United States v. Rishi Shah

Opinion

In the

United States Court of Appeals For the Seventh Circuit ____________________

Nos. 24-2230 & 24-2236 UNITED STATES OF AMERICA, Plaintiff-Appellee,

v.

RISHI SHAH and SHRADHA AGARWAL, Defendants-Appellants. ____________________

Appeals from the United States District Court for the Northern District of Illinois, Eastern Division. No. 1:19-cr-00864 — Thomas M. Durkin, Judge. ____________________

ARGUED FEBRUARY 10, 2026 — DECIDED AUGUST 6, 2026 ____________________

Before EASTERBROOK, SCUDDER, and KIRSCH, Circuit Judges. SCUDDER, Circuit Judge. In 2019, the federal government in- dicted Rishi Shah and Shradha Agarwal for orchestrating a multi-year, multi-million-dollar fraud scheme through their company, Outcome Health. Over three years later, and fol- lowing an 11-week trial, a jury convicted both Shah and Agarwal on multiple mail, wire, and bank fraud counts, and Shah on money laundering counts. 2 Nos. 24-2230 & 24-2236

Shah and Agarwal now appeal their convictions. Their main challenge focuses on an expansive, pretrial protective order that froze assets purportedly traceable to Outcome’s fraud. They contend that the order was overbroad and vio- lated their Sixth Amendment right to counsel of choice by im- properly restraining assets—so much so that they lacked the resources to continue paying the firms and lawyers they had hired to represent them in the case. Shah and Agarwal also allege that the government violated their Fifth Amendment due process rights when an FBI accountant knowingly made a false statement (related to the pretrial asset restraint) to the grand jury and the government failed to correct it. Finally, they argue that the district court abused its discretion by ad- mitting certain evidence at trial and in its instructions to the jury. While what transpired with the pretrial asset freeze is complicated and troubling, we see no error in the district court’s finding that Shah and Agarwal received sufficient in- formation during discovery to identify and challenge the over-restraint well before trial. Nor do we see any merit to Shah and Agarwal’s remaining challenges. All of this leads us to affirm. I A. Outcome Health In 2006, Rishi Shah founded Outcome Health, a healthcare technology company that provided television screens and tablets displaying ads and educational content in doctors’ of- fices. Shah served as Outcome’s CEO. Shradha Agarwal, Shah’s college classmate, joined the company in 2008 as its Nos. 24-2230 & 24-2236 3

Chief Marketing Officer and Chief Strategy Officer before transitioning to President and Co-Founder in 2013. Outcome generated revenue by selling ad space on screens to pharmaceutical companies and advertising agencies. To contract with Outcome, a company would provide a list of doctors’ offices it wanted an ad campaign to target. Outcome would then perform a “list match” of offices where its screens were installed to find the overlap. Ad campaign contracts of- ten promised the client a return on their investment, meas- ured by the amount of money earned in prescriptions or other revenue for every dollar spent on ads. By 2017, Outcome expanded from a 20-person startup to a company of over 500 employees with multiple offices. Tur- moil ensued in October 2017, when the Wall Street Journal re- ported that Outcome had been defrauding its clients by over- selling ad space and inflating performance metrics. Federal criminal charges followed, with the indictment alleging that Outcome’s multi-million-dollar fraud scheme ran from at least 2011 to 2017, and targeted both clients and investors. Outcome still operates today, rebranded as PatientPoint. 1. Fraud on Clients Outcome defrauded its clients to solve challenges associ- ated with its business model. To evolve from a startup to a nationwide enterprise, the company required enough screens to run ads and thereby attract advertisers. At the same time, it needed more ad revenue to purchase more screens. To bridge the gap between inventory and revenue needs, Out- come told clients it had more screens than it did to induce them to enter larger advertising contracts. When a client sent Outcome a list of doctors to target, the company inflated its 4 Nos. 24-2230 & 24-2236

list match using what was really a projected (and not actual) inventory figure. At least some clients did not know the list matches they received overstated Outcome’s current inven- tory. So they signed contracts believing their ads would run on screens that Outcome did not in fact have. And while some clients agreed to “weighted” contracts that promised growth over a certain period to reach an average projected figure, most contracted for current inventory. Outcome executives si- loed information about the list manipulation from sales em- ployees interfacing with clients. Outcome consistently overstated its inventory and under- delivered on ads, thereby also failing to generate promised returns, while all along billing clients for full service. The company referred to these performance gaps as “deltas.” It then hid the deltas from clients by manipulating information used to conduct ad campaign performance studies. Evidence at trial demonstrated that Shah, and to a lesser extent, Agarwal, knew of these problems but ignored or silenced em- ployees who tried to sound an alarm. 2. Fraud on Lenders and Investors Outcome defrauded its lenders and investors too. By 2016, the company began seeking outside investment. To help se- cure capital, Outcome hired Deloitte to audit its 2015 and 2016 financial statements. But in the process, the company hid its fraud by making misrepresentations to the auditors within misstated financial statements. Outcome nonetheless used those financial statements, and Deloitte’s resulting clean audit opinions, to secure $485 million in loans from JPMorgan and other lenders in April and December 2016, and over $487 mil- lion from equity investors from March to July 2017. From the proceeds of the financing, Outcome paid a $225 million Nos. 24-2230 & 24-2236 5

dividend to Gravitas Holdings, LLC, an entity controlled by Shah and Agarwal. B. Civil Settlement After Outcome’s fraud scheme became public in October 2017, its lenders and investors sued. The parties settled in Jan- uary 2018. Shah and Agarwal agreed to resign, surrender their controlling interest in Outcome, and pay about $190 mil- lion from Gravitas to the company and investors. The settle- ment allowed them to keep $31 million from Gravitas. In an- ticipation of a federal indictment, Shah and Agarwal paid $10.3 million of the $31 million to their attorneys as a retainer fee. C. Indictment and Forfeiture Allegations In November 2019, a grand jury indicted Shah and Agarwal on multiple counts of mail, wire, and bank fraud. Shah also faced a separate money laundering charge. Federal law allows the government to seek a pretrial pro- tective order freezing a defendant’s assets traceable to crimi- nal activity alleged in the indictment. See 21 U.S.C. § 853(e)(1)(A). The indictment against Shah and Agarwal in- cluded forfeiture allegations stating that “[a]ll right, title, and interest … including but not limited to” certain funds in cer- tain assets held by two of Shah and Agarwal’s companies (Gravitas Holdings and Jumpstart Ventures II) were traceable to the fraud and therefore forfeitable upon conviction. To arrive at the indictment’s forfeiture allegations prior to the grand jury, the government assembled a multi-agency team to identify Shah and Agarwal’s forfeitable assets. For her part, FBI forensic accountant Megan Poelking traced the as- sets. Given the complexity of the case, Poelking received 6 Nos. 24-2230 & 24-2236

assistance from the DOJ’s Money Laundering and Asset Re- covery Section and the U.S. Marshals Service Complex Asset Unit. DOJ attorney Daniel Olinghouse drafted the forfeiture allegations and protective order that would follow. Poelking testified before the grand jury as the govern- ment’s tracing expert about the criminal proceeds the govern- ment sought to freeze. After the grand jury returned the in- dictment and forfeiture allegations, the district court entered a protective order restraining assets pending trial. The frozen assets totaled in the tens of millions of liquid dollars, includ- ing the $10.3 million Gravitas dividend Shah and Agarwal re- tained through the settlement agreement (the “Settlement Funds”), which they had set aside for legal fees. It also in- cluded other mainly illiquid assets of unknown value (the “Other Assets”). Meanwhile, Shah and Agarwal took steps to retain coun- sel for trial. As early as 2017, two years before the indictment, Shah and Agarwal had paid a retainer to Quinn Emanuel Ur- quhart & Sullivan, LLP to represent them in potential investi- gations and lawsuits. After the indictment, William Burck and Jonathan Bunge of Quinn Emanuel entered a limited appear- ance for Shah, and Christina Egan of McGuireWoods for Agarwal. These are the lawyers and firms Shah and Agarwal wanted to represent them through trial. But the broad asset restraint presented a challenge, as both firms required pay- ment up front. So access to cash or other liquid assets became very important for Shah and Agarwal. Their focus naturally turned to the protective order. In January 2020, with Quinn Emanuel and McGuire- Woods still retained, Shah and Agarwal challenged the scope of the protective order by focusing on the frozen $10.3 million Nos. 24-2230 & 24-2236 7

