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

*You will grade each founder claim as confirmed, unconfirmed, or contradicted against an independent source and decide whether the background clears the bar to close.*

- Canonical URL: https://www.refolk.ai/guides/founder-claim-verification-standard
- Pillar: Investing and deal sourcing
- Format: Standard
- Published: 2026-08-10
- Last reviewed: 2026-08-10
- Reading time: 15 min

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.

> **Rule:** One statement, one source, one grade
>
> Every material claim must be rewritten as a single testable statement tied to a named independent source, then graded confirmed, unconfirmed, or contradicted. A claim with no independent source cannot be graded confirmed.

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.

**46% - Resumes with at least one discrepancy versus a background check**

Employment-verification discrepancies rose 44% between 2021 and 2024, so the base rate is not falling.

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.

> Degrees and titles have registries; a claimed user count often does not, which is exactly why it is the highest-leverage lie.

## 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.

| Claim | Authoritative source | Documented cost or limit |
|---|---|---|
| Degree | Clearinghouse DegreeVerify | 80M+ records |
| Federal legal | PACER | $0.10/page, $3 doc cap |
| Name search | PACER Service Center | $30 per name |
| Finance / adviser | IAPD / BrokerCheck | updated within 30 days |

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.

> **Watch out:** A clean federal search is not a clean record
>
> PACER covers federal courts only. State and county cases, and settled matters, are invisible to it. Run state and county searches alongside PACER before grading legal history as clean.

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.

#### Verification pipeline, screen to close

1. **Inventory** - Write each material claim as a testable statement with a named source
2. **Open-source screen** - Grade every claim on public data alone
3. **Formal verification** - Match degrees, titles, dates, and legal records to issuer sources
4. **Independent checks** - Verify traction, technical originality, and off-list references
5. **Adjudicate** - Apply materiality plus intent and write the decision

*Each stage produces graded claims; the pipeline stops when every material claim is graded and adjudicated.*

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.

| Market | "Due Diligence" / "Background Investigator" profiles |
|---|---|
| United States | 2,545 |
| United Kingdom | 12 |

## 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.

#### From claim inventory to a wire-ready decision

1. **Inventory material claims** - List every checkable factual claim from the deck, CV, and data room. Done means each is written as a testable statement with a named independent source. Analyst, about one hour, at initial screen.
2. **Run the open-source screen** - Cross-check LinkedIn, company registries, and press, contacting previous employers where useful. Done means each claim is graded confirmed, unconfirmed, or contradicted on public data alone. Analyst, one to two hours.
3. **Verify credentials formally** - Confirm degrees via Clearinghouse DegreeVerify or the registrar and licences via the issuer database. Done means each degree, title, and date is matched to an issuer record. Analyst or screening firm, one to five business days.
4. **Search legal and regulatory records** - Run PACER for federal cases plus state and county searches, and IAPD or BrokerCheck for finance history. Done means clean or documented hits with case numbers. Counsel or firm, one to three days.
5. **Verify prior exit and traction** - Independently confirm claimed exit value, acquirer, and the founder's actual role, and check live metrics against a party the founder does not control. Done means the exit or metric is confirmed externally before you wire. Deal lead, one to three days.
6. **Review technical originality** - Have an independent expert audit code and IP assignment and grade the originality claim. Done means IP ownership is confirmed and the claim is graded. Independent expert, two to five days.
7. **Run backdoor reference checks** - Contact ex-colleagues not on the founder's list, since named references are usually coached. Done means at least one off-list reference per material role. Deal lead, ongoing.
8. **Adjudicate and close** - Apply the materiality-plus-intent test in a partnership meeting. Done means every material claim is graded and there is a written decision to clear, remediate, or pass. Partnership, one meeting.

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.

I ran this search: `Former engineers and colleagues who worked with startup founders at prior companies between 2018 and 2021` - [see the full result list](https://www.refolk.ai/s/g5dqx8dwxq).

*Returns people who overlapped with the founder at a named prior employer, so you can arrange a reference the founder did not select or coach.*

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.

