How Investors Verify Startup Claims Before Writing a Cheque

Investors verify startup claims in two passes. Before a first call they run quick desk research on the team, market size, traction and competitors, looking for anything implausible. After a term sheet they run due diligence: reference calls, bank statements, contracts and legal checks. Claims that trace to an original, dated source survive both. Claims that don't usually end the conversation.

The extreme case shows what investors are really doing. JPMorgan Chase agreed to pay $175 million for Frank, a student financial aid start-up whose founder began looking for a buyer in 2021. During the process, the bank sought to verify how many users Frank really had, and treated that number as critical. Frank had told it, again and again, that it had 4.25 million.

It had about 300,000. According to the US Department of Justice, the founder, Charlie Javice, hired an outside data scientist to create a synthetic data set to show the bank. In September 2025 she was sentenced to 85 months in prison.

Users Frank told the bank it had 4.25 million

Users Frank actually had About 300,000

Frank's users: what JPMorgan was told and what existed

Users Frank told the bank it had	4.25 million
Users Frank actually had	About 300,000

Source: US Attorney's Office, Southern District of New York, sentencing press release, 29 September 2025. Chart by ThriveFinity.

That's a criminal case. Almost every founder who overstates a number is doing something far more ordinary: rounding up, quoting an old figure, or repeating a market size from someone else's slide. But the pattern is the same. Investors pick the one or two claims the whole deal rests on, and they try to confirm them from a source the founder doesn't control.

I wrote this for founders preparing to raise, and for the angels and advisers who back them. It covers:

• what investors check before a first call and after a term sheet
• the sources they use
• the ways decks tend to fail
• how to run the same checks on yourself first

The Desk Research Window

Desk research is the quick plausibility check an investor runs before deciding whether to meet you. It isn't due diligence. At this stage the investor is looking for a reason to say no quickly, because they see far more companies than they can ever back.

The best numbers on that funnel come from the largest survey of venture capitalists I know of. Paul Gompers, Will Gornall, Steven Kaplan and Ilya Strebulaev surveyed 885 institutional VCs at 681 firms for their paper How Do Venture Capitalists Make Decisions?, published as an NBER working paper in 2016 and in the Journal of Financial Economics in 2020. For every deal a firm closed, it had looked at roughly 100 opportunities.

Considered 101

Met management 28

Reviewed at a partners meeting 10

Went into due diligence 4.8

Offered a term sheet 1.7

Closed 1

Opportunities at each stage, for every one deal a VC firm closes

Considered	101
Met management	28
Reviewed at a partners meeting	10
Went into due diligence	4.8
Offered a term sheet	1.7
Closed	1

Source: Gompers, Gornall, Kaplan and Strebulaev, How Do Venture Capitalists Make Decisions? (NBER), Table 6, survey run November 2015 to March 2016. Averages across 442 responses, mostly US firms. Chart by ThriveFinity.

Look at the first two bars. Roughly three in four opportunities never got a meeting at all. That's where the desk research happens, and it's fast. The later stages are just as selective. The paper describes:

• one in four opportunities leading to a meeting
• a third of those reviewed at a partners meeting
• roughly half of those going into due diligence
• about a third of those getting a term sheet

Each step is a fresh round of checking, by more people, with more time.

How fast is the first step? Dropbox, which owns the DocSend deck-sharing tool, says in its pitch deck research that investors now spend less than 3 minutes reviewing a pitch deck. That's vendor data, so I'd treat the exact figure loosely. The point stands. Nobody checks 20 slides in three minutes. They check the few claims the story rests on, then decide whether the rest is worth an hour.

Investors usually don't explain a pass. A founder can send the same deck to 40 investors, get 40 polite declines, and never learn that slide four quoted a market size nobody could trace.

The Four Claims Investors Check First

Before a first call, investors usually check four kinds of claim: market size, traction, unit economics and competition. They're the claims that are both important and checkable from outside the company, which makes them the natural place to start. This is my reading of how the checks work, assembled from the sources below, not a published protocol.

Market size gets traced first

Market size is usually the first number to be traced, because it's the easiest. It's one figure, it nearly always cites a source, and that source is usually a press release from a research firm. An investor can open it in seconds and see whether it really measures your market.

