. This is an appeal from a final judgment following a jury verdict. Appellant Title Source, Inc. (“TSI”) sued appellee HouseCanary, Inc. f/k/a Canary Analytics, Inc. for breach of contract and fraud. HouseCanary filed counterclaims for breach of contract, fraud, misappropriation of trade secrets, unjust enrichment, quantum meruit, and intentional interference with an existing contract. The jury rejected TSI’s claims, found in favor of HouseCanary on its breach of contract, fraud, and misappropriation claims, and awarded HouseCanary both compensatory and punitive damages. HouseCanary elected to recover on its fraud and misappropriation claims, and the trial court signed a final judgment that ordered TSI to pay $235,400,000 in compensatory damages, $470,800,000 in punitive damages, $28,989,154 in prejudgment interest, and $4,528,711.79 in attorney’s fees. We affirm the trial court’s judgment in part, reverse it in part, and remand this cause for further proceedings consistent with this opinion. BACKGROUND TSI provides title insurance, property valuations, and settlement services as part of the Quicken Loans family of companies. One of the services TSI offers is appraisals of residential properties that are being sold or refinanced. TSI’s appraisals are performed by a few hundred TSI employees it refers to as staff appraisers, as well as approximately 20,000 independent contractors it calls panel appraisers. HouseCanary is a real estate analytics company that was founded in October of 2013. One of HouseCanary’s assets is a data dictionary, a list of attributes that can be used to determine a property’s value. HouseCanary’s co-founder and Chief of Research, Chris Stroud, spent two years building its data dictionary from at least ten sources, including a HouseCanary vendor now known as Black Knight. Although the first thirty-six pages of the data dictionary bear Black Knight’s logo, HouseCanary is listed as a source for the rest of the information in the 200-page data dictionary. HouseCanary considers its data dictionary a trade secret. Around the same time, Stroud also developed a home price index (“HPI”) model based on public and private property level data from 20,000 zip codes around the United States. He testified that no other real estate analytics provider has models built on that many zip codes, and that “the biggest value [of this data] is speeding up the development of models that require access to [it].” TSI and HouseCanary first crossed paths in late 2013, and on December 6, 2013, the two companies entered into a non-disclosure agreement (“the NDA”) so they could explore a contractual relationship without jeopardizing their intellectual property. In the NDA, both parties agreed not to “disassemble,” “decompile,” or “otherwise attempt to reverse engineer” the other party’s confidential information. They also agreed they would not “develop, manufacture, produce, and/or distribute any software or business system derived from or which otherwise uses any of the [other party's] Confidential Information.” In 2014, TSI asked HouseCanary to build an iPad application (“the app”) that TSI’s appraisers could use to perform appraisals more efficiently. Making the appraisal process more efficient would financially benefit TSI by allowing it to assign more appraisals to its salaried staff appraisers instead of its independent contractor panel appraisers. HouseCanary agreed to build the app, and the parties memorialized that agreement in a January 29, 2015 Master Software License Agreement (“the licensing agreement”). The licensing agreement incorporated the terms of the NDA by reference and further provided that TSI “may not decompile, disassemble, reverse translate, reverse engineer or otherwise attempt to discover or directly access the source code of [the app] or any component or portion thereof.” It required TSI to make the app available to both its staff appraisers and its panel appraisers and called for HouseCanary to be paid a per-use licensing fee for each appraisal completed with the app. After the parties signed the licensing agreement, HouseCanary began developing both the app and its first automated valuation models (“AVMs”). An AVM is a computer program that analyzes data to estimate a home’s value. HouseCanary’s first AVM was a “regression” model, which it “built into [its] appraisal software.” It then built the Cascade AVM, which combined the HPI Stroud previously developed with two different AVMs HouseCanary licensed from Black Knight. Stroud testified this combination maximized both coverage and accuracy, yielding “a better model than any of the inputs.” After HouseCanary completed the Cascade AVM, TSI compared its estimates with actual appraised values TSI received in the past. TSI’s testing showed that the Cascade AVM’s estimates were more accurate and had a higher “hit rate” than the service TSI was using at that time. Finally, HouseCanary built the HouseCanary AVM, which used a machine-based learning model that yielded better results than the Cascade AVM. HouseCanary considers its AVMs a trade secret. At some point during the parties’ relationship, HouseCanary began developing a model for a similarity score. A similarity score is created by an algorithm that compares two properties and rates their similarity on a scale of 1 to 100. HouseCanary did not pioneer the concept of a similarity score, and mathematical formulas for similarity scores are available on the internet. However, Stroud testified that HouseCanary did not use any publicly available formulas to develop its similarity score and that the specific data inputs it used “[came] from our research.” HouseCanary considers its similarity score a trade secret. In addition to the similarity score, HouseCanary began developing a prototype complexity score it showed to both TSI and another company that is not a party to this litigation. A complexity score is an estimate of how difficult an appraisal will be to complete and, by extension, how skilled an appraiser must be to complete it properly. Although Stroud testified HouseCanary “shelve[d]” full production of its complexity score in March of 2016, he also told the jury HouseCanary delivered a “ property score” to TSI, which some witnesses testified was the same thing as a complexity score. HouseCanary considers the work it did on its complexity score model a trade secret. HouseCanary also developed a product called a Value Report, which it described as “an overall report and tool to be able to value a property and have comparables around that property in a local market.” Each Value Report contains up to 1,000 lines of information, including similarity scores for 10 to 30 comparable properties per report. While HouseCanary does not identify the Value Report itself as a trade secret, it considers the data