in Settlement Funds. They maintained that the funds were both essential to continue retaining their counsel of choice and not subject to forfeiture. The district court disagreed and denied Shah and Agarwal’s motion, finding that the settle- ment did not somehow cleanse the funds of their allegedly fraudulent origin. The court gave Shah and Agarwal until June 30, 2020, to find new trial counsel. In June 2020, Quinn Emanuel and McGuireWoods with- drew from the case due to Shah and Agarwal’s inability to pay. Shah proceeded to trial with Hueston Hennigan LLP and Agarwal with Larson LLP and Blegen & Associates. D. Trial and Sentencing Trial began on January 30, 2023, and ran for 11 weeks, with the jury hearing from Outcome employees, pharmaceutical clients and ad agency representatives, lenders and investors, a Deloitte auditor, FBI forensic accountant Poelking, and var- ious experts. Several former Outcome employees involved in the fraud who testified at trial are relevant to this appeal: Da- vid Ma, a Senior Sales Analyst from mid-2014 until late 2015; Jason Ketchum, a Member Services Executive who assisted with sales, inventory projections, and list matches from 2012 to 2013; and Ashik Desai, Vice President of Research and An- alytics, who took over Ketchum’s sales role around 2013. Ma and Ketchum received immunity before trial while Desai pleaded guilty and agreed to testify as a government witness. During the trial, Shah and Agarwal challenged the gov- ernment’s motions to admit statements that Ma, Ketchum, and Desai made while testifying before the grand jury. The district court admitted the testimony as prior consistent state- ments under Federal Rule of Evidence 801(d)(1)(B) and the 8 Nos. 24-2230 & 24-2236

government introduced them during its redirect examina- tions of the witnesses. In the end, the jury convicted Shah on five counts of mail fraud (18 U.S.C. § 1341), 10 counts of wire fraud (18 U.S.C. § 1343), two counts of bank fraud (18 U.S.C. § 1344), and two counts of money laundering (18 U.S.C. § 1957). It convicted Agarwal of five counts of mail fraud, eight counts of wire fraud, and two counts of bank fraud. The district court sentenced Shah to 90 months’ imprison- ment and Agarwal to one day, with both defendants also re- ceiving three years of supervised release. The court further imposed a criminal fine of $1,900 on Shah and ordered him to forfeit $55 million. It fined Agarwal $1,500 and ordered her to forfeit $13.7 million. The district court has yet to set restitu- tion. E. Post-Trial Litigation A flurry of post-trial litigation related to Shah and Agarwal’s ability to retain counsel occupies much of this ap- peal. In March 2023, midway through trial, the government produced tracing spreadsheets and internal emails that Shah and Agarwal claim first revealed the over-restraint of their Other Assets. The protective order, informed by the forfeiture allegations, covered “all right, title, and interest” in certain as- sets held by Gravitas and Jumpstart II, “including but not lim- ited to” amounts identified in the order’s text. Dkt. 27 (empha- sis added). But as the documents produced at trial made clear, the government had intended only to restrain, and thus only traced, the specific amounts listed within each asset. Therein lies the over-restraint: the protective order’s “including but Nos. 24-2230 & 24-2236 9

not limited to” language resulted in more assets being frozen than those it traced. More specifically, as shown by docu- ments produced before trial and explained by the govern- ment, it sought to restrain only discrete amounts linked to the financing fraud that occurred in 2016 and 2017. How did this happen? Through carelessness on the gov- ernment’s part. DOJ attorney Daniel Olinghouse drafted the protective order and both he and everyone else involved in preparing the indictment and overseeing the protective order failed to catch the disconnect between its overbroad language and Poelking’s tracing analysis. As best we can tell, and as the district court found, the par- ties first recognized the over-restraint after trial. Following Shah and Agarwal’s convictions, the government moved to seize Shah’s forfeitable assets. During a hearing in July 2023, the government recounted its intent with the asset restraint— to freeze assets directly linked to the financing fraud—and de- scribed Poelking’s tracing method. In doing so, the govern- ment observed that some private equity entities targeted by the protective order held assets Poelking had not traced to any fraud. Shah’s counsel seized on that clarification to explain that the government must have over-restrained funds be- cause the “including but not limited to” language in the pro- tective order covered everything within a targeted entity or investment asset. Shah and Agarwal then filed motions for ac- quittal or for a new trial raising two claims related to the asset restraint. First, they claimed that the government violated their Sixth Amendment right to counsel of choice by over-restrain- ing their Other Assets. Second, they contended that the gov- ernment violated their Fifth Amendment due process rights 10 Nos. 24-2230 & 24-2236

when FBI forensic accountant Poelking expressly but incor- rectly represented to the grand jury that there was probable cause to believe all assets listed in the protective order were traceable to fraud. The government’s failure to correct the misstatement, Shah and Agarwal asserted, contributed to the Sixth Amendment violation. Extensive post-trial proceedings ensued. The district court invested substantial effort into getting to the bottom of what happened and, more specifically, whether the over-restraint of assets precluded Shah and Agarwal from going to trial with Quinn Emanuel and McGuireWoods. The court ordered dis- covery and held evidentiary hearings on the issue over the course of four months and, in the end, denied the post-trial motions. The district court determined that Shah and Agarwal for- feited their Sixth Amendment claim by raising it after trial when they could have identified the over-restraint years ear- lier based on the government’s pretrial disclosures. It then concluded in the alternative that Shah and Agarwal failed to show that they could have liquidated the improperly re- strained Other Assets for enough to afford their counsel of choice. The over-restraint therefore did not violate their Sixth Amendment rights. The district court also denied the defend- ants’ Fifth Amendment claim, concluding that it too was un- timely and, in any case, Poelking did not knowingly provide false testimony to the grand jury, nor did the government knowingly fail to correct her misstatement. Shah and Agarwal then appealed. Nos. 24-2230 & 24-2236 11

II We begin with Shah and Agarwal’s Sixth Amendment claims. By way of framing, keep in mind that Shah and Agarwal allege the government restrained two sets of assets resulting in a violation of their right to counsel of choice and raise sep- arate challenges to the district court’s handling of those assets. First, they challenge the district court’s denial of their pretrial motion to unfreeze $10.3 million in Settlement Funds. Second, they contest the court’s post-trial determination that they failed to preserve their challenge to the government’s admit- tedly improper restraint of their Other Assets and its finding that the restraint did not prevent them from hiring their coun- sel of choice. By any measure, the facts here raise complex questions about the effect of a pretrial asset restraint on a defendant’s right to counsel of choice. We approach the issue by first lay- ing the legal groundwork and then applying it to both chal- lenges. Ultimately, we see no Sixth Amendment violation. A. Legal Background on the Right to Counsel of Choice and Pretrial Asset Restraints The right to counsel of choice is a bedrock feature of the Sixth Amendment. “Given the necessarily close working rela- tionship between lawyer and client, the need for confidence, and the critical importance of trust, … ‘a fair opportunity to secure’” counsel of choice is essential. Luis v. United States, 578 U.S. 5, 11 (2016) (plurality opinion) (quoting Powell v. Alabama, 287 U.S. 45, 53 (1932)); see also United States v. Gonzalez-Lopez, 548 U.S. 140, 147–48 (2006) (recognizing that the right to coun- sel of choice is the “root meaning” of the Sixth Amendment). 12 Nos. 24-2230 & 24-2236

A defendant’s selection of counsel is of major conse- quence. The trust and confidence essential to the client-coun- sel relationship will inform key strategic decisions at every step. So, too, is it common for a defendant to seek and retain counsel with particular experience and skill. The Sixth Amendment protects this right to choose and affords it great significance. See Gonzalez-Lopez, 548 U.S. at 150. Indeed, when a defendant is wrongfully denied their counsel of choice, we consider the error structural, as “[i]t is impossible to know what different choices the rejected counsel would have made, and then to quantify the impact of those different choices on the outcome of the proceedings.” Id. Put another way, the er- ror defies harmless-error analysis and is deemed to affect the “framework within which the trial proceeds.” Id. at 148, 150– 51 (cleaned up). Structural errors are “subject to automatic re- versal.” Greer v. United States, 593 U.S. 503, 513 (2021) (cleaned up). But the right to counsel of choice is not without limits. Rel- evant here, “[t]he Sixth Amendment guarantees a defendant the right to be represented by an otherwise qualified attorney whom that defendant can afford to hire.” Luis, 578 U.S. at 12 (emphasis added) (cleaned up). Shah and Agarwal’s contention—that the government’s improper asset restraint prevented them from affording their counsel of choice—focuses us on the interaction between the right to counsel of choice and federal forfeiture law. Under 21 U.S.C. § 853, property derived from or used in the commis- sion of certain crimes is forfeitable to the federal government upon conviction. See United States v. Monsanto, 491 U.S. 600, 606 (1989) (“[T]he language of § 853 is plain and unambigu- ous: all assets falling within its scope are to be forfeited upon Nos. 24-2230 & 24-2236 13