Often it doesn't. A global AI market figure on a slide about AI tools for UK accountants tells the investor almost nothing, except that the founder reached for the biggest number available. Underscore VC's guide to bottom-up market sizing puts it bluntly: with top-down sizing, the data is often arbitrary and unreliable. A bottom-up estimate is better. Count the realistic customers, multiply by what each pays a year, and show the working in three lines so every input can be checked. Underscore's Richard Dulude makes the same case in two minutes, and I'm including it because he starts from a single customer, which is the part founders skip.

Play Video (Loads YouTube)

Traction gets checked for consistency

Revenue, users and growth can't usually be confirmed from outside before a meeting. What investors can check is whether they're consistent with everything that is visible:

• team size on LinkedIn
• how actively the company is hiring
• rough web traffic
• the size of the customers named on the logo slide
• in the UK, whatever's been filed at Companies House

None of those proves anything on its own. They're used to spot a mismatch. A company claiming fast enterprise growth with three people on LinkedIn and no hiring might be extraordinarily efficient. It might also be counting pilots as customers. The investor will want to know which, and they'll ask.

Unit economics get compared with what the investor has seen

Customer acquisition cost, lifetime value and payback period get compared with what the investor has seen in their own portfolio and deal flow for the same channel. An acquisition cost that looks implausibly low for the way you sell is a classic flag. So is a lifetime value built on an industry-average churn rate when you have cohort data of your own.

Competition gets searched

Investors look for competitors before they meet you. If they find one that raised a large round recently and isn't on your competition slide, they'll wonder whether you missed it or left it out. A deck that says no direct competitors invites exactly that search.

My advice is to name the strongest competitors yourself, including the well-funded ones, and say specifically why a customer would choose you. Investors don't need you to have no competition. They need to see you understand it. If you want a structured way to do that research, I've set it out in competitive intelligence for B2B companies.

Why the Team Gets Checked Hardest

Of all the things VCs weigh, the founding team comes first, so team claims get checked hardest. Every line of a founder's bio is checkable, and most of it is checked. In the Gompers survey, 95% of VC firms named the management team as an important factor, and 47% named it the single most important. Business model, product and market followed some way behind.

Management team 95%

Business model 83%

Product 74%

Market 68%

Industry 31%

Share of VC firms naming each factor as important in investment decisions

Management team	95%
Business model	83%
Product	74%
Market	68%
Industry	31%

Source: Gompers, Gornall, Kaplan and Strebulaev, How Do Venture Capitalists Make Decisions? (NBER), survey of 885 VCs, 2015 to 2016. Respondents could name more than one factor. Mostly US data. Chart by ThriveFinity.

So investors check job titles against LinkedIn and past employers, look up previous companies at Companies House to see how they ended, and call people who worked with you. Some may be people you didn't put forward, so it's worth assuming that could happen.

The practical lesson is dull but important. Make sure your deck, your LinkedIn profile and your company website all say the same thing about who you are and what you've done. Led growth at X on one and marketing manager at X on another is a small inconsistency that raises a much bigger question.

The Sources Investors Use

Most of the sources investors use for desk research are open to founders too. That's the useful part: you can run the same checks on your own deck before anyone else does.

What Companies House can and can't tell an investor

Companies House is the first stop for a UK company, and it's worth knowing its limits. Small companies don't have to file much. Under the current rules on GOV.UK, a company counts as small if it meets two of three tests, including turnover of £15 million or less, and it can choose whether or not to send a copy of the director's report and profit and loss account to Companies House. Micro-entities can send only a balance sheet with less information.

So for most start-ups, the public filing won't show revenue at all. That's why traction gets checked by consistency before a meeting and by bank statements after a term sheet. What a filing does show is the shape of the company: its officers, its shareholders with significant control, any charges over its assets, and whether accounts and confirmation statements have been filed on time. A late filing is a small thing. It's also an easy thing to find.