compilation underlying its Value Reports and other products to be a trade secret. In April of 2015, HouseCanary gave a TSI staff appraiser, Kenneth Bellew, login credentials for an early version of the app so he could test it on a real-world appraisal in Los Angeles. Although Bellew was not able to use the app to complete his appraisal, he was able to use the login information HouseCanary gave him to download the app onto his iPad. He also offered HouseCanary suggestions on how to improve the app. TSI repeatedly assured HouseCanary it had no intention of developing its own valuation models. Nevertheless, a TSI modeler named Tianqi “Ryan” Yang worked on a TSI AVM—which TSI referred to as an HVE[1]—both before and after the parties signed the licensing agreement. In April of 2015, after Yang had already begun working on the HVE, HouseCanary sent TSI its data dictionary. On August 12, 2015, Yang emailed one of his supervisors, Bryan Wang, to tell him that “[b]rainstorming data we need turns out a lot harder than I thought.” Wang responded, “Can you check the HC data dictionary to make a list?” Less than two hours later, Yang emailed Wang an Excel spreadsheet that stated it was “based on: a) HC’s data dictionary [and] b) Public source from ACS (American Community Survey) and AHS (American Housing Survey).” The document Yang sent to Wang indicates, “Note: this is from HC’s data dictionary” at the top of the first page. On September 4, 2015, a TSI intern sent an email with the subject line “some notes on the implementation of the HVE model, version 1.0″ to Yang and Wang. That email noted: We need to make a schedule to make sure that any data that is used to train the model that becomes available is added to our database. This might not be important if we use data from House Canary, since they already do this. We need to make sure that data that we keep our databases in sync with House Canary’s data [sic]. The email also explained, “[W]e want to design and deploy this model with an eye towards making it into a product, not just for internal use.” On December 14, 2015, Wang sent an email to several TSI employees with the subject line “The usage of the HC data (Modeling).” That email discussed “the following potential projects based on the HC data,” including to “Build our own products . . . After we receive significant enough data, we can develop our own HVM,[2] Similarity Score Model and Complexity Score Model Let’s think big and wide on how to maximize the value of the HC data to our business.” This is consistent with an earlier email Wang sent his team, which stated he was “more interested in knowing what data” HouseCanary had than he was in the app. While TSI exchanged these emails internally, HouseCanary became wary of the requests it was receiving from TSI for the analytics underlying its products. TSI asked HouseCanary “numerous times in pretty forceful ways to provide more detail” about the analytics behind its similarity score. HouseCanary’s CEO, Jeremy Sicklick, testified that “it seemed odd to us how much detail they really wanted.” Similarly, Stroud testified that “the questions we were getting seemed a little like they were prodding a bit too much, asking for more specifics than I would ask a vendor in a similar situation. this was just getting a bit too into the weeds of specifically what goes into the models and how they work.” Although TSI explicitly assured HouseCanary it was not in the business of building its own valuation models, former TSI employee Charles Watson testified he was aware of TSI discussions about creating models based on HouseCanary data. Watson identified Yang as the TSI employee who “wanted to make something better than what we were currently using.” In October of 2015, TSI executives and employees participated in multiple conference calls and meetings about the data TSI was receiving from HouseCanary. The notes of those calls and meetings indicate that TSI “want[ed] to make sure we’re getting the underlying data rather than just static pdfs/reports.” The notes of an October 13, 2015 conference call state: The subscription agreement/contract is currently in front of Jeff E[3] . . . Two remaining questions that need to be answered before signing this contract . . . are: We don’t just want the data in product form, we want product offering and underlying data behind the product (we don’t anticipate this to be a huge problem) We have to be able to figure out the right language in case we get cold feet while also figuring out what time & talk looks like when we do want to do bigger things with the data (client servicing, marketing, etc.) Approximately one month after that call, on November 11, 2015, TSI and HouseCanary signed an Amended Master Software License Agreement (“Amendment One”). Amendment One specified that by November 1, 2015—i.e., ten days before the parties signed the contract—HouseCanary would, inter alia, deploy the app in the field and have a system in place to allow specified information to flow between HouseCanary and TSI. That information included a “Property score for Appraisal assignment logic and & [sic] customary pricing.” Amendment One further provided that by January 1, 2016, HouseCanary would have a system in place to provide Value Reports to TSI for multiple uses. It also required HouseCanary to maintain a certain “hit rate”—which was defined as “the percentage of those residential properties submitted to [HouseCanary] by [TSI] for which [HouseCanary] can deliver a valuation using its customary practices”—and allowed TSI to terminate the contract if HouseCanary did not maintain the specified hit rate. Finally, Amendment One granted HouseCanary a license to use certain historical and ongoing appraisal data owned by TSI. Because HouseCanary believed this data was worth tens of millions of dollars per year, it agreed to accept a license to use that information plus a flat- rate price for TSI’s use of the app and HouseCanary’s other products instead of the potentially higher per-use price it would have received under the licensing agreement. Like the NDA and the licensing agreement, Amendment One restricted TSI’s use of HouseCanary’s confidential information. It specifically prohibited TSI from using HouseCanary’s data to create a database or derivative products unless the parties agreed to that use in writing. After the parties signed Amendment One, however, TSI arranged to send the data it collected from HouseCanary into both a “Data Warehouse” and a “Model Database.” Yang testified he believed TSI stored HouseCanary’s data on the “TS2dev1″ server, which he identified as a development/modeling server. On December 9, 2015, a TSI team leader sent an email to other TSI employees that joked, “Maybe we should call [the HouseCanary project] the Birdcage since we are capturing the