conviction ….”). But the government does not have to wait for a conviction to freeze assets subject to forfeiture. Indeed, as the government did here, it may include forfeiture allegations in an indictment identifying certain property as forfeitable. See 21 U.S.C. § 853(e)(1)(A). Upon indictment, a district court may then enter a protective order restraining those assets to ensure they are available after conviction. Id. To secure a pre- trial asset restraint, the government must show a grand jury that there is “probable cause to think (1) that the defendant has committed an offense permitting forfeiture, and (2) that the property at issue has the requisite connection to that crime.” Kaley v. United States, 571 U.S. 320, 323–24 (2014). The Sixth Amendment moderates the government’s pre- trial forfeiture authority. On the one hand, the government may restrain assets directly traceable to the crime alleged without running afoul of the Sixth Amendment, even if the defendant needs those funds to retain counsel. See Luis, 578 U.S. at 13–14 (discussing Caplin & Drysdale, Chartered v. United States, 491 U.S. 617 (1989) and Monsanto, 491 U.S. 600). That is so because § 853(c) grants the government a superior interest in the traceable assets upon commission of the crime. See id. at 13–15. We consider those assets “tainted” by the criminal act. Id. at 13. On the other hand, the government may not restrain a de- fendant’s “untainted” assets under § 853, even to substitute for tainted assets the defendant has already spent. See id. at 15–16. And “insofar as innocent (i.e., untainted) funds are needed to obtain counsel of choice … the Sixth Amendment prohibits” an asset restraint. Id. at 18. The Supreme Court’s decision in Luis illustrates these points. Sila Luis faced federal charges for allegedly reaping 14 Nos. 24-2230 & 24-2236

about $45 million through fraud. See id. at 9. By the time of indictment, she had spent most of the money. See id. The gov- ernment nevertheless obtained a protective order restraining all of her assets up to the $45 million she realized through the alleged fraud. See id. The order left Luis without money to retain counsel. See id. at 9–10. A plurality of the Supreme Court held the government’s pretrial restraint improper be- cause it targeted untainted, substitute assets. See id. at 23. The over-restraint, the Court further determined, violated Luis’s right to the counsel of choice she could otherwise afford be- cause it left her no funds to retain any attorney. See id. This appeal presents a more complicated fact pattern. No restraint, proper or improper, left Shah or Agarwal indigent. Not even close. Instead, they had sufficient resources to hire some of the most able trial lawyers in the country. But those lawyers were not Shah and Agarwal’s first pick. So we must determine whether the government’s pretrial over-restraint of assets was substantial enough to have prevented Shah or Agarwal from hiring their counsel of choice. Relying on Luis, the defendants urge us to reach the broader conclusion that any improper asset restraint that comes to light post-trial, no matter its dollar amount, is a per se Sixth Amendment violation. That is so, they continue, be- cause any restriction on a defendant’s assets reduces the range of attorneys available to hire and thereby denies him a “fair opportunity” to retain counsel of choice. Powell, 287 U.S. at 53. But Supreme Court precedent does not extend that far. This is not a Luis situation in which the over-restraint neces- sarily prevented the defendant from exploring or obtaining counsel. And the Court has never held that any over-restraint Nos. 24-2230 & 24-2236 15

automatically violates the Sixth Amendment by virtue of shrinking a defendant’s budget. The question presented here, then, as we see it, is not whether the over-restraint prevented Shah and Agarwal from retaining more expensive counsel, but whether it prevented them from being able to afford the counsel they expressly desired. Nobody disputes that Shah and Agarwal wanted to go to trial represented by specific attorneys at Quinn Emanuel and McGuireWoods. They revealed that preference when they freely selected those firms pre-indictment with all funds (tainted and untainted) at their disposal. And at no point did either defendant identify a second, backup choice more pre- ferred than the counsel they ultimately retained or show that the overbroad asset restraint dissuaded them from shopping for alternative firms. The essential question, then, is whether the government’s over-restraint precluded the retention of those firms’ attorneys for trial. By the time both firms withdrew from the case, Shah and Agarwal required an additional $8.5–9.5 million ($5–6 million for Quinn Emanuel and $3.5 million for McGuireWoods) to retain them through the trial. The district court credited their representations that they had paid counsel all they had avail- able at that time. Shah’s counsel, Bill Burck, later told the dis- trict court that $7.8 million would have sufficed for both firms. We too accept that representation and assume that Shah and Agarwal collectively needed $7.8 million for their counsel of choice. The remaining question is whether the gov- ernment’s pretrial over-restraint met or exceeded that amount. We tackle that question by focusing on the two cate- gories of restrained assets—the Settlement Funds and Shah and Agarwal’s Other Assets. 16 Nos. 24-2230 & 24-2236

B. Merits of Settlement Funds Challenge Recall that Shah and Agarwal challenged the restraint of $10.3 million they were allowed to keep as part of the civil settlement and attempted to use for attorneys’ fees. They do not dispute that the government traced these funds to the broader pool of $487 million that Outcome fraudulently raised from investors in 2017. As a result of this equity financ- ing, Outcome paid a $225 million dividend to Shah and Agarwal’s entity, Gravitas Holdings. Shah and Agarwal then retained $31 million of the $225 million in Gravitas equity through the civil settlement and set $10.3 million of it aside for legal fees. In the district court, and again on appeal, Shah and Agarwal insist that Outcome’s civil settlement washed or cleansed the $31 million, including the $10.3 million they sought to unfreeze, of all criminal taint. In their view, the fact that Outcome, its investors, and its lenders involved in the settlement knew the funds came from fraud and still agreed to let Shah and Agarwal keep those funds in return for the surrender of their indemnification rights and other things of value left the money free and clear to be used to retain crimi- nal defense counsel. The district court considered this argument attenuated in the extreme, as do we. First, and most obviously, the civil set- tlement was a negotiated agreement between private parties. The United States was not involved. So the agreement in no way affected the criminal proceeding against Shah and Agarwal. Put more directly, the civil settlement had no bear- ing on the nexus the government established before the grand jury between Shah and Agarwal’s fraud and Outcome’s fi- nancing in 2017. Nos. 24-2230 & 24-2236 17

Second, under 21 U.S.C. § 853(c), “all right, title, and inter- est” in forfeitable property “vests in the United States upon the commission of the act giving rise to forfeiture.” That means that once the criminal offense is complete, a defendant no longer holds “good title” to forfeitable property. Luis, 578 U.S. at 14 (quoting Caplin & Drysdale, 491 U.S. at 627). Nor does a defendant have a right to “give [the government’s] property to a third party.” See id. (quoting Caplin & Drysdale, 491 U.S. at 628). Shah and Agarwal therefore had no right to use any part of the allegedly tainted Gravitas dividend to pay their attorneys’ fees. And a private party’s assent to the use of forfeitable funds cannot override what the law tells us: the funds may be restrained. The district court committed no error in finding that the $10.3 million in Settlement Funds deriving from the Gravitas dividend remained tainted by their illicit origin. C. Shah and Agarwal’s Other Assets Shah and Agarwal also contend that the government’s im- proper restraint of their Other Assets prevented them from continuing with Quinn Emanuel and McGuireWoods through trial. Remember that the defendants first lodged this challenge in their post-trial motions. By then the government had conceded that the protective order was overbroad. So the focus in the district court was on whether the over-restraint was at least $7.8 million and thereby large enough to have prevented Shah and Agarwal from staying with Quinn Eman- uel and McGuireWoods. The Other Assets the government restrained were mostly illiquid—mainly investment interests Shah and Agarwal held in private equity entities and companies. The district court 18 Nos. 24-2230 & 24-2236

