The Founder Claim Verification Standard: When a Track Record Checks Out

Before you wire, you have to confirm that what a founder claims about their background is actually true - and know when you have checked enough to stop. This guide is for early-stage investors, platform and talent partners, and angels who need to grade a founder's factual claims to a wire-ready bar. It gives you a per-claim rubric mapping each claim type to the independent source that confirms it and the threshold that separates a fixable discrepancy from a deal-killer, so two partners grade the same founder the same way.

This is a standard, not a background-check pitch. The output is a decision: every material claim graded confirmed, unconfirmed, or contradicted, and a written call to clear, remediate, or pass.

What counts as a material founder claim

A material claim is any factual statement about the founder's background that, if false, would change your decision to invest. That is the working definition of materiality: had the fact been known, the decision would have been different.

Not every line on a deck is material. The founder's favourite framework is not checkable; the founder's degree, titles, dates, prior exit outcome, sole-founder status, technical originality, and legal history are. Those seven are the spine of this standard because each one has an independent source that can confirm or contradict it, and each has a documented pattern of inflation.

The grades are deliberately three, not two. Confirmed means an independent source matches the claim. Contradicted means an independent source disagrees with it. Unconfirmed means you could not source it either way - and unconfirmed is not a pass. The most common grading error is reading "no evidence against" as confirmation. It is not.

The scale of the population you are grading against matters for one claim in particular. In Refolk's index of professional profiles, US profiles carrying a "Founder" title outnumber "Co-Founder" profiles 2.17 to 1, and co-founder-titled profiles make up 31.6% of the combined pool. A sole-founder claim is common, but it is not the default, so it is a claim to check rather than assume.

Which claims lie most, and what each lie proves

Weight your effort toward the claim types with the highest discrepancy rates: education first, employment history and titles second, then licences for regulated roles and legal history for sensitive ones. Screening data is blunt about how often these are wrong.

Thirty-three percent of applicants admit to misrepresenting their education, including incomplete degrees, inflated GPAs, and false institutions. Yahoo's CEO resigned after four months when it was discovered he had lied about his education. The lesson is not that education fraud is rare - it is that it is the single most inflated claim and the easiest to verify against a registry, so a missed degree is inexcusable.

Employment departures are the quiet one. Candidates dismissed for performance or conduct almost universally describe the exit as voluntary, mutual, strategic, or a pursuit of new opportunities. The claim "I left to start this company" is not checkable from the founder alone; the reason for leaving is confirmed only by an off-list source.

The startup-specific fraud sits above all of these: inflated traction and prior-outcome metrics. Charlie Javice claimed Frank had 4.25 million users; the real number was under 300,000, about 7% of the claim. JPMorgan acquired Frank for $175M in 2021, and Javice was later sentenced to 85 months. The metric survived diligence because it came from a source the founder controlled, and only collapsed when the acquirer tried to contact the customer base after closing.

The source that confirms each claim

Match every claim to its authoritative source before you start, because the source determines whether a "yes" means anything. A degree confirmed from a PDF the founder supplied is not confirmed; a degree matched to an issuer record is.

The table below is the core of the standard. It maps the claim types with real registries to the source that owns the truth and what that source costs or covers.

Two properties of these sources decide how you read a result. First, coverage boundaries: PACER indexes federal courts and more than a billion filed documents, but state and county matters and settled cases sit outside it entirely. A clean PACER search proves the absence of a federal case, nothing more. Second, freshness: registered finance professionals must update Central Registration Depository information within 30 days, and it appears in BrokerCheck the next business day, so BrokerCheck is close to current. DegreeVerify draws on more than 80 million student records, which is broad but not universal - a gap there means run the registrar directly, not that the degree is false.

For finance backgrounds, IAPD and BrokerCheck report employment history, professional qualifications, disciplinary actions, criminal convictions, civil judgments, and arbitration awards drawn from the Central Registration Depository. That makes them unusually rich, but the same caution applies: an absence of entries is unconfirmed innocence, not proof of it.

Who does the work, and in what order

Split the work by who can defensibly do it: the investment team runs initial online research and informal reference checks, legal counsel handles formal background investigations and compliance, and an external firm may be engaged for comprehensive screening on larger rounds. Screening intensity varies by stage - angels lean on personal networks while institutional funds employ professional firms.

Timing is where sources genuinely disagree, so state it plainly rather than pretend there is consensus. Most place deep verification post-term-sheet, angels often defer legal and regulatory searches, and funds run them earlier. The order of the checks is stable; the trigger point is a fund policy choice. Set it once and apply it to every deal so grading does not drift with the deal lead's mood.

The supply of people who do this professionally is thin, which is why the written rubric matters more than the hire. In Refolk's index, exact-title matching returns 2,545 US profiles carrying "Background Investigator" or "Due Diligence" against 12 in the UK. Treat the international counts as directional only - exact-title matching materially undercounts non-US title conventions - but the direction is real: this is not a deep talent pool, so most funds run the standard themselves and reach for a firm only on the biggest rounds.