The register is also getting more reliable about who's behind a company. Identity verification became a legal requirement from 18 November 2025, after a voluntary period that started in April 2025, with timing that depends on your role. If you run or control a company, check yours is done before an investor looks. Companies House explains the process in three minutes:

Play Video (Loads YouTube)

Before you send a deck, pull up your own page on the Companies House register and read it the way an investor would. Ask yourself:

• Are the listed officers the people on your team slide?
• Does the register of people with significant control match your cap table?
• Are your accounts and confirmation statement up to date?
• Is there a charge over the company's assets that you haven't mentioned?

Each of those takes a minute to check and a lot longer to explain if someone else finds it first.

What UK Investors Add to the Checks

UK investors run the same core checks as anyone else, plus a few that are specific to the UK: tax relief eligibility, the shape of the round documents, and what Companies House shows.

They're also working in a smaller market than a year earlier. The British Business Bank's Small Business Equity Tracker says UK smaller businesses raised £12.3 billion of equity across 2,002 deals in 2025. Investment was down 4% on 2024 and deal numbers fell 17%. Seed-stage deal numbers fell 27%, the sharpest drop of any stage, and AI companies took 44% of the money, a record £5.4 billion. The tracker doesn't say how that changes investors' checks. My reading is that when fewer deals get done, a deck with a claim nobody can trace is an easy one to put down.

Total equity investment -4%

Deal numbers -17%

Seed-stage deal numbers -27%

UK smaller businesses: fall in equity investment, 2025 against 2024

Total equity investment	-4%
Deal numbers	-17%
Seed-stage deal numbers	-27%

Source: British Business Bank, Small Business Equity Tracker 2026 factsheet, 2025 data. UK smaller businesses raised £12.3 billion across 2,002 deals in 2025. Seed was the sharpest fall of any stage. Chart by ThriveFinity.

SEIS and EIS get checked early

For many UK angels, tax relief under the Seed Enterprise Investment Scheme (SEIS) or the Enterprise Investment Scheme (EIS) is part of the case for investing. So a line on your deck saying SEIS/EIS eligible will get checked. Investors want to know whether HMRC has given advance assurance, and what it covered.

Advance assurance is HMRC's view that an investment is likely to qualify, and it isn't the end of the matter. HMRC's own guidance says:

It should not be taken as a more general endorsement or as an indication of potential investment performance. Investors should consider conducting their own due diligence before proceeding with an investment.

It also says that if things change before you submit a compliance statement, the assurance will no longer apply. If your deck says eligible, be ready to show the letter and explain anything that has changed since.

Round documents set out what you're promising

The industry body UK Private Capital, formerly the BVCA, publishes model documents for early-stage investments, including a subscription agreement, a shareholders' agreement and articles. It says they have been drafted for use on a Series A funding round. The subscription agreement contains warranties, which are formal statements about the company that investors can rely on. When the templates were updated in February 2025, law firm Burges Salmon's summary of the changes listed revised intellectual property, business systems and data protection warranties among them.

That's the part founders often miss. The claims in your deck don't stay in the deck. Anything material tends to resurface as a question in due diligence, and the answers can end up shaping what you're asked to warrant. A number you wouldn't put your name to in a legal document probably shouldn't be on slide four either.

Three Ways Decks Fail Under Checking

Most deck claims that fail aren't invented. They're real numbers used in the wrong place. When an analyst has time to open the cited reports, I'd group the failures into three patterns. These are my groupings, drawn from checking decks, and not a published taxonomy.

• Category miscitation is using a benchmark from the wrong kind of product. A retention figure from a consumer app report on a B2B software slide is the classic case. The numbers look similar, but the populations behind them aren't comparable, and anyone who knows the sector will spot it.
• Recency misalignment is quoting data that's too old for a fast-moving market. A 2022 figure on AI infrastructure spending tells an investor in 2026 very little about today. My rule of thumb is to put the year next to any figure more than 18 months old, and to replace it if a newer one exists. Labelled old data is forgivable. Unlabelled old data looks like you hoped nobody would notice.
• Source laundering is citing a secondary source, like a blog post, someone else's deck or a news story, instead of the original research it quotes. Each retelling can round a number, shift its definition or drop its caveats. Analysts follow the chain back to the original, so you should too.

If you want to see how one of those checks runs end to end, from finding the original source to challenging it and signing it off, I walk through it in how Release Assurance checks a claim.