therefore assessed whether the defendants, if they had access to those assets from the outset, could have liquidated at least $7.8 million of the investments and then used the cash pro- ceeds to retain their counsel of choice for trial. In 2023, the district court devoted nearly four months of post-trial discovery and evidentiary hearings to the issue. In the final analysis, the district court concluded that Shah and Agarwal raised an untimely Sixth Amendment challenge be- cause they could have identified the over-restraint years be- fore trial, as early as January or February 2020, when they re- ceived the grand jury transcript, exhibits, indictment, protec- tive order, and bank records revealing the over-restraint. In the alternative, the district court addressed the merits and assessed whether the over-restraint of the Other Assets met or exceeded the $7.8 million needed to retain the firms. On this front, the court required Shah and Agarwal to show by a preponderance that they could have liquidated or other- wise realized enough money from the illiquid assets to pay Quinn Emanuel and McGuireWoods within a reasonable time before the scheduled trial. After extensive proceedings, the district court found Shah and Agarwal’s liquidity estimate in- flated and speculative. This left the court to conclude that Shah and Agarwal failed to establish a Sixth Amendment vi- olation. 1. Forfeiture “If a litigant believes that an error has occurred (to his det- riment) during a federal judicial proceeding, he must object in order to preserve the issue.” Puckett v. United States, 556 U.S. 129, 134 (2009). When a defendant preserves a claim of error, the burden rests on the government to show the error Nos. 24-2230 & 24-2236 19

was harmless. See Fed. R. Crim. P. 52; Chapman v. California, 386 U.S. 18, 24 (1967) (holding in the context of a preserved constitutional error that the beneficiary of the error must prove harmlessness). Conversely, “if [a defendant] fails to [object] in a timely manner, his claim for relief from the error is forfeited.” Puckett, 556 U.S. at 134. We review forfeited claims for plain error. See United States v. Jones, 844 F.3d 636, 640 (7th Cir. 2016). Under the plain error standard, “the tables are turned” and the defendant bears the burden of showing reversible er- ror. United States v. Vonn, 535 U.S. 55, 62–63 (2002). To do so, he must demonstrate (1) an error, (2) the error is plain, and (3) the error affects the defendant’s “substantial rights.” Greer, 593 U.S. at 507–08. If a defendant meets these threshold re- quirements, we may grant relief if the error “had a serious ef- fect on the fairness, integrity or public reputation of judicial proceedings.” Id. at 508 (cleaned up); see also United States v. Page, 123 F.4th 851, 864 (7th Cir. 2024) (reiterating the same standard for plain error review). Few courts have addressed the point at which a challenge to a pretrial asset restraint becomes forfeited. But all signs in- dicate that after trial is too late, at least when a defendant had the information necessary to challenge the restraint earlier. See United States v. Newton, 76 F.4th 662, 669 (7th Cir. 2023) (“[I]ssues that should have been brought at trial and are raised for the first time in a post-trial motion are forfeited.”). That view aligns with Rule 51(b), which instructs parties to object “when the court ruling or order is made.” Fed. R. Crim. P. 51(b); see also Puckett, 556 U.S. at 135 (“Failure to abide by this contemporaneous-objection rule ordinarily precludes the raising on appeal of the unpreserved claim of trial error.”). All 20 Nos. 24-2230 & 24-2236

for good reason, as the contemporaneous-objection rule en- sures a district court has an opportunity to correct any error. Just so here, a defendant who objects to a pretrial asset re- straint is entitled to an immediate hearing to challenge a pro- tective order “if the pretrial seizure of his assets would pre- vent him from hiring the counsel of his choice.” United States v. Kielar, 791 F.3d 733, 739 (7th Cir. 2015) (citing United States v. Moya-Gomez, 860 F.2d 706, 730 (7th Cir. 1988)); see also Ka- ley, 571 U.S. at 324 (observing that lower courts “have uni- formly allowed the defendant to litigate … whether probable cause exists to believe that the assets in dispute are traceable … to the crime charged in the indictment”). If the district court declines to modify the restraint, the defendant may ap- peal immediately. See United States v. Kirschenbaum, 156 F.3d 784, 788 (7th Cir. 1998). But if the court finds error, it may or- der a release of the funds the defendant needs for counsel, continue the trial, or declare a mistrial and start over. Any one of these options is less costly and time consuming than dis- missing the indictment or ordering a new trial on appeal. For these reasons, a defendant must challenge a pretrial asset restraint within a reasonable period after discovering or obtaining the means to discover the restraint may be im- proper. See, e.g., Jones, 844 F.3d at 639–41 (deeming pretrial asset restraint challenge raised for the first time on appeal for- feited and applying plain error review). Shah and Agarwal challenged the restraint of their Other Assets three months after their trial concluded. They claim, as they did in the district court, that they had no basis to do so earlier because, until the government produced revealing in- ternal documents mid-trial, they reasonably believed it in- tended to restrain all assets traceable to Outcome on the Nos. 24-2230 & 24-2236 21

theory that the “entire business” was a fraud. The district court examined this contention in great depth and found it unsupported by the record. We do too. No doubt the indictment advanced a broad theory of fraud by Outcome Health. Some counts charged Shah and Agarwal with mail and wire fraud based on payments clients made to Outcome pursuant to a misleading revenue model grounded in the manipulation of advertising data. Other counts charged them with wire and bank fraud based on transfers from investors or loans issued by banks to the com- pany in 2016 and 2017. Under § 853, the government could move to restrain assets traceable to any of these forms of fraud. And indeed, one forfeiture allegation included in the indictment expressly references counts related to both the fraud on Outcome’s clients and the fraud on its lenders and investors. But there is more, and the more is critical. Shortly after the indictment and the district court’s issuance of the protective order, the government made a discovery production. The pro- duction, which occurred in January 2020, included the tran- script of the grand jury proceedings and the exhibits the gov- ernment presented to the grand jury. The government also produced by February 2020 bank records and documents from private equity firms and other companies—all of which reflected the assets affected by the protective order. These documents together belied a view that the govern- ment sought to restrain anything beyond proceeds the of Out- come’s fraudulent financing in 2016 and 2017. Put more di- rectly, the documents showed that the government, despite the over-broad language in the protective order (the “includ- ing but not limited to” formulation), did not intend to restrain 22 Nos. 24-2230 & 24-2236

all of Shah and Agarwal’s assets bearing any connection to Outcome. Allow us to unpack these points. In her grand jury testi- mony, the government’s tracing expert, FBI forensic account- ant Megan Poelking, identified the assets the government sought to restrain and explained how it traced those assets to the crimes alleged. A grand juror directly asked Poelking what the “criminal proceeds” included in this case—in other words, what fraudulent funds the government sought to re- strain. And she responded in no uncertain terms: “[i]n this in- dictment, we’re alleging that the capital raise [and] the loans that they received will be criminal proceeds.” Dkt. 512-1 at 18 (em- phasis added). At no point did Poelking mention or trace funds from the fraud on Outcome’s healthcare clients or more broadly state that the government considered all of Out- come’s operations and assets the fruit of the fraud. Even more, Poelking itemized the amounts sought to be restrained in exhibits—spreadsheets that became part of the grand jury record. The spreadsheets tracked the flow of funds from Outcome’s 2016 loans and 2017 equity financing from investors and bore corresponding titles—“Assets Subject to Forfeiture Attributable to the Loans Obtained in 2016,” “As- sets Subject to Forfeiture Attributable to the Capital Raise in 2017,” and “Assets Subject to Forfeiture Attributable to the Loans Obtained in 2016 and the Capital Raise in 2017.” Every amount listed in each forfeiture allegation, which the protec- tive order mirrors, aligns with the traceable amounts itemized in the spreadsheets shown to the grand jury. So while one for- feiture allegation within the indictment referenced counts re- lated to the client fraud, no amount listed within the forfeiture allegations or on Poelking’s spreadsheets traced to that fraud. Nos. 24-2230 & 24-2236 23

