LARRY A. LAWSON, Plaintiff, v. SPIRIT AEROSYSTEMS, INC., Defendant Case No. 18-1100-EFM-ADM United States District Court, D. Kansas Signed November 24, 2020 Counsel F. James Robinson, Jr., Kimberly D. Farha, Hite, Fanning &amp; Honeyman, LLP, Wichita, KS, James C. Dugan, Pro Hac Vice, Matthew S. Freimuth, Pro Hac Vice, Joseph T. Niczky, Pro Hac Vice, Ravi Chanderraj, Pro Hac Vice, Martin L. Seidel, Pro Hac Vice, Jonathan David Waisnor, Pro Hac Vice, Willkie Farr &amp; Gallagher, LLP, New York, NY, for Plaintiff. Ann Marie Arcadi, Pro Hac Vice, John M. Farrell, Pro Hac Vice, Seema Tendolkar, Pro Hac Vice, T. Gregory Jackson, Pro Hac Vice, Arcadi Jackson, LLP, Dallas, TX, Charles E. McClellan, Gary L. Ayers, Jeff P. DeGraffenreid, Foulston Siefkin LLP, Wichita, KS, for Defendant. Melgren, Eric F., United States District Judge MEMORANDUM AND ORDER *1 Before the Court is Plaintiff Larry Lawson's Appeal from the Magistrate Judge's Order Shifting Costs (Docs. 398 and 372). After Lawson's repeated attempts at discovering electronically stored information (“ESI”) as part of the voluminous discovery in this case, Defendant Spirit Aerosystems, Inc. (“Spirit”) asked the Court to shift technology-assisted review (“TAR”) costs to Lawson. The Magistrate Judge granted that request, and Lawson now appeals. For the following reasons, the Court affirms the Magistrate Judge's order. I. Factual and Procedural Background Lawson is Spirit's former chief executive officer. He retired on July 31, 2016. His Retirement Agreement contained non-compete obligations lasting two years, until July 31, 2018. In early 2017, non-party investment firms Elliott Associates, L.P. and Elliott International, L.P. (collectively, “Elliott”) hired Lawson for consulting services in connection with a proxy contest Elliott initiated against non-party Arconic, Inc. (“Arconic”). When Spirit learned of this, it notified Lawson that his involvement constituted a breach of his non-compete. Spirit then ceased paying Lawson and demanded that he repay what the company had already paid him under the Retirement Agreement. Lawson disputes that he breached the non-compete and filed this lawsuit seeking to recover the withheld payments under his Retirement Agreement. The non-compete provision in Lawson's Retirement Agreement prohibited him from being involved with “any business that is competitive with the Business or any portion thereof.” The Retirement Agreement defined the term “Business” as follows: We [Spirit] are engaged in the manufacture, fabrication, maintenance, repair, overhaul, and modification of aerostructures and aircraft components, and market and sell our products and services to customers throughout the world (... the “Business”). Lawson alleges that Spirit is a tier-one manufacturer of aerostructures and aircraft components (i.e., it builds and sells large structures and components like fuselage, propulsion, and wing systems) whereas Arconic is a tier-three or tier-four manufacturer of lightweight engineered metal components (e.g., small fasteners, connectors, bolts, engine components, fan blades, etc.) that are supplied to tier-one manufacturers like Spirit. Lawson therefore contends that Spirit and Arconic are not in the same “Business” because they do not provide, market, or sell the same “specific products and services.” Lawson and Elliott entered into two agreements on January 31, 2017. The first was a Consulting Agreement for Lawson to provide Elliott with consulting services in connection with the Arconic proxy contest. By the time Elliott and Lawson entered into the Consulting Agreement, Spirit had already notified them that Spirit believed Lawson's consulting arrangement with Elliott would violate Lawson's non-compete. So Lawson and Elliott also entered into an Indemnification Agreement by which Elliott agreed to indemnify Lawson if Spirit failed to pay him under his Retirement Agreement, in which case Elliott would become subrogated to the extent of those payments to Lawson's rights of recovery from Spirit. Elliott paid Lawson tens of millions of dollars under the Consulting and Indemnification Agreements and retained Lawson's litigation counsel at Elliott's expense. Elliott is now funding this lawsuit to recover the amounts Spirit allegedly owes Lawson under his Retirement Agreement. *2 During discovery, the parties were unable to agree on ESI custodians or search terms and had difficulty conferring productively. As a result, Lawson filed motions to compel Spirit to produce