Not legal advice. This article is educational. Cases are described according to their public posture. Civil settlements resolve allegations without an admission of liability, and complaint-only matters are described as allegations.
The government just changed who it is looking for
For most of the False Claims Act’s modern history, the prototypical whistleblower was an insider: the billing manager who saw the upcoding, the program director who watched costs get shifted to the government, the engineer who knew the part failed inspection. The law rewards that person, the “relator”, with a share of whatever the government recovers.
On April 30, 2026, the Justice Department’s Civil Division announced an initiative that points in a different direction. It is called FOCUS, Fraud Oversight through Careful Use of Statistics, and it is a formal invitation to “data miners”: outside analysts with no insider access who detect fraud by working public government data. According to the Department, data miners have filed a large and growing share of new qui tam complaints in recent years. The government is, in plain terms, recruiting people who can do what a competent forensic analyst already does on a laptop.
There is a second audience for this story, and it is the one that pays our invoices. If you are buying a company that gets paid by the government, a healthcare provider, a defense or government contractor, a federal-grant recipient, the same public-data analysis is now being run against targets like yours, and under the False Claims Act you can inherit a seller’s pre-closing fraud. A whistleblower suit can even sit in a sealed court file straight through your diligence, invisible to a standard process. We take up that defensive angle in a companion piece, Don’t Inherit the Fraud; this article stays on offense.
That is worth sitting with. The skill set this initiative is courting, pulling public datasets, benchmarking a provider or a contractor against its peers, finding the outlier and explaining why it is an outlier, is the everyday work of a due-diligence tech lab. So the obvious question for anyone with those skills is: can you actually do this? Who has done it, what did they do, and what did it take?
This article answers that honestly. The short version: public data is an extraordinary engine for finding leads, and a lead is not a case. The people who have won did the analysis and the much harder work that comes after it. Here is the whole arc.
The numbers that frame the opportunity
The False Claims Act (FCA; 31 U.S.C. §§ 3729 to 3733) makes it illegal to knowingly submit a false claim for federal money. Its teeth are treble (triple) damages plus a penalty for every false claim, and its incentive is a relator’s share of the recovery. The scale is real and climbing: in fiscal year 2025 the Justice Department reported more than \$6.8 billion in False Claims Act settlements and judgments, the highest annual total in the law’s history, of which more than \$5.7 billion involved health care. Whistleblowers filed a record 1,297 qui tam suits that year, and qui tam cases accounted for more than \$5.3 billion of the recoveries.
The table below collects what outside, data-driven relators have actually recovered. Every figure traces to the public source noted in the source log; awards left blank were not publicly reported.
| Matter | What public data exposed | Resolution | Reported relator award |
|---|---|---|---|
| Implantable defibrillators (~500+ hospitals) | Medicare procedure/diagnosis codes showing devices implanted inside Medicare’s prohibited waiting windows | \$250M+ across 70+ settlements (2015) | \$38,227,500 |
| Vascular practice & physician, 2026 (Serrano) | Medicare claims for medically unnecessary vascular procedures, surfaced by a data-analytics relator (Lincoln Analytics) | \$6.73M settlement (May 2026) | ~\$976,000 |
| Empire Roofing network (PPP, the Paycheck Protection Program) | Public SBA (Small Business Administration) loan data showing affiliated entities each certifying under the 500-employee cap | \$9M settlement, N.D. Tex. (2023) | \$1,000,000 |
| YAPP USA (PPP) | Public SBA loan data + ownership records (affiliation / foreign-ownership) | \$14,208,496 settlement | \$1,420,849 |
| BWI-related entities (PPP) | Public loan and ownership data | \$21.6M settlement (2025) | \$2,100,000+ |
| NY skilled-nursing facilities | Statistical analysis of public CMS (Centers for Medicare & Medicaid Services) claims data (therapy-level / length-of-stay outliers) | DOJ (Department of Justice) intervened in 2021; complaint alleged ~\$129M | (not reported) |
Two statutory levers make those awards legible:
| The lever | Figure | Statute |
|---|---|---|
| Damages multiplier | Treble (3×) the government’s loss | 31 U.S.C. § 3729 |
| Civil penalty, per false claim | \$14,308, \$28,619 (2025 adjustment) | § 3729 |
| Relator’s share, government joins | 15%, 25% of the recovery | § 3730(d)(1) |
| Relator’s share, government declines | 25%, 30% of the recovery | § 3730(d)(2) |
The penalty stacks on top of treble damages and applies per claim, which is why a scheme with thousands of invoices can generate enormous exposure even when each individual claim is small.