What all of this means is that Shah and Agarwal, having received not only the indictment and protective order, but also Poelking’s testimony and exhibits (the tracing spread- sheets), as well as bank and private equity records, had eve- rything they needed to detect the over-restraint. All they had to do was realize that the government had frozen more assets and larger dollar amounts than Poelking indicated that the prosecution intended to target during her grand jury testi- mony, including in her tracing spreadsheets. To be sure, Poelking made a misleading statement to the grand jury when she confirmed that, having reviewed the for- feiture allegations in the indictment, to the best of her knowledge, they “list items for which there is probable cause that they came from proceeds of the fraud.” That was not ac- curate because the “including but not limited to” language in the forfeiture allegations expanded them beyond the amounts listed in the underlying tracing analysis. While we recognize that Poelking’s testimony injected confusion, it did nothing to alter the particular assets and amounts the government meant to freeze—those from the 2016 and 2017 financing activity and itemized with precision. Poelking told the grand jury what the government intended to restrain and specified those amounts on her spreadsheet, exposing a disconnect with the “including but not limited to” language in the protective or- der. Shah and Agarwal were positioned to question, if not outright challenge, the scope of the restraint. The district court reached this precise conclusion and em- phasized that the post-trial proceedings revealed that Shah and Agarwal received the grand jury materials in January 2020, just a couple of months after the indictment and three years before trial. Had counsel reviewed the materials, they 24 Nos. 24-2230 & 24-2236

could have seen that the government’s grand jury presenta- tion on traceability conflicted with the scope of the protective order, and then raised questions or moved for a pretrial hear- ing to require the government to “prove the likelihood that the restrained assets are subject to forfeiture.” Jones, 844 F.3d at 640 (quoting Moya-Gomez, 860 F.2d at 731). That is what should have happened here. If Shah and Agarwal had sought a hearing, or just pressed for more infor- mation, the government almost certainly would have realized its error, reinforced its intent to restrain just those assets of Shah and Agarwal’s traceable to the financing fraud, and moved to modify the protective order. Their failure to take any of these measures—even mid-trial when the govern- ment’s document production further revealed its limited asset tracing—leads us to agree with the district court’s finding that they forfeited their Sixth Amendment challenge to the gov- ernment’s over-restraint of their assets. 2. Burden Allocation But concluding a claim is forfeited does not mean it is un- reviewable. Instead, the Supreme Court has made clear time and again that when a defendant forfeits a claim of error, he bears the burden of demonstrating plain error. See Greer, 593 U.S. at 507–08; Vonn, 535 U.S. at 62–63; United States v. Olano, 507 U.S. 725, 734 (1993); Jones, 844 F.3d at 640. The government urges us to hold Shah and Agarwal to this burden by requir- ing them show that the over-restraint was at least $7.8 million and therefore precluded them from being able to afford Quinn Emanuel and McGuireWoods through trial. But Shah and Agarwal disagree, inviting us to flip the bur- den and place it on the government as the party responsible Nos. 24-2230 & 24-2236 25

for the overbroad protective order. In pressing this position, they contend that the most effective way to deter a mistake like the one that happened here is to require the government to show no Sixth Amendment violation, especially given that the government fully acknowledges it over-restrained. They point to several non-structural prosecutorial errors in which the government must show harmlessness for this reason. See Medina v. California, 505 U.S. 437, 451–52 (1992) (collecting cases concerning errors such as the waiver of Miranda rights, inevitable discovery of evidence obtained by unlawful means, voluntariness of consent to search, and voluntariness of a con- fession). Shah and Agarwal have not identified an example of a for- feited claim for which the defendant is relieved of the burden of showing reversible error, however. Indeed, the burden on plain error review falls on the defendant, even where the gov- ernment has made a mistake. See Greer, 593 U.S. at 507–08; see, e.g., United States v. Johnson, 415 F.3d 728, 731 (7th Cir. 2005) (concluding that even if the defendant showed a Mi- randa error, he could not satisfy all plain error prongs); United States v. Raney, 342 F.3d 551, 559 (7th Cir. 2001) (concluding that even if the defendant showed an unlawful seizure, he could not satisfy all plain error prongs). And while there are some “serious [non-structural] er- rors” to which we assign a rebuttable presumption of preju- dice, like jury tampering, see Hall v. Zenk, 692 F.3d 793, 799– 801 (7th Cir. 2012), those are generally errors where the harm is both weighty and difficult for defendants to prove, see Bracy v. Schomig, 286 F.3d 406, 428 (7th Cir. 2002) (Rovner, J., concurring in part). Here, despite the seriousness of the po- tential error, Shah and Agarwal are in the best position to 26 Nos. 24-2230 & 24-2236

estimate the value of their own assets that should have been available but for the over-restraint. Shah and Agarwal are also quick to emphasize the gravity of the government’s mistake: a pretrial over-restraint of assets that prevents a defendant from hiring his counsel of choice amounts to structural error. The observation is correct as a le- gal matter, but it does not resolve the question of who bears the burden of showing the Sixth Amendment violation. Be- cause the defendants forfeited their challenge, it follows that, to prevail on a Sixth Amendment claim, they must demon- strate the structural error by showing that the over-restraint deprived them of access to at least the $7.8 million of liquid assets necessary to retain Quinn Emanuel and McGuire- Woods through trial. This follows from the Supreme Court’s plain error precedent and our own. See Olano, 507 U.S. at 734; Greer, 593 U.S. at 507–08; United States v. Navarrete, 88 F.4th 672, 674 (7th Cir. 2023) (observing that “[t]o obtain the benefit of automatic reversal” when a claim of error is unpreserved a “defendant must establish structural error” (emphasis added) (cleaned up)). In short, the district court correctly placed the burden of proving the alleged Sixth Amendment violation on Shah and Agarwal. 3. Application Recall that Shah and Agarwal needed $7.8 million to re- tain Quinn Emanuel and McGuireWoods for trial and that the district court set a June 30, 2020, deadline for retaining trial counsel. The question then is how much money Shah and Agarwal could have secured but for the over-restraint. This analysis is complicated by the fact that the restrained assets contained illiquid investments in private equity funds (and Nos. 24-2230 & 24-2236 27

investments within those funds) and companies held by Gravitas and Jumpstart II. Shah and Agarwal’s expert, Kenneth Mathieu, estimated that the assets would have yielded a $1.739 million distribu- tion and sales proceeds of $6.29 million, for a total value of over $7.8 million. The district court found both estimates in- flated and the latter speculative. We see no clear error in this finding. Start with the distribution. Mathieu’s $1.739 million distri- bution estimate included a $563,030 distribution from 7Wire Ventures Fund, in which Shah and Agarwal’s company, Jumpstart II, was an investor. The district court found that the defendants did not receive that distribution for reasons unre- lated to the protective order. The district court also identified $38,943 in distributions that Shah and Agarwal did receive be- fore the protective order went into effect. In sum, then, the protective order improperly restrained about $1.137 million in distributions. Shah and Agarwal give us no reason to ques- tion this revised estimate. Next, Mathieu acknowledged the lack of a robust second- ary market for the private company investments. He instead estimated that the defendants could have sold the illiquid pri- vate equity interests for $6.29 million, calculated by applying a 10% discount to the capital account balances. The district court found that the estimate rested on a number of doubtful assumptions, as Mathieu assessed the overall secondary mar- ket for private equity interests, not the market for each spe- cific interest at issue. First, the district court doubted Shah and Agarwal could have sold the assets in 11 weeks—the time between its ruling 28 Nos. 24-2230 & 24-2236

on their challenge to the restraint of the Settlement Funds and the June 30, 2020, deadline for retaining trial counsel. The court used the defendants’ prior asset sales as a benchmark. Six weeks before the indictment, Shah and Agarwal sold Gravitas’s interest in another private equity fund. That deal took place in ideal conditions—the fund was “consistently oversubscribed,” meaning it had a large buyer market—and it still took them over six months to complete the sale. And that transaction, Shah acknowledged, had been the only time he managed to sell a fund interest or private security. The illiquid assets here faced far worse sale conditions. Transfer restrictions requiring certain legal approvals would have tightened the potential buyer pool and extended the timeline to sell. On top of that, any sale would likely have taken place early in the COVID-19 pandemic, depressing de- mand and the value of the assets. Due to these added chal- lenges, the district court determined that the interests would take even longer to sell than the “ideal” six-month sale. Of course, the defendants could have sought an extension of the deadline for retaining counsel at that point. But that decision would have been committed to the district court’s discretion. And the court had already granted an extension to obtain counsel from May to the June 30 deadline, commenting that “[n]o further extensions [would] likely be granted.” We see no clear error in the conclusion that the funds would not have been obtainable in time. Separate and apart from the feasibility of any sale, the dis- trict court deemed Mathieu’s valuation overambitious. Mathieu estimated the liquidity value of the over-restrained assets at $6.29 million by applying a 10% discount to the cap- ital account balances at the time of his analysis. But as the Nos. 24-2230 & 24-2236 29