ESI regarding the business overlap issue. Spirit responded, arguing Lawson's ESI demands were disproportionate to the needs of the case and that Lawson was intentionally burdening Spirit with discovery. As an example, Spirit highlighted that Lawson had demanded that Spirit search 69 custodians’ ESI in addition to each custodian's assistant's ESI. Lawson also demanded that Spirit run these searches using roughly 90 search terms, but many of the searches contained one or more “OR” connectors, effectively expanding the number of search terms to far more than 100. None of the search terms were tailored to specific custodians. Many of the search terms were not tailored to the issues in the case. Other search terms were overly generic and lacked appropriate limiting terms. In February of 2019, Spirit had identified four individuals (out of the dozens of custodians Lawson had proposed) that Spirit believed would be most likely to have relevant and responsive information. Spirit then ran ESI searches using Lawson's proposed search terms. These searches returned more than 320,000 documents, of which Spirit reviewed a sample of approximately 400 and determined that 85% were irrelevant. As a result, Spirit considered Lawson's proposed search terms ineffective and told Lawson that Spirit would craft its own search terms. Spirit also suggested limiting the ESI searches to the ten custodians it believed were most likely to have relevant information. On April 23, 2019, Magistrate Judge Mitchell held a hearing on Lawson's motion to compel. The Court consulted with the parties about a proposed plan for tailoring ESI custodians and search terms. Beginning first with the issue of custodians, the Court rejected Lawson's request for 69 custodians and encouraged Lawson to prioritize his list of custodians because at some point the Court would start shifting costs. In consultation with the parties, the Court developed the following ESI protocol: • Lawson would first identify up to seven categories for which he was seeking ESI; • For each category, Spirit would list the top three custodians most likely to have relevant ESI, from the most likely to the least likely, along with a brief explanation as to why Spirit believed the custodian would have relevant information; • Lawson would then serve a list of five custodians with proposed search terms for each, and a second set of five custodians and search terms a week later; and • Spirit would search those custodians’ ESI using Lawson's search terms, conduct a sampling to determine responsiveness rates, and suggest modified search terms if the sampling revealed an unreasonably large number of non-responsive or irrelevant results. The Court directed the parties to work together on search terms to try to achieve an 85% responsiveness rate. The parties proceeded according to this protocol. Spirit provided Lawson with a list of custodians it thought most likely to have relevant ESI. Lawson picked only three custodians from Spirit's list. He disregarded Spirit's advice in selecting the remaining seven, none of whom were on Spirit's list. Lawson provided Spirit with 803 search terms (counting terms with “OR” as multiples) and asked Spirit to run those search terms on all of the ten identified custodians’ ESI. Spirit harvested the ten custodian files. They consisted of 1.8 million documents—1.2 million after de-duplication. Spirit ran the search terms. They returned 304,272 documents, or 468,595 documents including families, for a total of approximately 200GB of data. Spirit reviewed a 384-document sample and determined that only 7.8% were responsive. Of those, many were technically responsive but were irrelevant to the claims and defenses in this lawsuit. Spirit provided Lawson with hit reports for the first five custodians. Spirit also proposed revised search terms with corresponding hit reports for those custodians. *3 On June 6, 2019, Magistrate Judge Mitchell convened a telephone conference to discuss various discovery issues, one of which was Spirit's concerns about the lack of efficiencies in the ESI process. Spirit told the Court that Lawson had selected three custodians from Spirit's list, that the number of Lawson's search terms had increased to 803, and that Spirit's corresponding searches had resulted in only a 7.8% responsiveness rate. Turning first to Lawson's selected custodians, the Court remarked that Lawson's decision to pick seven custodians that were not on Spirit's list would be at his peril. Turning next