data they provide.” Meanwhile, the app was still in development and was being tested for functionality. In January of 2016, HouseCanary hired a new director of appraiser product, Mike Poindexter, who testified HouseCanary “had a working [app] product” that “produced all the outputs it needed to” and “was a complete product” when he joined the company. Poindexter told the jury he “play[ed] with” the app before he agreed to work for HouseCanary and he would not have taken the job if the app did not work. Between January of 2016 and May of 2016, TSI downloaded 150,000 Value Reports, representing millions of data points. On March 8, 2016, TSI employee James Carson emailed TSI employees and executives to inform them, “[Yang] continues making progress on developing our complexity model using House Canary data[.]” On March 14, 2016, Carson sent an email announcing: As 2,000 or so House Canary files roll into our system each day, [Yang] will work on a data storage solution, as well as continue training the property Complexity model this week. He’ll also be validating the data to make sure we’re receiving a complete dataset. On March 31, 2016, Yang sent an email with an Excel spreadsheet called “Appraisal Complexity Model” attached to it. That document listed HouseCanary as the source for ten different types of information, including “Complexity Score.” The parties’ relationship began to deteriorate in early 2016. TSI believed HouseCanary had not satisfied its obligations under the parties’ contracts, while HouseCanary believed it had delivered all the required products and any usability issues with those products were caused by problems outside its control. On March 31, 2016, TSI asked HouseCanary to sign a second amendment to the licensing agreement, which would have required HouseCanary to send all of its underlying formulas and analytics to TSI. It also would have removed the restrictions on TSI’s ability to reverse engineer HouseCanary’s data and create derivative products. HouseCanary refused to sign the proposed amendment. At that time, TSI was still regularly requesting Value Reports on individual addresses. When a user requests a Value Report, HouseCanary’s system logs the property’s address, the requester’s identity, and whether the request was successful or unsuccessful. After HouseCanary refused to sign the proposed amendment, the number of unsuccessful requests from TSI escalated. Some of the “addresses” TSI sent to HouseCanary during this time included “99999 Anywhere Street Fricker,” “Billy’s Nerf Gun,” “the Bun’s Hun,” “Nick is awesome,” and “Wiping a Vendor Wipes the fee.” After reviewing thousands of unsuccessful requests, Sicklick concluded that someone at TSI “was just writing up addresses, random and fake addresses.” Each of those fake addresses resulted in an error that negatively affected HouseCanary’s hit rate. On April 12, 2016, TSI sued HouseCanary in Bexar County District Court for breach of contract and fraud. On April 19, 2016, TSI terminated the parties’ contract. HouseCanary answered TSI’s lawsuit and asserted counterclaims for breach of contract, fraud, unjust enrichment, and quantum meruit. HouseCanary later added a claim for misappropriation of trade secrets under the Texas Uniform Trade Secrets Act (“TUTSA”). After TSI terminated the parties’ contract, TSI assigned Claude Wang to develop an in- house AVM, which he named MyAVM. Because Claude Wang had no prior experience working with AVMs, he testified he “went to Google Scholar and searched the publications about AVM.” He explained that he chose what data to use, that he was in charge of developing the model, and that he did so by himself. He also testified MyAVM relied solely on TSI appraisal data and publicly available macroeconomic and property data and did not include any of HouseCanary’s data. Yang—who had previously worked with HouseCanary’s data—collected the publicly available demographic data for MyAVM. Although Claude Wang testified that was the full extent of Yang’s contribution to the project, internal TSI documents showed that Yang also worked on the appraisal data that went into MyAVM and on its user interface. Additionally, when Claude Wang presented MyAVM at an August 2016 Quicken Loans conference, he told the audience, “[Yang] has developed a similarity score.” A PowerPoint presentation he used during that conference lists both Yang and Bryan Wang, who had also worked with HouseCanary’s data, as part of the team that worked on MyAVM. Claude Wang finished MyAVM two months after he started the project, and he testified that “[i]t took [him] just a few minutes to create the data dictionary” he used. In contrast, Stroud testified that it took him two years to create HouseCanary’s data dictionary, and that it would have taken “a year or more” to build HouseCanary’s first AVM without HouseCanary’s HPI. The jury found in HouseCanary’s favor on its misappropriation of trade secrets, fraud, and breach of contract claims, rejected all of TSI’s affirmative claims, and found that HouseCanary was entitled to actual and punitive damages. HouseCanary elected to recover on its misappropriation and fraud claims, and the trial court signed a judgment consistent with the jury’s verdict on those claims. TSI filed a motion for new trial, which the trial court denied. This appeal followed. ANALYSIS TSI presents eight issues challenging the trial court’s judgment. We will consider only those issues that are necessary for the final disposition of this appeal. TEX. R. APP. P. 47.1. HouseCanary’s TUTSA Claim TSI first contends the evidence is legally and factually insufficient to support the jury’s finding that it misappropriated HouseCanary’s trade secrets. It also argues the liability and damages questions the jury answered on HouseCanary’s TUTSA claim mixed valid and invalid theories of recovery. HouseCanary responds that the evidence supports the jury’s findings and the court’s charge was properly worded. Standard of Review When an appellant challenges the legal sufficiency of an adverse finding on an issue on which it did not have the burden of proof, it must show there is no evidence to support the finding. Graham Cent. Station, Inc. v. Pena, 442 S.W.3d 261, 263 (Tex. 2014). “In reviewing a finding for legal sufficiency, we consider the evidence in the light most favorable to the finding, crediting favorable evidence if a reasonable factfinder could and disregarding contrary evidence unless a reasonable factfinder could not.” John Deloach Enters., Inc. v. Telhio Credit Union, Inc., 582 S.W.3d 590, 595 (Tex. App.