None of these failures is about the idea. Each is checkable in minutes, which is why they matter. An investor deciding whether to give you an hour won't spend that hour finding out. They'll spend five minutes, and if what they find doesn't match the deck, they'll move on to the next one in the pile. Good ideas get passed over because something in the deck made the investor doubt the people presenting it.

Questions That Come Later in a Round

A second set of questions usually arrives later, in partner meetings or legal due diligence, once the investor is seriously interested. These are the nine I'd prepare for, based on the checks above and on how due diligence is described by the investors cited below. Each one is easier to answer in the deck than in a follow-up email.

1. Market size method. Where does this number come from, and how was it worked out? A top-down 1% of a £4bn market is the usual trigger.
2. Benchmark category. Which report is this retention figure from, and does the product type match yours?
3. Lifetime value inputs. What churn rate is this built on, and is it yours or an industry average?
4. Competitors. Have you looked at the company that launched in your space last quarter? Any only or first claim gets searched.
5. Acquisition channel. Is the cost per customer based on a channel you've actually run, or one you plan to run?
6. Customer evidence. How many customers is this satisfaction score from, and how were they chosen?
7. Regulation. Which licences or approvals do you need, and where are you with them? Sharpest in fintech, health and AI.
8. Team depth. Has the team done this before, and if not, what replaces that experience?
9. Intellectual property. Does the company own everything it's built? Code written before incorporation, or by contractors without an assignment, surfaces here.

The last question deserves a UK footnote. Under UK law, a freelancer or contractor usually keeps the copyright in what they create unless a contract says otherwise. GOV.UK's guidance on copyright ownership says that for commissioned work, the first owner is the person or organisation that created it, unless you otherwise agree it in writing. If an agency or freelancer built your first product, check the paperwork before an investor's lawyer does. I cover this and the rest of the UK legal checks in the guide to what UK investors verify in a Series A data room.

What Due Diligence Adds After the Term Sheet

Due diligence is the formal investigation an investor runs once they want to invest, usually after a term sheet. It's slower, deeper and done with documents and people rather than public sources.

The Gompers survey put numbers on it. On average:

• a deal took 83 days to close
• a firm spent 118 hours on due diligence
• a firm called 10 references, with early-stage firms calling about 8 and late-stage firms about 13

That's mostly US data from 2015 and 2016, so read it as a rough guide.

Jason Lemkin, the SaaStr founder and investor, describes early-stage due diligence as checking that what you said is true. In his words, it's mostly confirmatory: do the beta customers really use the product the way you said, is the team as committed as you said, is the rest of the round as solid as you said. He also shared the best due diligence question he says he ever got from a VC:

X Jason Lemkin (@jasonlk) · SaaStr founder and investor · 26 January 2025

The best diligence question I ever got from a VC: What am I going to learn at the first board meeting that I don't know today?

Read the original post on X Show the Original Post

That question is due diligence in one line. Anything an investor would learn at the first board meeting, they'd rather learn now.

In a separate post on due diligence before and after a term sheet, Lemkin sets out how the depth changes. Before a term sheet, he describes a couple of customer calls, often just two or three, and a basic look at the financials and model. After it come deeper customer calls, bank statements with a full financial review, formal reference calls, and legal due diligence covering former founders, the cap table and how much of the founders' stock is still unvested.

Read that list again with your deck open. Every item tests something you've already claimed. The customer calls test the logo slide and the case studies. The bank statements test the revenue chart. The legal review tests the team slide and the cap table. Due diligence rarely discovers brand new facts about a company. It mostly finds out whether the facts you've already presented were true. The usual checks also set signed contracts against the logo slide, employment and contractor agreements against who built the product, and the cap table against the register of members.

If you'd like to hear what that process feels like from the investor's side, Underscore VC's Brian Devaney walks through it in four minutes:

Play Video (Loads YouTube)

When the Checks Miss

Due diligence isn't a guarantee, and investors know it. Return to Frank. JPMorgan didn't skip the check. It asked the right question, about the number of users, and it asked to see the data. The DOJ says the data it was shown had been fabricated for the purpose.