district court observed, the defendants had sold another pri- vate equity interest in 2022 at a huge loss, nearly a 35% dis- count. Given the limited window for sale and the volatile COVID-19 market, a “forced immediate liquidation” of the Other Assets would likely have resulted in a similar discount if the assets could be sold at all. Shah insists the district court overlooked that an asset sale at a 35% or even slightly higher 40% discount would have suf- ficed for him to retain Quinn Emanuel. His calculation ap- pears to proceed on the assumption that he would have re- tained 100% of the sale proceeds. Agarwal, by contrast, insists that they would have split any proceeds 50/50 based on past practice. The district court considered these positions and de- termined that Shah’s decision to share 50% of an attorney re- tainer with Agarwal earlier in the case was entirely discretion- ary. Instead, the more sensible split between Shah and Agarwal would have been 80/20 or 85/15 based on their pro- portional interests in Gravitas and Jumpstart II (a point Shah did not address on appeal). We see no clear error in that find- ing. Regardless, precise calculations are unnecessary. The dis- trict court did not find that the assets would have sold for a maximum discount of 35%, only that it was a far more proba- ble figure than the 10% discount Mathieu used. It further ob- served that Mathieu assumed that the discount of the capital account balance was the relevant sale price for private equity interests, which the studies Mathieu drew on did not address. The district court ultimately concluded that Shah and Agarwal, due to the inflated estimate, did not show by a pre- ponderance that they could have sold their assets in time or for enough money to afford their counsel of choice. Nor have 30 Nos. 24-2230 & 24-2236

they pointed to other assets that would adequately supple- ment a plausible sale of the illiquid assets. In this final deter- mination, we see no clear error and, by extension, no Sixth Amendment violation. III We next turn to Shah and Agarwal’s contention that the government violated their Fifth Amendment due process rights by knowingly making a false statement to the grand jury and failing to correct it. See United States v. Burke, 425 F.3d 400, 412 (7th Cir. 2005) (“The government’s knowing use of false testimony, or failure to correct testimony, violates due process.”). They point to FBI forensic accountant Poelking’s testimony that, to the best of her knowledge, the indictment’s forfeiture allegations “truly and accurately … list items for which there is probable cause that they came from proceeds of the fraud.” Shah and Agarwal assert that Poelking’s testimony amounted to the government falsely telling the grand jury that it traced all the property covered by the protective order to the alleged fraud when, in fact, it had not—an error war- ranting at least a new trial. The district court found this claim untimely for largely the same reasons as the defendants’ Sixth Amendment claim. We agree that the claims fold together. Until their post-trial mo- tions, Shah and Agarwal did not question the veracity of Poel- king’s statement by emphasizing that her tracing analysis supported a forfeiture limited only to the financing fraud, nothing broader. But they were positioned to have done so, as we have explained, by pointing to Poelking’s grand jury testimony. On these facts, raising a Fifth Amendment Nos. 24-2230 & 24-2236 31

challenge for the first time three months after trial and three years after receiving pertinent documents is too late. We may review an untimely challenge to an error in the grand jury proceedings only on a showing of good cause. See Fed. R. Crim. P. 12(b)(3)(A)(v), 12(c); United States v. Acox, 595 F.3d 729, 730–31 (7th Cir. 2010) (explaining that an untimely challenge under Rule 12(b)(3) is only reviewable for plain er- ror where the defendant has shown good cause). Shah and Agarwal did not show good cause for the same reasons that they forfeited the claim. We therefore will not entertain rever- sal on Poelking’s testimony itself. Instead, we focus on Shah and Agarwal’s contention that the government’s knowing failure to correct the misleading statement affected the trial. We review that challenge for plain error. We are skeptical that a Fifth Amendment violation, much less a plain one, occurred. We see no evidence that Poelking or anyone else involved in the prosecution knew she made an inaccurate statement or recognized the over-restraint. The district court examined this question at length, allowing four months of post-trial discovery and holding multiple hearings. The court reviewed internal government communications about the forfeiture allegations and protective order, sworn affidavits from prosecutors and DOJ and FBI personnel in- volved, and permitted defense counsel to question Poelking (four times), as well as DOJ’s Daniel Olinghouse who drafted the forfeiture allegations, and Assistant United States Attor- ney Matthew Madden who served as a point of contact for the private equity entities and companies affected by the protec- tive order. The district court also considered the knowledge of prosecutors who presented the case to the grand jury, in- cluding the prosecutor who examined Poelking. 32 Nos. 24-2230 & 24-2236

After all of that, the district court found that neither Poel- king nor any other government official recognized the over- restraint until after trial. This may appear improbable given the magnitude of assets restrained. But the disconnect stemmed from a right hand-left hand problem: multiple gov- ernment actors took responsibility for different pieces of the restraint—tracing the assets and drafting and monitoring the protective order—without sufficient communication to real- ize that more assets had been restrained than traced or in- tended. The district court, in short, found that the government stumbled but did not engage in willful misconduct or know- ingly present false information to the grand jury. We see no error in the district court’s findings. Shah and Agarwal instead urge us to find a Fifth Amendment due pro- cess violation based on the government’s collective failure to put two and two together, even though they themselves could have caught the error with more diligence. They cite no prec- edent supporting that conception of a due process violation, nor have we found clear support. Any error, then, is not plain. See United States v. Clark, 935 F.3d 558, 571 (7th Cir. 2019) (cit- ing Henderson v. United States, 568 U.S. 266, 279 (2013)). Further, to succeed on plain error, Shah and Agarwal must identify an error that affected their substantial rights—an er- ror that is not harmless. Greer, 593 U.S. at 507–08; see also United States v. Maez, 960 F.3d 949, 961 (7th Cir. 2020). They have not. A petit jury’s conviction renders most grand jury errors, including the use of and failure to correct false or per- jured testimony, harmless unless they are structural. See United States v. Mechanik, 475 U.S. 66, 71 (1986); United States v. Vincent, 416 F.3d 593, 601 (7th Cir. 2005); United States v. Harmon, 833 F.3d 1199, 1203–05 (9th Cir. 2016). The only Nos. 24-2230 & 24-2236 33

structural harm resulting from the government’s failure to correct Poelking’s misstatement that Shah and Agarwal posit is a Sixth Amendment violation, which they failed to estab- lish. By the same token, then, their Fifth Amendment claim fails for lack of prejudice. IV That brings us to Shah and Agarwal’s evidentiary chal- lenges. Recall that former Outcome employees David Ma, Ja- son Ketchum, and Ashik Desai each testified about Shah and Agarwal’s involvement in the fraud during the government’s case-in-chief. On cross-examination, the defendants at- tempted to impeach these witnesses on various grounds such as faulty memory and that favorable (or the prospect of favor- able) treatment by the government motivated their testimony. During the redirect examination of each witness, and over Shah and Agarwal’s objections, the district court granted the government’s motions to admit each witness’s grand jury tes- timony as a prior consistent statement under Federal Rule of Evidence 801(d)(1)(B). Shah and Agarwal now challenge those rulings. We review the district court’s evidentiary rulings for abuse of discretion and underlying factual determinations for clear error. See United States v. Medrano, 83 F.4th 1073, 1076 (7th Cir. 2023). “We give special deference to a district court’s evidentiary rulings and we reverse … only if no reasonable person could take the judge’s view of the matter.” United States v. Pulliam, 973 F.3d 775, 782 (7th Cir. 2020) (cleaned up). We will not reverse an error that is “harmless in light of the trial record as a whole.” Medrano, 83 F.4th at 1076 (cleaned up). 34 Nos. 24-2230 & 24-2236

Rule 801(d)(1)(B) allows the admission of a prior con- sistent statement of a witness subject to cross-examination as non-hearsay in two situations: “(i) to rebut an express or im- plied charge that the declarant recently fabricated [his testi- mony] or acted from a recent improper influence or motive in so testifying; or (ii) to rehabilitate the declarant’s credibility as a witness when attacked on another ground.” A prior con- sistent statement may only be admitted under Rule 801(d)(1)(B)(i) if it precedes “the alleged influence, or motive to fabricate.” Tome v. United States, 513 U.S. 150, 159 (1995). The district court admitted Ma’s, Ketchum’s, and Desai’s grand jury testimony under both Rule 801(d)(1)(B)(i) and (ii). Beginning with Ma, Shah and Agarwal impeached Ma exten- sively on independent grounds justifying the admission of portions of his prior consistent statement to rehabilitate his credibility under Rule 801(d)(1)(B)(ii). See United States v. Be- gay, 116 F.4th 795, 799–802 (8th Cir. 2024) (affirming the intro- duction of a prior consistent statement under Rule 801(d)(1)(B)(ii) even though Tome’s pre-motive rule blocked its admissibility under Rule 801(d)(1)(B)(i)). For instance, on cross-examination they suggested Ma remembered certain events inaccurately and had no contemporaneous notes about a particular interaction with Agarwal that demonstrated her knowledge and approval of fraud within Outcome. Shah and Agarwal insist that the district court neverthe- less abused its discretion in admitting Ma’s grand jury testi- mony as a prior consistent statement. Doing so, they contend, allowed the government to impermissibly bolster Ma’s direct testimony as a cooperating witness. See Fed. R. Evid. 801(d)(1)(B) advisory committee’s note to 2014 amendment (stating that Rule 801(d)(1)(B)(ii) “does not allow Nos. 24-2230 & 24-2236 35