—San Antonio 2019, no pet.). Evidence is legally sufficient to support a jury’s verdict if it would allow reasonable and fair-minded jurors to reach the challenged finding. See, e.g., City of Keller v. Wilson, 168 S.W.3d 802, 827 (Tex. 2005). In a factual sufficiency challenge to an adverse finding on an issue on which the appellant did not have the burden of proof, the appellant must show that there is insufficient evidence to support the finding. See Johnson v. Waters at Elm Creek, L.L.C., 416 S.W.3d 42, 47 (Tex. App.— San Antonio 2013, pet. denied). This court must consider and weigh all the evidence, including the evidence that is contrary to the verdict. See De los Santos v. Comm’n for Lawyer Discipline, 547 S.W.3d 640, 645 (Tex. App.—San Antonio 2017, pet. denied). Evidence is factually insufficient to support a jury’s verdict only if the challenged finding is so contrary to the overwhelming weight of the evidence as to be clearly wrong and unjust. Id. This court reviews the trial court’s submission of jury questions, instructions, and definitions for abuse of discretion. Bexar Cty. Appraisal Dist. v. Abdo, 399 S.W.3d 248, 257–58 (Tex. App.—San Antonio 2012, no pet.). The court may not reverse a judgment for charge error unless the error probably caused the rendition of an improper judgment or probably prevented the appellant from properly presenting its case on appeal. TEX. R. APP. P. 44.1; Ramos v. City of Laredo, 547 S.W.3d 651, 654 (Tex. App.—San Antonio 2018, no pet.). Applicable Law To bring a successful TUTSA claim, the claimant must first show that it owns a trade secret in the information it seeks to protect. TEX. CIV. PRAC. & REM. CODE ANN. § 134A.002(3-a), (6); Morgan v. Clements Fluids S. Tex., Inc., 589 S.W.3d 177, 186–87 (Tex. App.—Tyler 2018, no pet.). Under both the current statutory definition and the jury charge in this case, “trade secret” is defined as: all forms and types of information, including business, scientific, technical, economic, or engineering information, and any formula, design, prototype, pattern, plan, compilation, program device, program, code, device, method, technique, process, procedure, financial data, or list of actual or potential customers or suppliers, whether tangible or intangible and whether or how stored, compiled, or memorialized physically, electronically, graphically, photographically, or in writing if: the owner of the trade secret has taken reasonable measures under the circumstances to keep the information secret; and the information derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable through proper means by, another person who can obtain economic value from the disclosure or use of the information. TEX. CIV. PRAC. & REM. CODE § 134A.002(6). The claimant must also demonstrate that it is “the person or entity in whom or in which rightful, legal, or equitable title to, or the right to enforce rights in, the trade secret is reposed.” Id. § 134A.002(3-a). If a TUTSA claimant demonstrates that the information it seeks to protect meets the statutory definition of “trade secret,” it must then show that the defendant misappropriated the trade secret. Id. § 134A.002(3); Morrison v. Profanchik, No. 05-17-01281-CV, 2019 WL 3798182, at *5 (Tex. App.—Dallas Aug. 13, 2019, no pet.) (mem. op.). Here, the jury was asked to determine whether TSI: acquired the trade secrets and knew or had reason to know that the trade secrets were acquired by improper means; or disclosed or used the trade secrets without HouseCanary’s express or implied consent, and that TSI used improper means to acquire knowledge of the trade secrets; or disclosed or used the trade secrets without HouseCanary’s express or implied consent, and that TSI, at the time of the disclosure or use, knew or had reason to know its knowledge of the trade secrets were acquired under circumstances giving rise to a duty to maintain secrecy or limit their use. TEX. CIV. PRAC. & REM. CODE § 134A.002(3)(A), (B)(i), (B)(ii)(b). When a single broad-form question commingles valid and invalid theories, a new trial is required if the appellate court cannot determine whether the jury’s verdict is based on an invalid theory. Harris County v. Smith, 96 S.W.3d 230, 232–33 (Tex. 2002) (citing Casteel v. Crown Life Ins. Co., 22 S.W.3d 378, 388 (Tex. 2000)). Such an error is presumed harmful because it “affirmatively prevent[s] the appellant from isolating the error and presenting its case on appeal.” Id. In order to preserve this error for appellate review, the complaining party must make a timely and specific objection. Thota v. Young, 366 S.W.3d 678, 691 (Tex. 2012); see also TEX. R. CIV. P. 274; TEX. R. APP. P. 33.1. A complaint of Casteel error need not specifically reference Casteel, but the objection must be sufficient to inform the trial court that the submission includes an invalid theory of liability. See Thota, 366 S.W.3d at 691. Analysis Question 37—Ownership of Trade Secrets In its answer to Question 37, the jury found HouseCanary owned trade secrets in its AVMs, similarity score, data dictionary, data compilation, and complexity score. TSI argues the evidence is legally and factually insufficient to support this finding. It also contends the question itself fails due to Casteel error. See Casteel, 22 S.W.3d at 388. The evidence shows the parties signed three contracts, all of which restricted TSI’s use and storage of HouseCanary’s confidential and proprietary data. The jury also heard evidence that HouseCanary actively resisted TSI’s attempts to obtain more data and did not relent until TSI promised it would use that data “for definition only.” Based on this evidence, a reasonable factfinder could conclude HouseCanary took reasonable measures under the circumstances to keep the information at issue a secret. TEX. CIV. PRAC. & REM. CODE § 134A.002(6)(A); City of Keller, 168 S.W.3d at 827. While TSI has argued HouseCanary did not take reasonable measures to keep the information secret because it publicly disclosed eight specific exhibits at trial, that disclosure occurred more than a year after the alleged misappropriation in this case. Under these circumstances, we decline to hold that HouseCanary’s use of those eight exhibits at trial conclusively shows it did not take reasonable measures to protect its trade secrets during the course of its relationship with TSI. See City of Keller, 168 S.W.3d at 827. Moreover, the jury’s finding is not so contrary to the overwhelming weight of the evidence as to be clearly wrong and unjust. See De los Santos, 547 S.W.3d at 645. Additionally, the jury heard evidence about HouseCanary’s development of each trade secret, how those trade secrets built on each other to help HouseCanary develop its products, and why HouseCanary was reluctant to share the data and analytics underlying those trade secrets. It also heard evidence of difficulties TSI experienced while trying to create similar products. Finally, the jury saw multiple internal TSI documents showing TSI intended to use HouseCanary’s data to develop its own products and