Co-Founder Vetting: The Checklist, Questions and Trial That Prevent a Bad Split

Co-founder failure is expensive in a way that a bad hire is not. A bad hire costs a few months and some money. A bad co-founder costs a year, a share of the company that you cannot easily recover, and often the company itself, because investors will not fund a contested cap table.

The uncomfortable part is that founders usually knew. In almost every post-mortem I have been part of, the founder can name the moment in the first six weeks when they saw the thing that later ended the partnership, and talked themselves out of it because the alternative was starting the search again. Vetting is largely a mechanism for making yourself act on what you already noticed.

The four layers of co-founder vetting

Run them in that order. Each layer is more expensive than the last, so the cheap ones exist to prevent you spending three months on someone the first hour would have ruled out.

Layer 1: verify the factual claims

This is unglamorous and takes about ninety minutes per candidate. The purpose is not suspicion; it is that a specific class of problem, once discovered late, is unfixable.

• Employment dates and scope. Compare the profile against any public record you can find: company announcements, conference bios, published work, patent filings. Look for gaps described as consulting, and ask about them plainly.
• Actual scope of a claimed leadership role. Led engineering can mean forty people or two contractors. Ask for the headcount, the reporting line and who else could have claimed the same sentence.
• Current commitments. Is the person a director or shareholder of another active company? Company registries in most jurisdictions are searchable and free. An undisclosed active venture is the most common serious finding.
• Restrictive covenants. Non-compete, non-solicit and IP assignment clauses in their current contract. In some jurisdictions these are weakly enforceable and in others they are not. Either way you want to know before, not after.
• Immigration and residency constraints where relevant, because they determine whether full-time commitment on a specific date is even possible.
• Anything that would surface in investor diligence. Assume anything you can find in an hour will be found by someone with a professional budget.

Layer 2: twenty-four questions that produce real answers

Most co-founder conversations are a pleasant exchange of intentions. Intentions are free. The questions below are built around things that already happened, because past decisions are the only evidence available about future ones. Spread them over two or three conversations rather than running an interrogation.

Motivation and horizon

1. 1What made you start looking at this now, rather than a year ago or a year from now?
2. 2What does this company need to become for you to consider the next seven years well spent?
3. 3What outcome would you be happy with that I might consider a failure?
4. 4What is the smallest exit you would accept, and under what circumstances?
5. 5What would you be doing instead if this conversation did not exist?
6. 6Which part of company building do you actively not want to do?

Judgement and track record

1. 1Tell me about a decision you made that you now think was wrong. What did you believe at the time and what changed your mind?
2. 2Describe a project that failed. What was your contribution to the failure?
3. 3When did you last change your position on something significant because of someone else's argument?
4. 4What is a widely held belief in your field that you think is wrong?
5. 5Walk me through the hardest technical or commercial trade-off you have made, including the option you rejected.
6. 6What did you do when you realised a deadline was not going to hold?

Working style and conflict

1. 1How did the worst professional disagreement of your career start, and how did it end?
2. 2What do people find difficult about working with you? Who would say so?
3. 3How do you want to receive criticism, and what does bad criticism look like to you?
4. 4When we disagree about something where neither of us can prove we are right, how should we decide?
5. 5What working pattern do you need to be effective, and what would break it?
6. 6How much do you want to be told versus asked?

Commitment and circumstances

1. 1What are your financial constraints over the next eighteen months, honestly?
2. 2What else has a claim on your time: other companies, advisory roles, notice periods, family obligations?
3. 3What would make you leave this company in year two?
4. 4If we raised nothing in twelve months, what would you want to do?
5. 5How would your partner or family describe what you are about to take on?
6. 6What is the question you hoped I would not ask?

Two rules of use. First, follow every answer with what happened next, twice, until you reach a specific event rather than a summary. Second, notice who takes responsibility. Candidates who describe every past failure as caused by someone else are giving you a preview of how your partnership will be described later.

Layer 3: references, including the one you were not offered

Offered references are curated and mostly useless as a filter, but they are useful as a starting point. The technique is to use each call to obtain the next name.

1. 1

   Take the two offered references, but ask peers not managers

   A former manager will tell you about performance. A peer or a direct report will tell you about behaviour, which is what you are buying.

2. 2

   End every call with the same question

   Who else worked closely with them during that period and would have a different perspective? Then call that person. The second-order reference is where the useful material is.

3. 3

   Ask about the worst period, not the best

   Tell me about a stretch where the project was going badly. What was this person like in that month? Consistent answers across two independent sources are close to reliable.

4. 4

   Ask the counterfactual

   If you were starting a company, would you want them as a partner? Then, after the answer, ask what would have to be true for the answer to change. The hesitation is the data.

5. 5

   Ask what they need from a partner

   This is the only question referees answer generously, because it is framed as helping rather than judging, and it often surfaces the operating requirement the candidate did not mention.

Layer 4: the paid trial

Everything above is filtering. The trial is the assessment. If you take one thing from this article, take this: do not issue equity to someone you have not worked with for at least sixty days.

Designing a trial that produces signal

The most informative moment in a trial is almost never the deliverable. It is the first time something goes wrong. Watch for how long the gap is between the candidate knowing there is a problem and you knowing. In my experience that lag is the strongest single predictor of how the partnership behaves in a crisis.

Red flags, ranked by how often they end companies

What to agree before equity is issued

Once shares are issued, changing the terms requires the other person's agreement. Before issuance you are two people negotiating; afterwards you are one person asking. Settle the following in writing first, with a lawyer for the parts that bind.

• The split, and the reasoning behind it. Write the reasoning down. It is what you will both reread in eighteen months.
• Vesting. Four years with a one-year cliff is the market standard. Consider whether time already contributed counts as credited vesting.
• What happens on departure, distinguishing between leaving voluntarily, being asked to leave for cause, and being asked to leave without cause.
• Roles and final decision rights. Name the domains where each of you decides alone. Ambiguity here is the most common source of the fight that ends companies.
• IP assignment. All work assigned to the company, including anything created during the trial period.
• Full-time start date and what commitment means until then.
• Deadlock resolution. What happens on a 50/50 split when you genuinely cannot agree. An outside advisor with a casting vote on defined categories is the simplest mechanism.

If a candidate resists putting any of this in writing on the grounds that it is too early or shows a lack of trust, treat it as a finding rather than a negotiation. People who intend to honour terms rarely object to recording them.

A short version you can run this week

1. 1Ninety minutes of verification: registries, dates, scope, current commitments.
2. 2Two conversations using the questions above, with four seconds of silence after each answer.
3. 3Three reference calls, at least one of which was not offered to you.
4. 4A written trial scope with success criteria, signed by both of you before it starts.
5. 5A two-way written review at the end, written independently before either is read.

If you are still building the shortlist rather than assessing a specific person, start with How to find a co-founder, and if the person you are assessing is technical and you are not, the additional instruments are in How to find a technical co-founder.