Who has actually done it
Two outside analysts versus 500 hospitals
The most striking proof that an outsider can do this predates the “data miner” label. Thomas Schuhmann and Leatrice Ford Richards were not employees of any hospital they sued. They were Medicare reimbursement specialists, one a consultant, one a cardiovascular nurse, who understood how hospitals code and bill.
Working from the procedure and diagnosis codes hospitals submit to Medicare, they identified implantable cardioverter defibrillators (ICDs) placed inside the waiting periods that Medicare’s National Coverage Determination prohibits, too soon after a heart attack, bypass, or angioplasty. They did not need a single hospital’s internal files; the pattern was legible in the claims data itself, and it repeated across the country. The Justice Department pursued it as a nationwide sweep. By 2015, more than 500 hospitals had settled for over \$250 million, and the two relators were reportedly awarded \$38,227,500. It remains the clearest demonstration in the set that fluency in a billing rule, applied to public claim codes, can scale into a historic recovery.
A data firm reads the SBA’s own loan file
Fast-forward to the pandemic-relief era, where the model gets cleaner and more modern. Sidesolve, Inc., an analytics firm with no insider knowledge, worked the public Paycheck Protection Program loan data the SBA itself published. The Small Business Administration’s affiliation rules require commonly owned businesses to count all of their affiliates’ employees toward the 500-employee eligibility cap. Sidesolve’s analysis surfaced a roofing company and eight affiliated entities that had each certified under the cap separately, collectively well over it.
The matter settled for \$9 million in late 2023 in the Northern District of Texas, and the data firm received \$1 million as the relator. One defendant, one clean public dataset, one affiliation rule, a seven-figure award, it is the tightest single illustration of the whole idea.
The serial data-relator
GNGH2, Inc., the vehicle of a New York lawyer named David Abrams, turned the approach into a practice, filing many PPP cases nationwide built on public SBA loan data and public ownership records, targeting affiliation-rule violations and undisclosed foreign ownership. Public settlement announcements include YAPP USA at \$14,208,496 (a reported relator share of \$1,420,849), Horn USA at roughly \$4.15 million, and a \$21.6 million resolution involving BWI-related entities in 2025. The lesson is not that one case is a lottery ticket; it is that a repeatable public-data method, run as a disciplined pipeline, compounds.
The firm that lost twice, then made history
No honest account of this skips Integra Med Analytics, because its story is the whole thesis in miniature. Integra was built for exactly this purpose: a data-forensics firm that ran statistical analyses against public CMS Medicare claims data to find hospitals coding high-value secondary diagnoses far more often than their peers, then filed qui tam suits alleging upcoding.
It lost, twice, publicly. Its case against Baylor Scott & White was dismissed, and the Fifth Circuit affirmed in 2020; its case against Providence Health was ordered dismissed by the Ninth Circuit in 2021. We will come back to why in the next section, because the reasoning is the single most useful thing in this article.
And then the same firm, using the same kind of public-data analysis, found a different target, a group of New York skilled-nursing facilities, and in June 2021 the Justice Department intervened, joining a complaint that alleged roughly \$129 million in fraud. It was reported as the first time the government had stepped into a case originated purely by a data-analytics firm. Rejection, refinement, and then validation from the government itself. That arc, not any single payout, is the realistic shape of the work.
And the healthcare model keeps producing. In May 2026, the data-analytics relator Lincoln Analytics resolved allegations that a California vascular practice billed Medicare for medically unnecessary procedures for more than \$6.73 million, taking roughly \$976,000 as its relator share, an outside analytics shop, a Medicare-billing pattern, a multimillion-dollar healthcare recovery.
The reality check: why data alone usually loses
Here is the part most “you can get rich as a whistleblower” pitches leave out, and the part that makes a forensic firm’s voice worth trusting.
When the Integra courts threw out the upcoding cases, they did not say statistics are inadmissible. They said something more precise and more important: a statistical outlier is, by itself, equally consistent with fraud and with simply being good at lawful billing. A hospital that codes more high-severity diagnoses than its peers might be cheating, or it might be an early, accurate adopter of coding guidance that CMS itself encouraged. Under the pleading standard the Supreme Court set in Twombly and Iqbal, when there is an “obvious innocent explanation,” the relator has to plead facts that tend to exclude it. Numbers that merely raise a suspicion are not enough. (Notably, neither Integra ruling was published as binding precedent, but both are widely followed, and the Justice Department now cites the same logic in its own FOCUS guidance.)
