Startup due diligence checklist for angel investors and family offices

What a startup due diligence checklist should cover

A seed-stage diligence checklist for an angel, syndicate lead or family office covers seven areas: the entity and who owns the IP, the founders, whether the technology is built or just claimed, customers and revenue, the market including the cheap alternatives, the deal terms, and legal and regulatory exposure. It's only useful if it ends in a written verdict.

Start with the cheap checks that can end a deal by themselves, and stop when something can't be fixed.

The list below is ordered by what kills deals most cheaply. Checking the entity chain takes an hour and can end a deal by itself. A market study takes a week and rarely does. Work top to bottom, and stop when you hit something that can't be fixed before closing.

1. The entity and the IP

Confirm that the company raising is a real, properly formed entity, that it's the entity your money actually goes into, and that it owns the intellectual property the business depends on. Ask for the certificate of incorporation, the current cap table, and signed invention-assignment agreements from every founder and early engineer.

Make sure the company raising the money is the one that actually holds the key to its IP.

The usual failure is IP that lives somewhere else: with a founder personally, with a founder's previous employer, with a university lab, or, for Israeli companies, in a local entity with grant obligations attached. If a founder built the first version while employed elsewhere, ask for that employer's release or a lawyer's view of the assignment clause. For Israeli companies, see how to diligence an Israeli startup as a US investor.

2. The founders

Check each founder's stated background against sources the founder didn't write: prior employers, degrees, prior companies and how they ended, litigation and regulatory records, and people who worked with them. Reference calls matter most, and the useful ones are with references the founder didn't give you.

The references that matter most are the ones the founder didn't hand you.

Give prior companies extra attention. A founder whose last company "was acquired" may have sold it for less than the money raised. One who "stepped back" may have been removed. Neither is disqualifying, but the story should match the record. A founder who shades the record in a pitch will shade it in a board meeting too.

3. Built versus claimed

Separate what the product does today from what the deck says it'll do. Ask for a live demo run by the technical founder (a video doesn't count), and ask what's built in-house versus licensed, open-source or called from another company's API. For AI companies, ask what data the company owns that a competitor couldn't buy.

A live demo run by the technical founder shows what's built. The deck shows what's claimed.

If you can't read architecture or code yourself, this is the section where an outside technical read pays for itself. Outright fraud is rarely what you find. What you usually find is a thin layer over someone else's product, priced as if it were the product. More on this in can you use AI for startup due diligence.

4. Customers and revenue

Confirm that revenue is real, recurring and not concentrated in one friendly customer. Get the customer list with contract values, talk to two customers the founder picks and one the founder didn't pick, and reconcile the deck's revenue figure with bank statements or the accounting system.

Call two customers the founder picks and one they didn't, then check revenue against the bank.

Watch the labels: pilots counted as customers, letters of intent counted as contracts, a one-month peak annualized and called ARR, and revenue from the founders' own network.

5. Market and alternatives

Size the market bottom up (how many buyers, at what price) instead of accepting a top-down industry figure, and list what a customer would do if this company didn't exist. At seed, the competitor to worry about most is usually a free feature in a product the customer already owns, more than any other startup.

Count the buyers one by one. The real competitor is often a free feature they already own.

6. Deal terms

Read the terms against the stage. Check that the round label matches the price, that a post-money SAFE's cap translates into the ownership you expect, how many other SAFEs and notes are outstanding, what the option pool does to your stake, and whether you get pro-rata and information rights.

Model the SAFE cap, the other SAFEs and the option pool before you trust your ownership number.

Y Combinator's post-money SAFE fixes the holder's ownership at conversion no matter how many other SAFEs are raised, which is why the cap is the number to model. And a round called "seed" at a pre-money that prices like a Series A is common enough that you should check every time.

7. Legal and regulatory

Flag the questions only counsel can answer, and leave the answers to counsel: foreign-investment review, export controls and sanctions for defense and dual-use companies, sector regulation, pending litigation, and open-source license obligations. Your job in diligence is to know these questions exist before you wire. The lawyer's job is to answer them.

Know which legal questions exist before you wire. Answering them is counsel's job.

How much time is enough

More than most investors spend. The best-known data point is Robert Wiltbank and Wade Boeker's 2007 study of 539 angels and 1,137 exits for the Kauffman Foundation and the Angel Capital Education Foundation. In it, 52% of exits returned less than the money invested, and the median deal got 20 hours of diligence.

In the Wiltbank study, deals with more diligence hours returned far more. Spend yours where it counts.

Exits from deals above that median returned 5.9 times the money invested, against 1.1 times for deals below it, and the top quartile, with more than 40 hours, returned 7.1 times. The authors are careful to call this a correlation, and they say the quality of the hours probably matters as much as the count. (You'll also see it said that the Angel Capital Association recommends 40 hours per deal. Its 2007 best-practice guidance on diligence gives no hour figure.)

Institutional venture firms spend far more. A survey of VCs by Gompers, Gornall, Kaplan and Strebulaev found the average firm spends 118 hours on due diligence over an average of 83 days from first meeting to close, and calls 10 references. Early-stage firms average 81 hours and 8 references.

Few individual investors have that kind of time for every deal. So front-load the cheap, decisive checks, and spend real hours only on the deals that survive them.

Write it down: the one-page verdict

Diligence that isn't written down won't survive your next conversation with the founder. We use and recommend a one-page verdict: one of four words (invest, invest with conditions, pass for now, pass), the three things that would change it, the cheapest experiment to run before wiring, and a score on three independent axes (business, integrity, and fit to your criteria).

One page, one of four words, and three scores kept separate so they can't average out.

Keeping the three axes separate is the whole point. A strong business with a founder-integrity problem should read as exactly that, instead of averaging out to "promising". The evidence behind the verdict goes on the following pages, with every claim sourced and every inference labeled as one.

When to bring in an outside read

If you work through this list and the open items are the ones you can't close yourself (the codebase, the cap table mechanics, an Israeli grant agreement, a founder's history in another country), an independent read costs a small fraction of the check. Olivent Diligence runs this checklist on a fixed clock and delivers the one-page verdict plus the evidence: a Pre-Wire Screen is $2,500 in 48 hours, a Full Diligence $8,500 in five business days, paid only by the investor and never by the company.

When the open items are ones you can't close yourself, get an independent read.