#### Grading a discrepancy

Horizontal axis runs from Immaterial to Material. Vertical axis runs from Honest error to Fabricated.

| Quadrant | What it means |
| --- | --- |
| Honest, immaterial | Note and move on |
| Honest, material | Remediate: correct the record and reprice if needed |
| Fabricated, immaterial | Hard stop: fabrication predicts more fabrication |
| Fabricated, material | Hard stop and walk: this is the Frank pattern |

*Intent and materiality together decide whether a discrepancy is fixable or a deal-killer.*

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.

> **Tip:** The engineer is your best independent witness
>
> In the Frank case, the chief of engineering refused a request to generate synthetic user data. On any technical claim, the person closest to the build is the source least able to launder a fabrication - reach them off-list.

## 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.

| Title | Count | Share of pool |
|---|---|---|
| Founder | 424,069 | 68.4% |
| Co-Founder | 195,865 | 31.6% |

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.

**Per-claim grading row**

```
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
```

*One row per material claim; the file is wire-ready when every row is filled and the decision line is written.*

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.

#### Wire-ready verification checklist

- [ ] Every material claim is written as a testable statement with a named independent source.
- [ ] Each degree, title, and employment date is matched to an issuer record, not a founder-supplied document.
- [ ] PACER plus state and county searches are run, and any hits are documented with case numbers.
- [ ] Finance history is checked in IAPD or BrokerCheck where the role warrants it.
- [ ] The prior-exit outcome and any live traction metric are confirmed against a party the founder does not control.
- [ ] IP ownership is confirmed and the technical-originality claim is graded by an independent expert.
- [ ] At least one off-list reference per material role has been contacted.
- [ ] Every claim carries a grade of confirmed, unconfirmed, or contradicted, and no unconfirmed claim is treated as a pass.
- [ ] Each discrepancy is adjudicated on materiality plus intent, and any fabricated evidence is treated as a hard stop.
- [ ] A written decision to clear, remediate, or pass is recorded.

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.

## Frequently asked questions

### How do I verify a founder's degree before investing?

Verify degrees against the issuer, not a document. Use National Student Clearinghouse DegreeVerify, which draws on more than 80 million current and former student records, or contact the awarding registrar directly. Never accept a transcript or diploma PDF as proof, because fraudulent documents are easy to produce. A polished transcript is a false positive; only an issuer-matched record counts as confirmed.

### When has a founder's background been verified enough to wire?

When every material claim is graded confirmed, unconfirmed, or contradicted against an independent source, and the partnership has applied the materiality-plus-intent test in writing. There is no published universal bright line; the test is whether the fact, if known, would have changed the decision, and whether the discrepancy was honest or fabricated. Any fabricated evidence is a hard stop regardless of size.

### How do I confirm a founder's prior exit or traction claims?

Confirm them against a party the founder does not control: the acquirer, the payment processor, the platform, or early employees who were there. Self-reported spreadsheets and internally generated user counts are not confirmation. In the Frank case a claimed 4.25 million users was really under 300,000, about 7%, and it survived until the acquirer tried to contact the customer base after closing.

### Does a clean PACER search mean a founder has no legal history?

No. PACER covers federal courts only, so state and county matters and settled cases are invisible to it. Run state and county searches alongside PACER, and for finance backgrounds check IAPD and BrokerCheck. Treat the absence of a hit as unconfirmed, not confirmed, since not every bad actor is caught and one entry matters less than a pattern of disputes.

### How do I check a sole-founder claim?

Treat it as a checkable population claim. Co-founder-titled profiles are 31.6% of the founder pool in Refolk's index, so a solo claim is common but not the default. Cross-check incorporation filings and cap-table records for silent or departed co-founders, and search who else was listed at the company at formation. A silent co-founder with equity is a material discovery.

### Should angel investors run the same verification as funds?

Yes, and they need the written rubric more. Angels rely on personal networks and informal checks while funds use professional firms, and diligence-professional supply is thin, with 2,545 US background and due-diligence profiles in Refolk's index. The registries angels need are low-cost or free, so a written per-claim standard closes the gap without a firm's budget.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/guides/founder-claim-verification-standard*