believed having access to the data and analytics underlying HouseCanary’s products would help it do so. This evidence supports a finding that the five trade secrets derived independent economic value from not being generally known or readily ascertainable through proper means. TEX. CIV. PRAC. & REM. CODE § 134A.002(6)(B). While the jury heard conflicting evidence on each of these elements, we cannot say that there is no evidence to support its finding that the five categories of information constituted trade secrets. See City of Keller, 168 S.W.3d at 827. Nor can we conclude that the jury’s finding is so contrary to the overwhelming weight of the evidence as to be clearly wrong and unjust. See De los Santos, 547 S.W.3d at 645. The evidence also supports a finding that HouseCanary is the entity with “rightful, legal, or equitable title to, or the right to enforce rights in” the trade secrets. TEX. CIV. PRAC. & REM. CODE § 134A.002(3-a). It is undisputed that HouseCanary alone assembled its data compilation. While TSI has argued HouseCanary’s AVMs and data dictionary are based on information that belongs to Black Knight, one of TSI’s own trial exhibits shows that HouseCanary owns all of the derivative works it created using Black Knight’s data. Additionally, Stroud testified HouseCanary’s AVMs, similarity score, and data dictionary include HouseCanary’s own research and analytics in addition to the data it obtained from Black Knight and other public and private sources. Finally, while TSI argues the evidence shows HouseCanary never developed a complexity score, the jury heard testimony that HouseCanary presented a prototype complexity score model to both TSI and to another company that is not a party to this case. The jury also saw an internal TSI document titled “Appraisal Complexity Score” that listed HouseCanary as the source for several of its attributes. Moreover, the jury heard evidence that HouseCanary delivered a property score to TSI, and one of TSI’s own witnesses testified that a property score and a complexity score are the same thing. This evidence is both legally and factually sufficient to support the jury’s finding that HouseCanary owned trade secrets in all five categories of information. Id. § 134A.002(3-a), (6); City of Keller, 168 S.W.3d at 827; De los Santos, 547 S.W.3d at 645. In addition to challenging the sufficiency of the evidence supporting the jury’s answers to Question 37, TSI contends the question itself fails due to Casteel error. See Casteel, 22 S.W.3d at 388. It argues that at trial, HouseCanary “referred to” multiple AVMs and complexity scores; that Question 37 did not give the jurors sufficient guidance “on which AVM or which complexity score they were being asked to assess”; and that “the broad-form nature of [Question 37] requires a new trial.” HouseCanary argues, however, that TSI waived this objection to Question 37. We agree with HouseCanary. While TSI made global objections to the whole of Question 37, it did not specifically identify any issues with the “complexity score” and “AVM” portions of that question. TEX. R. CIV. P. 274; TEX. R. APP. P. 33.1; see also Watts v. Watts, 396 S.W.3d 19, 23 (Tex. App.—San Antonio 2012, no pet.) (“In order to preserve error for appellate review, a party’s argument on appeal must comport with its argument in the trial court.”). Nor did it object to the broad-form nature of the question. See In re A.V., 113 S.W.3d 355, 363 (Tex. 2003) (holding Casteel complaint not preserved because objection did not “put [the] trial court on notice to submit a granulated question”); Watts, 396 S.W.3d at 23. Finally, the heart of a Casteel objection is that a broad-form question combines multiple legal theories, some of which are valid and some of which are not. See, e.g., Laredo Med. Grp. Corp. v. Mireles, 155 S.W.3d 417, 426–27 (Tex. App.— San Antonio 2004, pet. denied.). The objections TSI raised in the trial court—that “the alleged items that are claimed to be trade secrets are not clearly identified specifically for the Jury”; that the terms used in Questions 37 “are vague, nebulous, misleading, and it’s impossible to determine what they mean”; and “there’s no evidence that any of the items listed are trade secrets”—were not specific enough to put the trial court on notice that TSI believed the “AVM” and “complexity score” portions of Question 37 mixed valid and invalid legal theories. See Duradril, L.L.C. v. Dynomax Drilling Tools, Inc., 516 S.W.3d 147, 157 (Tex. App.—Houston [14th Dist.] 2017, no pet.). We overrule TSI’s Casteel challenge to Question 37. Question 38—Misappropriation of Trade Secrets Question 38 asked the jury to determine whether TSI misappropriated HouseCanary’s trade secrets. The jury was instructed that it could find misappropriation based on either a “use” theory or an “acquisition by improper means” theory. TEX. CIV. PRAC. & REM. CODE §§ 134A.002(3)(A), (B)(i), (B)(ii)(b). TSI objected to Question 38 on the basis that it “obtained the alleged trade secrets properly under the agreements between the parties” and “there is no evidence of . . . improper means.” The trial court was required to give the jury “such instructions and definitions as shall be proper to enable the jury to render a verdict.” TEX. R. CIV. P. 277; Columbia Rio Grande Healthcare, L.P. v. Hawley, 284 S.W.3d 851, 855 (Tex. 2009). A jury instruction is proper if it “(1) assists the jury; (2) accurately states the law; and (3) finds support in the pleadings and evidence.” Columbia, 284 S.W.3d at 855. Neither the statute nor the jury charge define “use,” but both define “improper means” as “theft, bribery, misrepresentation, breach or inducement of a breach of a duty to maintain secrecy, to limit use, or to prohibit discovery of a trade secret, or espionage through electronic or other means.” TEX. CIV. PRAC. & REM. CODE § 134A.002(2). HouseCanary asserts that the evidence shows TSI relied on HouseCanary’s trade secrets to assist or accelerate its own research and development, and that this evidence supports a misappropriation finding under a “use” theory. See id. § 134A.002(3)(B)(ii)(b); Bohnsack v. Varco, L.P., 668 F.3d 262, 279 (5th Cir. 2012); see also Univ. Computing Co. v. Lykes-Youngstown Corp., 504 F.2d 518, 538–39 (5th Cir. 1974). HouseCanary also contends “there is abundant evidence” that TSI acquired the trade secrets through misrepresentation and that this evidence supports a misappropriation finding under an “improper means” theory. See TEX. CIV. PRAC. & REM. CODE §§ 134A.002(3)(A), 134A.002(2). Even if we assume these assertions are true, however, the jury was also instructed that “improper means” includes bribery, espionage, and “breach or inducement of a breach of a duty to maintain secrecy, to limit use, or to prohibit discovery of a trade secret.” This instruction