to search terms, the Court limited Lawson to 25 search terms and instructed him to tailor them according to custodian rather than running the same search terms across all custodians. The Court again told the parties to work together to try to achieve an estimated responsive hit rate of at least 85%. On June 28, Lawson sent Spirit revised proposed search terms. Many of Lawson's revised terms were again common aviation-related terms as well as verbs commonly used in many industries. Spirit conducted new searches of the ten custodians’ ESI using Lawson's revised terms, which returned approximately 322,000 documents. A sample revealed that the response rates for each custodian ranged from 0.5% to 13.5%, with an average across all custodians of 5.1%. Spirit again characterized many of the responsive documents identified in the sampling exercise as “technically responsive” but “largely irrelevant to this dispute.” By August 9, the parties again conferred about search terms. Spirit stated that it believed continuing to discuss individual search terms and custodians would not be productive. Ten days later, Spirit produced responsive documents from its July sampling exercise, totaling only 173 documents. Spirit then produced 77 non-responsive documents to assist Lawson in determining why his search terms were resulting in such few responsive documents and next-to-no relevant documents. Spirit further advised Lawson that it believed reviewing the remaining approximately 322,000 documents “is not proportional to the needs of this case and will likely result in a small number of relevant documents.” Spirit predicted that “[b]ased on the most recent sampling exercise, it is likely that only 5% of these documents are responsive to outstanding discovery requests, and that these technically responsive documents are largely irrelevant to the dispute.” Around that time, the parties abandoned efforts to refine search terms to meet the 85% responsiveness-rate goal, and they began discussing the option of conducting a TAR of the 322,000-document set identified in July. Spirit's ESI vendor Legility offers a TAR tool called “Predict.” After an initial set of documents is coded for responsiveness, the Predict tool uses continuous active learning to code additional documents. Predict ranks coded documents from the most likely responsive to the least, and then humans review the top-ranked documents. When Predict determines the pool of responsive documents is depleted such that the effort of continued review is disproportionately outweighed by the possibility of additional gain, review ceases. Legility then conducts a statistical validation of the TAR's results. On September 17, the Court convened another telephone conference to discuss the parties’ progress on ESI and the case schedule. Lawson explained that the parties had discussed the TAR process and that Lawson wanted to proceed in that fashion. Spirit explained that it had been proceeding with document discovery on two different paths: (1) the ESI protocol and the process Lawson had discussed, and (2) separately, the “old-fashioned way” of targeted productions via custodian interviews and collections. According to Spirit, the second method had proven to be more efficient and effective. Using that method, Spirit had already produced about 39,000 pages of documents primarily on the issue of the “Business,” and Spirit wanted to continue to proceed down that path. Meanwhile, the ESI process was costly and yielded exceptionally low responsiveness rates. Spirit explained that the issue of business overlap between Spirit and Arconic was incredibly broad and disagreed with Lawson that using TAR would fix it. *4 By that time, Spirit had already spent hundreds of thousands of dollars on document collection, processing, and hosting, as well as the sampling exercises, and the parties had yet to achieve a 15% responsiveness rate. Given this, the Magistrate Judge raised the possibility of adjusting the case schedule to allow the parties to proceed with TAR, with Lawson bearing the TAR costs. The parties did not agree as to the allocation of costs at that time, but they agreed to move forward with the TAR process subject to Spirit filing a motion to shift those costs to Lawson. On September 19, Spirit reached out to Lawson in a last attempt to try to avoid a lengthy and expensive TAR that was unlikely to yield many responsive documents. Spirit reiterated that the sampling exercise it conducted in July suggested that only 5% of the 