So what does survive? The instructive contrast is Customs Fraud Investigations, LLC v. Victaulic, where a data-driven complaint cleared the bar, because the relator paired the statistics with concrete, particular evidence: import records, photographs of specific unmarked products, 221 researched product listings, and an expert declaration. The data was the connective tissue; hard, specific facts of an actual false claim were the case. Even then it was a divided decision, with a dissent warning courts not to be “fooled by the numbers.” The signal is consistent: statistics get you in the door; particularized facts win.
Three more rules shape whether an outsider’s case is viable at all:
- The public-disclosure bar (31 U.S.C. § 3730(e)(4)). If the fraud has already been publicly disclosed, your case can be barred, unless you qualify as an “original source” whose independent analysis “materially adds” to what was public. For a data miner working public files, planning to qualify as an original source is not optional; it is the strategy. It generally means making a documented, voluntary disclosure of your analysis to the government before you file.
- First to file (31 U.S.C. § 3730(b)(5)). Only the first relator on a given fraud recovers. Public-data anomalies are reproducible by definition, so two analysts can independently find the same outlier, and the second one home gets nothing. Speed and confidentiality are competitive assets.
- The knowledge element (“scienter”). The Act reaches a defendant who acted “knowingly”, actual knowledge, deliberate ignorance, or reckless disregard. Honest billing disputes and good-faith reading of an ambiguous rule are not fraud. Your evidence has to speak to a knowing falsehood, not just an unusual number.
The Justice Department’s FOCUS initiative points the same way: it says it will prioritize data miners who show “pre-filing diligence,” “familiarity with program rules,” and “legally sufficient allegations.” Translation, the data finds the lead; particular facts make the case.
The whole life cycle: what it actually takes
If you clear that bar and have something real, here is the path from suspicion to settlement. Knowing the shape of it is the difference between a hobby and a plan.
- Detection and evidence. You move from outlier to evidence, a specific false claim, tied to a specific program rule, with something that speaks to the defendant’s knowledge. Gather only what you can lawfully access; improperly taken records can sink a case.
- Counsel. You effectively cannot do this alone. A qui tam suit is brought on the government’s behalf, and courts will not let a relator litigate it pro se. Reputable whistleblower firms work on contingency, no upfront cost, and they vet hard, because a case can run for years. Choosing the right firm is part of the work.
- Filing under seal. The complaint is filed in camera and stays sealed for at least 60 days; the defendant is not told (31 U.S.C. § 3730(b)(2)). Simultaneously, you serve the government with a written “disclosure of substantially all material evidence.” That disclosure, not the complaint, is usually the most labor-intensive and most persuasive document in the matter. (More on who builds it below.)
- The seal period. The 60 days is a formality; in practice the seal is extended, often for one to three years, while the government investigates, issuing civil investigative demands, interviewing witnesses, quantifying damages. You keep the entire matter secret the whole time, frequently while still working alongside the target.
- The intervention decision. The government either intervenes (joins, and brings its resources and credibility) or declines (you may proceed alone). Intervention is the single biggest predictor of recovery; declination is not a verdict on the merits, but declined cases are far harder to win.
- Resolution. Most cases settle rather than go to trial, defendants face treble damages, per-claim penalties, and collateral consequences like exclusion from federal programs. The relator’s counsel advocates for the relator’s share in that negotiation.
- The relator’s share. If the government joins, you receive 15 to 25% of the recovery; if it declines and you litigate alone, 25 to 30% (31 U.S.C. § 3730(d)). A case resting mainly on already-public disclosures is capped lower, and a relator who helped plan the fraud can be cut with no floor. You also recover reasonable attorneys’ fees and costs from the defendant, on top of the share.
- Timeline and odds. Plan for three to seven years. The government intervenes in only a minority of qui tam cases, practitioners often estimate roughly one in five, yet those cases account for the overwhelming majority of dollars recovered. This is an endurance test, not a quick hit.
- Protection. The Act’s anti-retaliation provision (31 U.S.C. § 3730(h)) entitles an employee fired or harassed for lawful whistleblowing to reinstatement, double back pay, and special damages, separate from, and available even without, a successful qui tam recovery.