impermissible bolstering of a witness”); United States v. Echols, 104 F.4th 1023, 1027 (7th Cir. 2024) (reiterating the rule against improper bolstering (citing Tome, 513 U.S. at 157–58)); United States v. Bonner, 302 F.3d 776, 780 (7th Cir. 2002) (defining im- proper witness bolstering as “offering evidence solely for the purpose of enhancing a witness’s credibility before that cred- ibility is attacked” (cleaned up)). As a general matter, we agree with Shah and Agarwal that district courts must take care in tailoring prior consistent statements, including from cooperating witnesses. But here, the district court did scope the admissible portions of Ma’s grand jury testimony. Indeed, after much debate, the court agreed with defense counsel that a prior consistent statement is “not meant to bolster a witness’s testimony by simply re- peating what they said on direct exam and was not shaken on cross.” It then directed the government to “discuss with the defense if there are portions of [the statement] that don’t sat- isfy the requirements of the rule.” And then it went through the testimony with the parties. The final statement read into evidence ran about four pages of the transcript. We see no abuse of discretion. We have reservations about the district court’s treatment of Ketchum’s and Desai’s grand jury testimony, however. It admitted each of their testimony in nearly its entirety under Rule 801(d)(1)(B)(i) based on its view that their motives to lie arose after their testimony before the grand jury—Ketchum later received statutory immunity and Desai entered into a plea agreement. See United States v. Nelson, 39 F.3d 705, 709 (7th Cir. 1994) (“Both immunity grants and plea bargains po- tentially give witnesses motives to testify falsely.”). But the record shows that Ketchum already had an immunity 36 Nos. 24-2230 & 24-2236

agreement before he testified in the grand jury. Desai too an- ticipated reaching a cooperation agreement with the govern- ment. All of this sufficed to give Ketchum and Desai motives to lie before testifying in the grand jury. Under Tome, Rule 801(d)(1)(B)(i) was an inappropriate vehicle for admitting the statements. At the very least, the district court should not have admit- ted all of Ketchum’s and Desai’s grand jury testimony. The testimony was substantial (11 and 25 pages for Ketchum and Desai, respectively) and adverse to both defendants. The dis- trict court, at a minimum, should have tailored the testimony as a prior consistent statement. The wholesale admission of prior testimony in circumstances like these amounts to bol- stering that risks undue prejudice to defendants. Even so, the admission of Ketchum’s and Desai’s grand jury statements does not warrant reversal. “[T]he test for harmless error is whether, in the mind of the average juror, the prosecution’s case would have been significantly less per- suasive had the improper evidence been excluded.” Medrano, 83 F.4th at 1077 (cleaned up). Put another way, “an eviden- tiary error is harmless if it did not have a substantial influence on the verdict.” Id. at 1078 (cleaned up). The government’s case against Shah was overwhelming. The jury heard about the fraud (and related money launder- ing) he perpetrated in detail from Outcome’s clients, inves- tors, and other former employees. The government also pre- sented mounds of documents exposing the fraud, including internal emails, text and voice messages, contracts, invoices, and other corporate documents showing that Outcome know- ingly overrepresented its inventory to clients to induce them to purchase larger contacts, underdelivered, and then hid the Nos. 24-2230 & 24-2236 37

performance issues. Shah was at the epicenter of all of this and more, including making false statements to Deloitte during its audits to secure funding. On this record, the district court’s admission of Desai’s and Ketchum’s prior grand jury testi- mony, even if overbroad, was harmless. It is a closer call when it comes to Agarwal, though, as she played a much more limited role in the fraud than Shah. The issue before us is the effect of the admission of Ketchum’s and Desai’s grand jury testimony under Rule 801(d)(1)(B). We nevertheless cannot conclude that allowing the jury to hear the prior testimony substantially influenced the verdict. To her credit, Agarwal acknowledges that Desai said little about her at trial, nor did he say much in his grand jury testi- mony, since he did not interact with her at Outcome as closely as with Shah. Jason Ketchum, by contrast, worked directly with Agarwal and his testimony said much more. As Agarwal sees it, Ketchum supplied nearly the entire case against her. We disagree. What jumps out from our review of the trial transcript is that Ketchum turned out to be a thoroughly inef- fective witness. As the district court emphasized in denying Agarwal’s sufficiency of the evidence challenge, and as Agarwal recognizes on appeal, Ketchum imploded on cross- examination. He contradicted his direct testimony multiple times and, in the district court’s summation, indiscriminately answered “yes” to “nearly every question posed by both the government and defense.” Normally, a prior consistent statement would help reha- bilitate a witness in this situation. But Ketchum became so detrimental that the government jettisoned him as a witness by the end of trial. In closing, it described him as a witness 38 Nos. 24-2230 & 24-2236

who resembled a blindly loyal “circus pony”—able to be steered in any direction by any party—and recommended the jury altogether disregard his testimony. That would include Ketchum’s prior grand jury testimony, on which the govern- ment placed no reliance in the end. Nor did Ketchum’s prior testimony go to the jury as an exhibit, a step further diminish- ing any influence. And the district court, having greater proximity to the ev- idence and witnesses than a court of review, determined that the jury “in all likelihood did not find [Ketchum] credible” because of his persistent flip-flopping. Defense counsel also re-crossed him after he read his grand jury statement, further impeaching his credibility. At bottom, it is hard for us to see Ketchum’s grand jury testimony as adding much, if anything, to the government’s case against Agarwal. On the other side of the ledger, the government presented documents showing Agarwal’s role in the fraud. As early as February 2011, Agarwal knew Outcome was presenting over- stated inventory levels to clients to boost sales. In an email exchange from this time period, Agarwal indicated to Shah that it was inaccurate to state that Outcome had screens in one-third of waiting rooms, as a sales email had represented to a potential client. She also observed that Shah had been making such misrepresentations “a lot.” In response, Shah acknowledged that one-third was the weighted average tar- get for the year and suggested they “get to” it “quickly”—an implicit recognition that Outcome’s inventory projections ex- ceeded actual levels. Further, Agarwal herself directed at least one employee (Ketchum) in an email to conduct a list match using projected inventory, later admitting it was not the “true list.” Evidence Nos. 24-2230 & 24-2236 39

showed that she knew Outcome presented those figures to cli- ents without indicating that they reflected projections, not current inventory. At one point, she confirmed in an email with Shah that the sales team did not know the figure was a projection because “they get confused about how to represent it.” At the same time, Agarwal directed Desai to remove sales- people from emails when discussing “what data to use” be- cause their confidence level when presenting to clients changed depending on whether they believed the data was “accurate vs made-up.” The evidence also showed that Agarwal was aware of Outcome’s persistent “delta” gaps between contracted and delivered services. Stated differently, she knew that the com- pany struggled to deliver on time and yet simultaneously re- ported increased revenue. As the district court observed, “[i]f Outcome was only billing on what was delivered, it would not have met its revenue targets.” Agarwal’s role in manipulating performance metric data, as described by Ketchum, is less clear. But she knew about reports that Desai was manipulating data and yet, around the time of the loans and equity financing in 2016 and 2017, pro- vided information to Deloitte in fraud inquiry meetings and management representation letters without ever apprising the auditors of Outcome’s true financial health. Agarwal invites a much different picture, painting herself as uninvolved in the fraud and, if aware, as someone who pushed against using inflated metrics and advocated for fix- ing the company. But good intentions do not absolve fraud. She may not have originated the scheme, but the evidence showed Agarwal knew of it and still took affirmative steps to further or hide it. 40 Nos. 24-2230 & 24-2236