tracks TUTSA’s definition of “improper means” and is therefore a correct statement of law. TEX. CIV. PRAC. & REM. CODE § 134A.002(2); Columbia, 284 S.W.3d at 855. But HouseCanary conceded at oral argument that there is no evidence TSI acquired the trade secrets through bribery, and our review of the record reveals no evidence that TSI acquired the trade secrets through espionage. Because those theories are not supported by the evidence, they should have been omitted from the “improper means” definition that was submitted to the jury. TEX. R. CIV. P. 278; Regal Fin. Co., Ltd. v. Tex Star Motors, Inc., 355 S.W.3d 595, 601 (Tex. 2010) (noting that while “a jury charge submitting liability under a statute should track the statutory language as closely as possible,” the statutory language “may be slightly altered to conform the issue to the evidence presented”); see also Tex. Comm’n on Human Rights v. Morrison, 381 S.W.3d 533, 537 (Tex. 2012) (“A broad-form question cannot be used to put before the jury issues that have no basis in the law or the evidence.”) (internal quotation marks omitted). Similarly, the “breach or inducement of a breach of a duty to maintain secrecy, to limit use, or to prohibit discovery of a trade secret” instruction should have been omitted from the charge. Under the plain language of both TUTSA and the court’s charge in this case, TSI needed to acquire the trade secrets as a result of the alleged breaches to support a misappropriation finding. TEX. CIV. PRAC. & REM. CODE § 134A.002(3)(A) (“‘Misappropriation’ means: (A) acquisition of a trade secret of another by a person who knows or has reason to know that the trade secret was acquired by improper means[.]“) (emphasis added). A post-acquisition breach of a confidentiality or non- disclosure agreement—the only breaches HouseCanary presented evidence of here—”is irrelevant to the method by which [TSI] obtained access to the trade secrets in the first instance” and thus cannot support an improper means finding as a matter of law. Educ. Mgmt. Servs., LLC v. Tracey, 102 F. Supp. 3d 906, 914 (W.D. Tex. 2015); Educ. Mgmt. Servs., LLC v. Cadero, No. SA-14-CA- 587, 2014 WL 12586781, at *2 (W.D. Tex. Nov. 18, 2014). During oral argument and in its post-submission briefing, HouseCanary noted that a Casteel issue is not reversible if the reviewing court can be “reasonably certain” the jury did not base its findings on the invalid theories. See Romero v. KPH Consol., Inc., 166 S.W.3d 212, 227– 28 (Tex. 2005). Because HouseCanary did not pursue bribery or espionage theories at trial, it contends the record does not support a conclusion that the jury based its findings on those theories. However, we are reasonably certain that the jury was significantly influenced by the erroneous inclusion of the “breach or inducement of a breach of a duty to maintain secrecy, to limit use, or to prohibit discovery of a trade secret” instruction.[4] See id. at 228. HouseCanary argued throughout the seven-week trial—and continues to argue on appeal—that TSI breached a duty to limit its use of HouseCanary’s trade secrets. But there is no evidence that TSI actually acquired the trade secrets through those breaches. Instead, the evidence shows those breaches, if any, occurred after HouseCanary willingly turned over its data under the NDA, the licensing agreement, and Amendment One. As a result, those breaches do not support a misappropriation finding. See Tracey, 102 F. Supp. 3d at 914. Nevertheless, because HouseCanary so heavily emphasized the evidence it presented to the jury of TSI’s alleged post-acquisition breaches, we cannot rule out the possibility that the jury found misappropriation based on those breaches. See Morrison, 381 S.W.3d at 537. (“[W]hen a broad-form question allows a finding of liability based on an invalid theory, an appealing party does not have to prove that the jury actually relied on the invalid theory.”). As a result, the inclusion of that issue in the charge constitutes harmful error. Casteel, 22 S.W.3d at 389. In short, this charge included “multiple liability theories, several of which are not supported by legally sufficient evidence.” Mireles, 155 S.W.3d at 427. This court has previously held that “when only one subpart [of a liability question] is supported by legally sufficient evidence,” we must reverse the trial court’s judgment and remand the cause for a new trial. Id. at 426. For this reason, we must reverse the trial court’s judgment on HouseCanary’s TUTSA claim and remand that claim for a new trial. HouseCanary’s Fraud Claim TSI argues the jury’s findings on HouseCanary’s fraud claim cannot stand because that claim is: (1) preempted by TUTSA; and (2) not supported by legally or factually sufficient evidence. HouseCanary responds that the jury’s fraud findings are supported by both Texas law and the evidence presented at trial. Standard of Review Whether a common law cause of action is preempted by a statutory remedy is a question of law the court reviews de novo. See Brocail v. Detroit Tigers, Inc., 268 S.W.3d 90, 99 (Tex. App.—Houston [14th Dist.] 2008, pet. denied). Applicable Law TUTSA “displaces conflicting tort, restitutionary, and other law of this state providing civil remedies for misappropriation of a trade secret.” TEX. CIV. PRAC. & REM. CODE ANN. § 134A.007(a). It does not affect “other civil remedies that are not based upon misappropriation of a trade secret.” Id. § 134A.007(b)(2). When the gravamen of a common law claim duplicates a TUTSA claim, the common law claim is preempted. Super Starr Int’l, LLC v. Fresh Tex Produce, LLC, 531 S.W.3d 829, 843 (Tex. App.—Corpus Christi–Edinburg 2017, no pet.). This occurs if the factual basis of the common law claim, as pleaded, would not exist “without the use of alleged trade secrets.” Id. However, because TUTSA’s preemption provision applies only to conflicting common law remedies, a common law claim is not preempted by TUTSA if it addresses harm separate from the trade secret misappropriation. TEX. CIV. PRAC. & REM. CODE § 134A.007(a); see Lifesize, Inc. v. Chimene, No. 1:16-CV-1109-RP, 2017 WL 1532609, at *12–13 (W.D. Tex. Apr. 26, 2017). Analysis TSI contends HouseCanary’s fraud claim is directly premised on the misappropriation of a trade secret. In support of this contention, it notes HouseCanary argued at trial that its fraud and TUTSA claims were based on the same facts. It also argues HouseCanary’s fraud claim is “a recycled version of its ‘improper means’ theory of misappropriation under TUTSA.” HouseCanary responds its fraud claim is not preempted because the evidence shows it entered into Amendment One and “invested substantial time and resources” in justifiable reliance on: (a) TSI’s representations about paying for and rolling out the app to TSI’s staff and panel appraisers; and (b) TSI’s promise to share its valuable historical appraisal data with HouseCanary. It also argues TSI never intended to pay for or roll out the app or share its historical data as promised, and HouseCanary lost profits as a result. HouseCanary contends these allegations show it “had a valid fraud claim even if it never had trade secrets nor established misappropriation.” We conclude different aspects of the fraud allegations in HouseCanary’s live petition support both parties’ preemption arguments. HouseCanary’s petition explicitly asserts that TSI “misappropriated HouseCanary trade secrets” and fraudulently induced HouseCanary to enter into the contract so it could “obtain HouseCanary’s data and analytics, gain insight into HouseCanary’s intellectual property, and use the data, analytics, and insight for its own software models, analytics, and/or products for use by TSI and/or its affiliate Quicken Loans without fairly compensating HouseCanary.” Because the foundation of these complaints is an assertion that TSI misappropriated HouseCanary’s trade secrets, these allegations duplicate HouseCanary’s TUTSA claim and are therefore preempted. See Super Starr, 531 S.W.3d at 843. But HouseCanary’s fraud allegations also contend that TSI never intended to pay HouseCanary the agreed-upon fee for its services and that HouseCanary was induced to enter into Amendment One and relinquish its rights under the “potentially more lucrative” licensing agreement “based on the representation that TSI would provide its historical appraisal data to HouseCanary.” These allegations, standing alone, are not based on any claims that TSI misappropriated any information from HouseCanary. See TEX. CIV. PRAC. & REM. CODE 134A.007(b)(2) (TUTSA preemption does not affect claims “that are not based upon misappropriation of a trade secret”); Lifesize, 2017 WL 1532609, at *12–13. In its briefing, TSI cites federal authority applying TUTSA to situations where “potentially preempted claims are founded on” the misappropriation of both trade secrets and non-trade secret confidential information. See Embarcadero Techs., Inc. v. Redgate Software, Ltd., No. 1:17-cv- 444-RP, 2018 WL 315753, at *2–3 (W.D. Tex. Jan. 5, 2018); see also Steves & Sons, Inc. v. JELD- WEN, Inc., No. 3:16-cv-545, 2018 WL 1796293, at *11–13 (E.D. Va. April 16, 2018). Those courts concluded that “TUTSA’s preemption provision encompasses all claims based on the alleged improper taking of confidential business information.” Embarcadero, 2018 WL 315753, at *3; Steves, 2018 WL 1796293, at *13–14. The Western District of Texas also recently considered common law conversion claims that involved the improper taking of both trade secret information and the physical devices on which those trade secrets were stored. See Lifesize, 2017 WL 1532609, at *12–13. It concluded the claims based on the conversion of the trade secrets were preempted but the claims based on the conversion of the physical devices were not, and it resolved that issue by granting a pretrial motion to dismiss the common law claims “to the extent they seek recovery of the value of Plaintiff’s trade secrets.” Id. While this authority is instructive, it is not dispositive where, as here, HouseCanary obtained a jury verdict on a common law claim that alleged both the unauthorized taking of trade secret information and other wrongful acts that did not involve the trade secrets. Based on this record, we cannot definitively resolve this issue. Because the jury answered a single broad-form question on all of HouseCanary’s fraud allegations, we cannot affirm its finding without ignoring the Legislature’s directive that TUTSA wholly displaces conflicting tort law. TEX. CIV. PRAC. & REM. CODE § 134A.007(a); Super Starr, 531 S.W.3d at 843. Nor can we render a take-nothing judgment in TSI’s favor without ignoring the evidence that a reasonable factfinder could have credited in favor of the non-preempted fraud allegations. See City of Keller, 168 S.W.3d at 820–21. Under these unique circumstances, we hold that the trial court’s judgment on HouseCanary’s fraud claim must be reversed, and we remand this cause for a new trial on that claim. Cf. Morrison, 381 S.W.3d at 537 (“When a jury question contains both valid and invalid theories, [an] appellate court cannot determine whether the jury based its verdict on an improperly submitted invalid theory, and thus remand for retrial is the only option.”) (internal quotation marks omitted). HouseCanary’s Breach of Contract Claim Because HouseCanary did not elect to recover on the jury’s findings on its breach of contract claims, neither the trial court’s judgment nor our original opinion and judgment in this appeal address those claims. Accordingly, our original opinion and judgment did not remand those claims to be retried alongside HouseCanary’s TUTSA and fraud complaints. In its motion for rehearing, TSI argues that all of HouseCanary’s causes of action “relied on the same core of interrelated facts and overlapping measures of damages” and therefore cannot be separated without unfairness to the parties. It contends that to avoid potentially conflicting judgments, HouseCanary must elect on remand to either: (1) recover on the jury’s contract verdict; or (2) retry all of its claims. In response, HouseCanary argues: (1) its TUTSA and contract claims are not inseparable; (2) its fraud claim should not be remanded at all; and (3) TSI’s arguments could be more efficiently raised in a petition for review to the Texas Supreme Court. We agree with TSI. A partial remand is appropriate only if the error in a trial court’s judgment “affects part of, but not all, the matter in controversy and that part is separable without unfairness to the parties.” TEX. R. APP. P. 44.1(b); see also Jerry L. Starkey, TBDL, L.P. v. Graves, 448 S.W.3d 88, 113 (Tex. App.—Houston [14th Dist.] 2014, no pet.). We have previously held, “If the jury finds in favor of the plaintiff on multiple alternative theories of recovery, but the theory of recovery on which the trial court based its judgment is reversed on appeal, the court of appeals must remand all theories of recovery to the trial court.” Cotter & Sons, Inc. v. BJ Corp., 549 S.W.3d 715, 729 (Tex. App.