322,000 documents would be responsive. Spirit once again proposed that, in place of TAR, it continue to engage in its efforts to identify custodians who likely had information responsive to discovery requests, reviewing that information, and producing it. Through this process, Spirit had already produced approximately 4,700 documents, totaling approximately 40,000 pages. Spirit explained that the TAR could cost “$250,000-$400,000 in eDiscovery and document review costs, and $40,000-$60,000 in outside counsel time, as well as additional costs not yet identified.” Lawson reiterated that he believed the TAR was an effective and efficient means to review the documents from the custodians. He stated that he expected Spirit to produce documents located through the TAR on a rolling basis to be completed by November 1. On September 26, the parties met and conferred regarding the final TAR protocol, including a first-level review by contract attorneys and a second-level quality-control review by Spirit's counsel's law firm. After initiating the TAR protocol, the parties followed up on its results at a discovery conference on November 8. Spirit reported that it estimated ending the TAR upon achieving a 65% recall rate (the percent of the 322,000 documents searched) and substantially completing document production by December 6. On January 10, 2020, Spirit reported that it had reached a 68.5% recall rate. Lawson did not believe that was sufficient. Spirit agreed to keep running the TAR to an 80% recall rate but reminded Lawson of an eventual motion to shift costs. Spirit completed production of the TAR documents in mid-January after reaching an 85% recall rate. Only 3.3% of the documents in the TAR set of 322,000 documents were responsive. Of those documents, Spirit produced 23,951 documents, only 9,128 of which were deemed relevant. The rest were irrelevant or non-responsive. Notwithstanding this exceedingly low response rate, Lawson filed a motion to compel Spirit to produce the remaining TAR documents beyond the 85% recall rate. The Magistrate Judge denied this motion because Lawson refused to bear Spirit's costs to review and produce the residual TAR documents, no authority supported what Lawson was effectively seeking (a 100% recall rate), and further review was not proportional to the needs of the case. By late January 2020, Spirit estimated its TAR expenses to be approximately $400,000 in vendor costs and $200,000 in legal fees. Spirit moved for the Magistrate Judge to shift all costs and legal fees associated with the TAR to Lawson under Rule 26(c), which the Court granted on June 18, 2020. Lawson moved for reconsideration on July 2, which the Court denied on July 6. Lawson filed this appeal on July 20. II. Legal Standard *5 Upon objection to a magistrate judge's order on a non-dispositive matter, the district court may modify or set aside any portion of the order that it finds to be “clearly erroneous or contrary to law.” To be clearly erroneous, a decision must strike the court as “more than possibly or even probably wrong.” Thus, the court is required to affirm the magistrate judge's order unless the entire evidence leaves it “with the definite and firm conviction that a mistake has been committed.” III. Analysis In support of its original motion to shift costs, Spirit argued that it spent months collecting, processing, hosting, and searching millions of documents from custodians selected by Lawson and using search terms selected by Lawson, noting that this process cost hundreds of thousands of dollars and yielded only a tiny percentage of responsive or relevant documents. Spirit also noted that its auxiliary path of conducting custodian interviews and gathering targeted files resulted in far more significant and fruitful productions on the business overlap issue. Spirit therefore moved to shift the TAR expenses to Lawson to enforce proportionality standards. In her order on June 18, Magistrate Judge Mitchell granted Spirit's motion to shift the TAR costs to Lawson. The Court held that even though the default discovery rule is that the producing party should bear the costs of production, this case presented good cause to allocate the TAR expenses to Lawson. The Court explained that Spirit needed protection and relief from Lawson's burdensome and costly discovery tactics. Importantly, the Court noted that it had repeatedly cautioned Lawson to better focus his electronic discovery because the Court would eventually shift costs. The Court reiterated