Two lessons come out of that, one for each side of the table. For investors, verify from the source system rather than from an export the company prepares. A spreadsheet of users is a claim about the database, not the database itself. A bank statement pulled directly is stronger than a revenue summary, and live access to a dashboard is stronger than a screenshot.

For honest founders, the lesson is that investors have every reason to check thoroughly, and they'll check you as hard as anyone. That's not personal. The easiest way through is to make every claim in the deck trace to something they can open themselves.

What to Do When an Investor Finds a Problem

If an investor finds a claim that doesn't hold, the best response is fast, specific and in writing. Say what was wrong, what the correct figure is, where it comes from, and what else you've re-checked as a result.

Most errors in decks are honest ones: an old figure, a mislabelled chart, a benchmark from the wrong category. Investors know that. What worries them is how you handle it. A founder who says you're right, here's the corrected number and the source, and I've been through the rest of the deck has turned a problem into a small piece of evidence that they're careful. A founder who argues, goes quiet or offers a new number with no source has done the opposite, and the response becomes the thing the investor remembers.

Lemkin makes the same point about timing in that post on pre and post term sheet checks:

If you have something to hide here, it will likely come out later. Post term sheet.

His advice is to get bad news out early. I'd add one thing. Get it out with a source attached.

What Founders Can Do Before Pitching

Run the investor's checks on your own deck first. Nearly everything above can be done by a founder in a few days, and the free Series A claim checklist sets it out as 23 checks across five claim categories.

1. Source every number to an original. If the market size can't be traced to a source you've opened yourself, replace it with a bottom-up estimate you can show in three lines.
2. Test your unit economics against comparables. Find two or three companies at the same stage, selling the same way. If your numbers are much better, explain why on the slide.
3. Search for competitors before investors do. Look for everyone in your space who has raised in the last two years. Name the strong ones and say why customers pick you.
4. Check your outside footprint. LinkedIn, your website, Companies House and the deck should tell the same story about the team, the headcount and the company. If there's a gap, explain it before anyone asks.
5. Keep a claims log. One row per claim, with its source, date and who checked it. My guide to fact-checking a pitch deck walks through it step by step.
6. Get it read by someone who didn't write it. The claims you've repeated for months are the ones you've stopped seeing. An AI assistant isn't that reader, because it tends to agree with you, as I explain in why most AI deck checks skip the hard part.

Put the evidence where investors will look for it

A short source line under each chart does more work than any amount of confident wording. UK mid-market accountancy firms, Companies House and ONS data, pulled March 2026 tells an investor you've done the work and lets them check it in a minute. An appendix slide listing every source with its date does the same job for the whole deck.

For numbers that come from your own systems, say which system and what period. ARR from Stripe, 30 June 2026 or Net revenue retention, 2025 customer cohort, 31 customers reads like a measured fact, because it is one. I'd much rather see a modest number with a clear source than a big one with none, and I'd expect most investors to agree.

Then build your data room as if every claim will be checked, because the important ones will be. If a claim on a slide can't be backed by a document in the room, either find the document or change the slide.

Where ThriveFinity Fits

We check claims for a living, using the same approach investors use: find the original source, test the claim against it, and record the evidence. Release Assurance does that for every headline claim in a deck, grades the evidence A to E, rewrites the weak claims with a citable source, and a named verifier signs the result.

• First Signal (free, under an hour): one claim or one positioning check, so you can see the method before you pay anything.
• Release Assurance Pack (£699, 3 business days): every headline claim in one deck, with ranked rebuttals and a signed Assurance Note.
• Release Assurance Suite (£1,749, 5 business days): a deck, one-pager and website checked against each other, which is close to what an investor does with a data room.

If you're on the other side of the table, our service for investors covers per-deal diligence briefs, deck claim checks and data-room consistency checks. And if you back or advise a founder and want their deck checked before your committee sees it, you can refer the founder to us. There's no cost to you, and we don't pay or accept referral fees.

We're not the right choice for legal due diligence (you need a solicitor), a financial audit (you need an accountant), or a valuation opinion. We don't give investment advice. We check whether the claims in a document are supported by evidence, and we tell you plainly where they aren't.