—San Antonio 2017, pet. dism’d). Here, the record shows that on both its TUTSA and contract claims, HouseCanary sought “[t]he value of the HouseCanary trade secrets to [TSI]” at the time of the alleged breach or misappropriation, and the jury awarded the same amount of actual damages for each claim. Similarly, on both its contract and fraud claims, HouseCanary sought to recover “ [l]ost profits, if any, that were a natural, probable, and foreseeable consequence of” TSI’s alleged fraud or breach, and again, the jury awarded the same amount of actual damages on each claim. Finally, as TSI notes, HouseCanary argued throughout the trial that its causes of action were based on the same facts. Cf. State Dep’t of Highways & Pub. Transp. v. Cotner, 845 S.W.2d 818, 819 (Tex. 1993) (claims are not properly severable if they “involve the same facts and issues”). Because HouseCanary’s TUTSA, fraud, and contract claims rely, at least in part, on the same facts and the jury found those acts caused the same damages, it is possible that a retrial of the TUTSA and fraud claims will result in a verdict that conflicts with the first jury’s contract findings. We therefore conclude HouseCanary’s claims are not separable from each other without unfairness to the parties. See Cotter & Sons, 549 S.W.3d at 728–29; Jerry L. Starkey, 448 S.W.3d at 93. As a result, if HouseCanary wishes to retry its TUTSA and fraud claims on remand, it must also relitigate its contract claims. See TEX. R. APP. P. 44.1(b); Cotter & Sons, 549 S.W.3d at 728–29. Alternatively, it may choose to forego a new trial and recover on the jury’s contract findings that were not successfully challenged in this appeal. See Jerry L. Starkey, 448 S.W.3d at 93; see also Boyce Iron Works, Inc. v. Sw. Bell Tel. Co., 747 S.W.2d 785, 787 (Tex. 1988) (“When the jury returns favorable findings on two or more alternative theories, the prevailing party need not formally waive the alternative findings. That party may seek recovery under an alternative theory if the judgment is reversed on appeal.”). TSI’s Breach of Contract Claim TSI also challenges the jury’s rejection of its own breach of contract claim. That claim asked the jury to determine whether HouseCanary failed to comply with the NDA, the licensing agreement, and Amendment One by failing to provide a working app. HouseCanary argues that the jury’s verdict on this issue is supported by the evidence. Standard of Review When a party challenges the legal sufficiency of an adverse finding on which it had the burden of proof, it must show that the evidence establishes all vital facts in support of the issue as a matter of law. See Zuniga v. Velasquez, 274 S.W.3d 770, 773 (Tex. App.—San Antonio 2008, no pet.). This court first “examine[s] the record for evidence supporting the finding, while ignoring all evidence to the contrary. If there is no evidence to support the finding, we then examine the entire record to determine if the contrary proposition is established as a matter of law.” Id. (internal citations omitted). When a party challenges the factual sufficiency of an adverse finding on which it had the burden of proof, the court examines all of the evidence and reverses the trial court’s judgment “only if the finding is so against the great weight and preponderance of the evidence as to be manifestly wrong or unjust.” Id. Applicable Law The court’s charge instructed the jury that any failure by HouseCanary to comply with the parties’ contracts must be material. See Mustang Pipeline Co., Inc. v. Driver Pipeline Co., Inc., 134 S.W.3d 195, 199 (Tex. 2004). The jury was further instructed to consider the following factors in determining materiality: the extent to which the injured party will be deprived of the benefit which he reasonably expected; the extent to which the injured party can be adequately compensated for the part of that benefit of which he will be deprived; the extent to which the party failing to perform or to offer to perform will suffer forfeiture; the likelihood that the party failing to perform or to offer to perform will cure his failure, taking into account the circumstances including any reasonable assurances; the extent to which the behavior of the party failing to perform or to offer to perform comports with standards of good faith and fair dealing. See id. Analysis The only failure by HouseCanary that TSI complains of on appeal is a failure to deliver a working app. The evidence shows that TSI’s appraisers tested the app at least twice—once when Bellew tried the app in Los Angeles, and then later in Phoenix when several TSI staff appraisers tried to use the app to complete test appraisals. On February 22, 2016, HouseCanary sent login information for the app to a TSI employee, Mike Brocker-Querio, who responded, “This is seriously cool.” TSI’s then-vice president and chief appraiser, Jordan Petkovski, stated in a July 2015 email that he had seen the app and it was “impressive.” However, Petkovski later testified HouseCanary never delivered an app that met all of the contract’s requirements. Eisenshtadt testified that the app only allowed appraisers to use one of four required government forms, and that appraisers cannot complete their work without access to all four. Other TSI employees testified the versions of the app they saw did not include a sketch feature that was necessary to complete an appraisal. While HouseCanary’s witnesses agreed the version of the app that existed when TSI terminated the contract did not include all four forms, witnesses from both TSI and HouseCanary testified that the single form the app supported covered 80% of TSI’s market. Additionally, HouseCanary’s witnesses testified that they had not implemented the three missing forms into the app because TSI asked HouseCanary to prioritize other features. HouseCanary’s witnesses also explained that although they disabled the sketch feature in the iPad version of the app while their engineers worked to improve it, that feature could have been turned back on at any point if TSI had asked HouseCanary to do so. Poindexter, HouseCanary’s director of appraiser product, also testified that while the sketch function could not be used in the iPad version of the app, it was available to use in the web-based version of the app on any device with a web browser, including an iPad. Based on this evidence, we hold TSI did not conclusively establish that HouseCanary’s failure to comply with the parties’ contracts, if any, was material. See Mustang Pipeline, 134 S.W.3d at 199; Zuniga, 274 S.W.3d at 773. We also hold that the jury’s rejection of TSI’s breach of contract claim is not so against the great weight and preponderance of the evidence as to be manifestly wrong or unjust. See Zuniga, 274 S.W.3d at 773. We therefore affirm the trial court’s take-nothing judgment on TSI’s breach of contract claim. CONCLUSION For the foregoing reasons, we affirm the trial court’s take-nothing judgment on Title Source’s breach of contract claim. We reverse the trial court’s judgment on HouseCanary’s TUTSA and fraud claims and remand this case with instructions that HouseCanary must elect to either: (1) recover on the jury’s breach of contract findings in its favor; or (2) retry all of its claims against TSI. In light of our disposition of these issues, we need not consider the parties’ remaining arguments. TEX. R. APP. P. 47.1. Beth Watkins, Justice