The verification procedure

Run these eight steps in order. Each has a named owner, a rough duration, and a definition of done, so a partner reviewing the file can see exactly where a claim stands.

Cost is not the reason to skip any of this. PACER is $0.10 per page with a $3 cap per document and a $30-per-name search at the service centre, and Clearinghouse and BrokerCheck are low- or no-cost. A missed federal case or an unverified degree reflects a process failure, not a budget one.

The hardest steps to staff are the off-list references and the independent technical review, because both require reaching people the founder did not hand you. That is a sourcing problem, and it is where a search tool earns its place.

Naming the prior company and the window turns "find me a backdoor reference" into a specific search across public LinkedIn and the open web. Refolk resolves that to reachable people in plain English, which is the friction the reference step otherwise imposes.

How this goes wrong: false positives and deal-killers

The failure modes below are the most valuable part of the standard, because each is a way a claim gets graded confirmed when it should not be. For every check, know what a false positive looks like.

The materiality-plus-intent test is what turns a list of discrepancies into a decision. Honest misrepresentation in the absence of fraud typically caps founder liability at the value of their stock; fabrication becomes criminal. So the size of the error is not the axis that matters - intent is. A small fabricated fact belongs in the same column as a large one.

Here are the specific traps, each with the false positive that fools an analyst and the fix.

• Degree "verified" from a document. The false positive is a polished PDF transcript. Fraudulent documents are easy to produce; confirm against the issuer or Clearinghouse, never against paper the founder supplied.
• Clean PACER read as a clean record. State and county cases and settled matters are invisible in federal PACER. Run state and county searches before grading legal history.
• "Amicable departure" from a prior role. Dismissed founders almost universally frame exits as voluntary or strategic. Confirm the reason for leaving with an off-list source, not the founder.
• Named references all glow. Three warm calls the founder arranged is a false positive; named references are usually coached. Require backdoor references.
• Traction self-reported. A spreadsheet that "confirms" the number is the Frank pattern. Verify metrics against the platform, payment processor, or a third party.
• Absence of a hit read as innocence. Not every bad actor is caught. Grade a missing disciplinary entry as unconfirmed, not confirmed.
• A single old disclosure over-weighted. Patterns - multiple disputes or sequential regulatory actions - are far more telling than one entry. Distinguish one old item from a pattern before you pass.
• IP originality assumed from a demo. Generative tools can fabricate interfaces and simulations. Require code and IP-assignment review by an independent expert.

The technical-originality trap deserves emphasis because it is now spoofable by default. Nearly 60% of deals fall through over problems found in technical review, and technical issues can cut a valuation by up to 20%. A slick demo proves nothing about ownership. The only reliable originality tests are an IP-assignment audit and an independent expert review - and if founders or contractors have not assigned all code to the company, that is non-negotiable and investors walk away.

Sizing the sole-founder and traction claims

Two claims resist the registry approach and need their own method: sole-founder status and traction. Treat both as claims to be sourced against a party the founder does not control.

Sole-founder status is a checkable population claim. With co-founder-titled profiles at 31.6% of the founder pool in Refolk's index, a solo claim is common but not the default, so cross-check incorporation filings and cap-table records for a silent or departed co-founder. Search who else was listed at the company at formation. A co-founder written out of the deck is material because it changes both the equity story and the founder's account of their own history.

Traction is the claim that put Frank's founder in prison, so verify it against the platform, the payment processor, or people who were there before you wire. Early employees at the claimed prior startup can confirm both the outcome and the founder's actual role in it. This is the step most likely to be skipped under time pressure and the one most likely to be catastrophic if it is.

Claim (testable statement): ____________________
Claim type: degree | title/dates | prior exit | sole-founder | technical/IP | legal
Independent source used: ____________________
Grade: confirmed | unconfirmed | contradicted
If unconfirmed: what I would check next: ____________________
Materiality: would this fact, if false, change the decision? yes | no
Intent read (if discrepancy): honest error | fabrication
Adjudication: clear | remediate | pass

Post-FTX, VCs prioritise founders' backgrounds and reputational checks have become a cornerstone of diligence. That shift is a reason to standardise, not to run every check ad hoc. A rubric that produces the same grade from two partners is worth more than a firm engaged inconsistently.

The close checklist and keeping it current

The background clears the bar when every material claim is graded against an independent source and the partnership has adjudicated each discrepancy on materiality and intent in writing. Use this before you call it done.

To keep the standard current, re-check the mechanisms, not the values. BrokerCheck reflects Central Registration Depository updates within 30 days plus a business day, so a finance check older than about a month is stale and worth refreshing before wiring. Registry coverage changes: DegreeVerify's record count and DiplomaVerify's school coverage grow, so if a check comes back with no match, confirm the source covers that institution before concluding the degree is false. PACER's fee schedule is published and occasionally revised, so treat the per-page and per-name figures here as the mechanism, and pull the current numbers when you budget a large search.

The one thing that does not change is the grading discipline. Three grades, one source per claim, and the materiality-plus-intent test at the end. Adopt those as team policy and two partners will grade the same founder the same way, which is the entire point of a standard.