Where the forensic skill set earns its keep
Notice where the leverage actually sits in that life cycle. It is not the anomaly score. It is Stage 1 (turning a number into provable evidence of a knowing falsehood) and Stage 3 (the written disclosure that makes the government care). That is forensic-accounting work, and it is exactly the discipline a Certified Fraud Examiner is trained for.
The framing here is the ACFE’s Report to the Nations (the Association of Certified Fraud Examiners), which makes a point that should reframe how anyone thinks about detection: tips are the number-one way occupational fraud is caught, roughly three times more often than the next method. The whistleblower is not a sideshow to fraud detection; the whistleblower is the detection mechanism. Data analytics does not replace that. It aims it.
The ACFE’s Fraud Tree, the taxonomy of asset misappropriation, corruption, and financial-statement fraud, tells a data miner what kind of scheme an outlier likely is and therefore what evidence proves it. Corruption schemes (kickbacks, conflicts of interest) map directly onto the kickback-tainted claims that drive a large share of health-care FCA cases. And the Fraud Triangle (pressure, opportunity, rationalization) supplies the narrative that makes the “knowing” element credible to a prosecutor. A CFE’s contribution to a qui tam case is concrete: reconstruct the claims, quantify the treble-damages-and-penalties exposure, document the red flags, and assemble the §3730(b) disclosure into something a line attorney at the Justice Department will act on. Done well, that contribution is also what moves the relator’s share toward the top of the statutory band, because the government weighs how substantially the relator helped.
So, can you do this yourself?
Honestly: the front half, yes. The back half is the job.
The lead-generation engine is genuinely reproducible. The data is free and official, CMS Medicare physician and Part D files, Open Payments, the OIG (the HHS, Department of Health and Human Services, Office of Inspector General) exclusions list, USAspending, SAM.gov (the System for Award Management, the federal contractor registration and exclusion system). The methods are published in the academic literature. And it is precisely the work our due-diligence tech lab already does: our walkthrough on screening a state’s Medicaid market with public data builds the same kind of peer-relative outlier list, with the same free APIs and the same allegation-vs-finding discipline, that a data-miner relator starts from. The same instinct underlies our franchise case studies: the public record usually already contains the pattern; the skill is knowing where to look and how to test what you find.
But every honest practitioner, and the government’s own declination rates, will tell you the last mile is where the case lives. An outlier is a question, not an answer. Turning it into a recovery takes the proof of a knowing false claim, an original-source disclosure done right, a whistleblower firm willing to carry years of contingency risk, and the patience to let it run. The numbers find the lead. Forensic work and counsel make the case.
That is the grassroots reality, and it is genuinely encouraging: the barrier to finding government fraud has collapsed, and the government is openly asking skilled outsiders to help. The barrier to proving it is still high, which is exactly why the people who can do both are about to be very valuable.
Two companion pieces build on this. A step-by-step Medicare outlier screen on public data, with code you can run, stands the front half up in the open. And Don’t Inherit the Fraud turns the same screen around to the deal table, what an FCA recovery in a target’s history should tell a buyer or a lender before they sign.
Primary Sources
- U.S. Department of Justice, Civil Division announces FOCUS initiative for data-miner qui tam complaints (2026)
- U.S. Department of Justice, False Claims Act recoveries exceed \$6.8B in FY2025 (record)
- U.S. Department of Justice, Nearly 500 hospitals pay over \$250M to resolve ICD allegations
- U.S. Department of Justice, Vascular practice pays more than \$6.73M; Lincoln Analytics as relator
- U.S. Attorney’s Office, N.D. Texas, National roofing company settles PPP allegations for \$9M
- Crowell & Moring, Recent PPP False Claims Act settlements (YAPP, Horn, BWI)
- Arnall Golden Gregory, Fifth Circuit affirms dismissal in Integra v. Baylor Scott & White
- 31 U.S.C. § 3730, Civil actions for false claims (qui tam, seal, intervention, relator share, anti-retaliation)
- 31 U.S.C. § 3729, False claims liability, treble damages, scienter
- Sidley, False Claims Act civil monetary penalty figures
- ACFE, Occupational Fraud 2026: A Report to the Nations, key findings
Prepared by Noah Green CPA CFE. Sheepdog Prosperity Partners provides financial due diligence and analytics-enabled transaction support. This article is educational and is not legal advice.