that Spirit had already shouldered its fair share of the expense by accommodating Lawson's many requests for ESI custodians and search terms, by running sampling exercises, and by facilitating an auxiliary discovery process utilizing traditional discovery means, which ended up producing more responsive documents than Lawson's overwhelming electronic discovery. After Spirit's extensive cooperation with Lawson and the Court's repeated warnings, Lawson nevertheless decided to proceed—at his own peril—with the costly and burdensome TAR. Holding that the electronic discovery process had become disproportionate to the needs of the case, Magistrate Judge Mitchell granted Spirit's motion to shift the TAR costs to Lawson. In his current appeal, Lawson argues that the Magistrate Judge's order should be overturned because it was clearly erroneous and contrary to the law. Lawson argues that the extensive and costly TAR was necessitated by Spirit's “broad affirmative defense” that it competed with Arconic and that Lawson's contract with Elliott was therefore a breach of contract. Regardless of Lawson's labeling, however, the interpretation of the Retirement Agreement's non-competition provision is the crux of this case, not an affirmative defense asserted solely by Spirit. Contrary to Lawson's implication, Spirit does not carry the sole burden of proving its interpretation of that provision. Both parties have a significant interest in discovery related to the business overlap issue. As such, Spirit's responsibility to cooperate during discovery and its responsibility to pay the associated costs is no different from a typical case. What is different is that, up until the Magistrate Judge's order, Spirit disproportionally carried the financial burden. *6 Lawson further argues that the Magistrate Judge's order shifting costs was clearly erroneous since the TAR uncovered some useful evidence to support his theory of the case. The Court is not persuaded by this “ends justify the means” argument. Even though a costly and overly thorough electronic discovery process produces some fruit does not prove that the discovery was proportionate to the case. The Magistrate Judge did not order costs shifted based on the forecast of an entirely fruitless TAR search. Rather, the Judge simply decided—within her sound discretion—that what little fruit would come from the search did not justify Spirit solely bearing its financial burden. Given the highly deferential standard of review, the Court concludes that Lawson has failed to carry his burden to prove that the Magistrate Judge's order was clearly erroneous or contrary to law. Lawson makes much of the fact that the Magistrate Judge issued her order based on briefings submitted in October 2019—before the TAR's final results were reported in January. Lawson argues that this renders the order indefensible. However, the TAR's final results were even worse than Spirit's predictions in its October briefing, which the Court relied upon in its order shifting costs. Based on its extensive sampling exercises, Spirit predicted that only 5% of the 322,000 would be responsive. Once the TAR was completed—at a substantial cost of time and money—only 3.3% of the documents proved responsive. While post-facto reasoning alone is insufficient to justify the Court's order, it does provide strong evidence for the fact that the Magistrate Judge's decision was based on sound principles and projections. The Magistrate Judge considered all the relevant facts and concluded that Lawson's persistence in pursuing the costly, ineffective TAR was disproportional to the needs of the case. Nothing in Lawson's appeal points to anything in the Magistrate Judge's order that this Court considers clearly erroneous or contrary to law. As such, the Court affirms the Magistrate Judge's order and denies Lawson's appeal. IT IS THEREFORE ORDERED that the Magistrate Judge's Orders Shifting Costs (Doc. 372) is AFFIRMED. IT IS FURTHER ORDERED that Plaintiff Larry A. Lawson's Appeal of Magistrate Judge Decision to District Court (Doc. 398) is DENIED. IT IS SO ORDERED. Footnotes  Doc. 1-3 at 8.  Id. at 2.  28 U.S.C. § 636(b)(1)(A); see also First Union Mortg. Corp. v. Smith, 229 F.3d 992, 995 (10th Cir. 2000) (citation omitted); Fed. R. Civ. P. 72(a).  United States v. Ludwig, 641 F.3d 1243, 1247 (10th Cir. 2011) (citation omitted).  Ocelot Oil Corp. v. Sparrow Indus., 847 F.2d 1458, 1464 (10th Cir. 1988) (quoting United States v. Gypsum Co., 333 U.S. 364, 395 (1948)).