V Finally, Shah and Agarwal contend that the government argued, and the district court instructions to the jury allowed convictions on, invalid theories of fraud. See Hedgpeth v. Pu- lido, 555 U.S. 57, 58 (2008) (citing Yates v. United States, 354 U.S. 298 (1957)); United States v. Pramaggiore, 178 F.4th 1071, 1077– 78 (7th Cir. 2026). At no point did either defendant object to the jury instruc- tions on this basis, however. See Fed. R. Crim. P. 30(d); Ewing v. 1645 W. Farragut LLC, 90 F.4th 876, 886 (7th Cir. 2024) (“The [jury instruction] objection must be specific enough that the nature of the error is brought into focus.” (cleaned up)). Rec- ognizing the challenge this creates on appeal, Shah and Agarwal backpedal and assert that their general Rule 29 mo- tion for acquittal preserved the objection to the jury instruc- tions. We disagree. A Rule 29 motion preserves challenges to the sufficiency of the evidence. See United States v. Sorensen, 134 F.4th 493, 498–99 (7th Cir. 2025). Nothing suggests it also pre- serves a jury instruction challenge, at least not one that presses legal questions not presented to the district court. Cf. United States v. Rivers, 108 F.4th 973, 985 (7th Cir. 2024) (Kirsch, J., concurring) (explaining that new, purely legal ar- guments fall “outside the narrow bounds of a Rule 29 mo- tion”). To stretch the preservative power of a general Rule 29 mo- tion this far would create “perverse incentives” to neglect timely objections to jury instructions, preventing what might be an easy fix in the district court. Cf. Sorensen, 134 F.4th at 498–99 (quoting Rivers, 108 F.4th at 978 n.1). Nor would it Nos. 24-2230 & 24-2236 41

make sense in light of Rule 30(d)’s more specific guidance about how to preserve a jury instruction challenge. See United States v. Fuertes, 805 F.3d 485, 497 (4th Cir. 2015) (rejecting the position that a Rule 29 motion preserved a jury instruction challenge). Shah and Agarwal point us to United States v. Borrero, where we seemed to entertain a challenge to a jury instruction preserved only by a Rule 29 motion. 771 F.3d 973, 976 (7th Cir. 2014). But no court has read Borrero this way and our prece- dent overwhelmingly weighs against it. We review an unpreserved jury instruction challenge for plain error. See United States v. Leal, 72 F.4th 262, 265 (7th Cir. 2023). “When a jury is instructed on alternative theories of guilt, one of which is legally invalid, and returns a general verdict … a plain error occurs.” United States v. Cardena, 842 F.3d 959, 998 (7th Cir. 2016). The defendant must show harm by demonstrating that “there is a reasonable probability that the outcome of the proceeding would have been different with a proper jury instruction.” Id. at 999; see also United States v. Cook, 970 F.3d 866, 881 (7th Cir. 2020) (discussing the “reasonable probability” standard). The federal fraud statutes at issue here require a “scheme or artifice to defraud, or for obtaining money or property by means of false or fraudulent pretenses, representations, or promises.” 18 U.S.C. §§ 1341, 1343; see 18 U.S.C. § 1344 (simi- lar). A defendant accordingly commits wire fraud when they (1) “devis[e]” or “inten[d] to devise” a scheme (2) to “obtai[n] money or property” (3) “by means of false or fraudulent pre- tenses, representations, or promises.” Kousisis v. United States, 605 U.S. 114, 121 & n.2, 123, 132 (2025) (addressing wire fraud but relying on precedent relating to mail and bank fraud). The 42 Nos. 24-2230 & 24-2236

falsehood must also be material, meaning likely to affect the recipient’s decision to enter the transaction. See id. at 131. The jury instructions conveyed these exact requirements and aligned with our pattern instructions. See United States v. Al-Awadi, 873 F.3d 592, 597 (7th Cir. 2017) (observing that a pattern instruction “is presumed to accurately state the law”). They also included a pattern good-faith instruction: “If a de- fendant acted in good faith, then he or she lacked the intent to defraud required to prove … [fraud] …. A defendant acted in good faith if, at the time, he or she honestly believed the truth- fulness of the representations or promises that the govern- ment has charged as being false or fraudulent.” The William J. Bauer Pattern Criminal Jury Instructions of the Seventh Cir- cuit § 6.10 (2023 ed.). Shah and Agarwal allege the jury instructions permitted the jury to convict on two invalid theories of fraud. First, they contend the instructions allowed for conviction on a “breach- only” theory, meaning that “overpromising and underdeliv- ering” qualified as a “false promise” that satisfied the ele- ments of fraud. In other words, the defendants claim that the instructions allowed the jury to convict them for simple breach of contract regardless of their intent to perform. We do not see it. Indeed, the instructions left no room for this theory because the good-faith instruction protects a defendant who believes the truthfulness of his representations from criminal conviction if his promise simply does not bear out. Second, Shah and Agarwal contend that the jury instruc- tions also allowed for conviction on a “fraudulent induce- ment” theory—specifically, that duping clients into contract- ing with Outcome (by providing misleading list matches) amounted to fraud even if the defendants fully intended to Nos. 24-2230 & 24-2236 43

later make good on the contract. The instructions did allow this inference and properly so. Mirroring the district court’s good faith instruction, the pattern instructions provide that “[a] defendant’s honest and genuine belief that he will be able to perform what he promised is not a defense to fraud if the defendant also knowingly made false and fraudulent repre- sentations.” The William J. Bauer Pattern Criminal Jury In- structions of the Seventh Circuit § 6.10. These instructions reflect a valid theory of fraud. Indeed, while this appeal was pending, the Supreme Court affirmed a conviction based on “fraudulent inducement” whereby a defendant (1) “devise[s]” a “scheme” (2) to induce the victim into a contract to “obtai[n]” her “money or property” (3) “by means of false or fraudulent pretenses.” Kousisis, 605 U.S. at 123. In plain terms, “a defendant commits federal fraud whenever he uses a material misstatement to trick a victim into a contract that requires handing over her money or prop- erty—regardless of whether the fraudster … seeks to cause the victim net pecuniary loss.” Id. at 118. Although the defendant’s subjective intent did not feature in Kousisis, the Court confirmed that “a fraud is complete when the defendant has induced the deprivation of money or property under materially false pretenses.” Id. at 127 n.5. Our precedent is on all fours with these points. As we ex- plained in United States v. Spirk, “[s]omeone who obtains a loan from a bank by [falsely] representing that his income is $1 million a year … has committed fraud even though he sin- cerely believes that he can repay. The lie exposes the bank to a materially greater risk of nonpayment than the lender knew it was accepting, and the creation of this risk by deceit is a crime even if the borrower plans to repay—indeed, even if the 44 Nos. 24-2230 & 24-2236

borrower actually repays.” 503 F.3d 619, 621–22 (7th Cir. 2007); see also United States v. Radziszewski, 474 F.3d 480, 485– 86 (7th Cir. 2007), as amended on denial of reh’g (May 14, 2007); In re Sentinel Mgmt. Grp., Inc., 728 F.3d 660, 668 (7th Cir. 2013) (“[S]omeone who has the best intentions can still pos- sess an actual intent to defraud.”); United States v. Hamilton, 499 F.3d 734, 737 (7th Cir. 2007) (overruling United States v. Bessesen, 445 F.2d 463 (7th Cir. 1971), which did not consider a defendant to have committed fraud where he obtained money through a material misrepresentation but did not in- tend to deprive the victim of the funds even temporarily). Just so here, it does not matter if Shah or Agarwal hoped they could fulfill a contract (by, for example, obtaining the screens promised by the time the ad campaign began) or re- turn the value of the contract over time through a weighted average approach the client did not approve. Intentionally in- ducing clients to part with money through a contract they would not have entered but for the material false promise of certain ad space constitutes fraud. See Kousisis, 605 U.S. at 127 at n.5 (observing that a fraud injury “has occurred when a fraudster ‘obtain[s] from an owner, by a false representation of a fact which he deems material, property which he would not otherwise have parted with upon the terms which he is thus induced to accept’” (quoting Williams v. Kerr, 25 A. 618, 619 (Pa. 1893))); id. at 131 (explaining “materiality”). A final point warrants mention. The government’s theory at trial was not that Shah and Agarwal intended to fulfill the contracts they entered—for instance to obtain the screens falsely advertised to a client by the time an ad campaign be- gan. To the contrary, even if they hoped their growth strategy would play out this way at first, the thrust of the evidence Nos. 24-2230 & 24-2236 45

showed that Shah and Agarwal knew from persistent “del- tas” (under-deliveries) that Outcome could not deliver what was promised on time and yet continued inducing contracts using projected inventory figures and pursued external fi- nancing using false financial information. This is not a case, in short, where the defendants under-delivered once or twice. See United States v. Kelerchian, 937 F.3d 895, 913 (7th Cir. 2019) (“[S]chemes to defraud a party into entering a contract it would not enter if it had been told the truth, but where the fraudsters deliver the agreed money, goods, or services are close to the edge of the reach of the wire and mail fraud stat- utes.”). But after years of under-delivery, the knowledge that they could not timely perform became undeniable and sup- ported a finding of intent to defraud. Even under a narrower jury instruction, then, the outcome of Shah and Agarwal’s trial would likely have been the same. See Cardena, 842 F.3d at 998. *** For these reasons, we AFFIRM.