Conflict and repair (the ones that predict breakups best)
How a pair recovers matters more than how often they fight. These come from Gottman's work on repair and contempt and from attachment research on pursue-withdraw cycles.
1. Contempt instead of criticism
Eye-rolls, mockery, "obviously", explaining you to yourself. Gottman's research found contempt is the single strongest predictor of relationship dissolution — a complaint is workable, contempt is corrosive.
2. No repair attempts
Healthy pairs argue and then reach back: a joke, an apology, a change of tone. If nobody ever repairs, every fight is added to a stack instead of resolved.
3. One founder always pursues, the other always withdraws
One escalates to get engagement, the other goes quiet to lower the heat. Both are trying to be safe and both read the other as the problem. It gets worse under fundraising stress, not better.
4. Conflict is avoided entirely
A founding team with no visible disagreement in month three isn't aligned — it's postponing. The bill arrives when the first real trade-off is forced.
5. Disagreements get re-litigated forever
If a decision never actually closes, there is no decision process, only whoever has more energy that week.
6. Third parties get recruited
Investors, employees, or a spouse used as leverage in a founder dispute. It converts a two-person problem into a company-wide one.
Power and decision rights
Most founder disputes people describe as “personality clashes” are actually undefined authority. Faultline research on subgroup splits applies directly to three- and four-person teams.
1. Decisions get made when the other founder isn't there
Blind decision-making is the fastest route to resentment. It rarely feels malicious to the person doing it — that's exactly why it repeats.
2. "We'll figure out titles later"
Ambiguity favours whoever is more comfortable taking ground. Undefined decision rights don't stay neutral.
3. A 50/50 split with no tie-breaker
Equal equity is fine; equal authority on every domain is not. Name who decides what when you disagree, in writing.
4. One founder owns all external relationships
Investors, customers, and press all routed through one person creates a structural power gap regardless of the cap table.
5. Information asymmetry
One person holds the numbers, the bank access, the roadmap. Anyone can look reasonable when only they can see the data.
6. Subgroups form (a "faultline")
Two technical founders and one commercial one, or two who went to school together. Faultline research shows teams split along pre-existing lines under stress.
Money, equity and fairness
Noam Wasserman's The Founder's Dilemmas is the reference here: rushed splits, no vesting, and unspoken definitions of fairness recur across thousands of founding teams.
1. Equity conversation keeps getting deferred
Wasserman's Founder's Dilemmas data is blunt here: teams that split fast and never revisit it are the ones that blow up. If the conversation is uncomfortable now, it will be unaffordable later.
2. No vesting or cliff
Without vesting, a founder who leaves in month five keeps a chunk of the company forever, and every future investor makes it your problem.
3. Fairness is defined differently and nobody notices
One person means equal shares, the other means shares proportional to contribution. Both say "fair" and mean opposite things.
4. Unequal financial runway that's never discussed
One founder can go two years unpaid, the other can go four months. That difference silently sets the company's risk appetite.
5. Side projects and consulting kept quiet
Not the moonlighting itself — the not mentioning it.
6. Spending is decided by whoever moves first
No threshold above which both founders sign off. Small at pre-seed, catastrophic at Series A.
Trust and reciprocity
Behavioural economics gives you cleaner signal than conversation: the investment (trust) game and public-goods games both expose how someone treats shared upside when it costs them something.
1. Reciprocity doesn't come back
In the classic behavioural trust game, people who receive generously and return little are consistent about it. Watch what returns, not what's promised.
2. Free-riding on shared effort
In public-goods experiments a few people always under-contribute while expecting the group payout. This shows up as slipping commitments, not stated refusals.
3. The story changes with the audience
One version for investors, another for the team, a third for you. Not always lying — often just conflict-avoidant. Both are dangerous at scale.
4. Credit drifts toward one person
"I built" in public, "we struggled" in private. Track pronouns in their public posts.
5. Bad news arrives late
A founder who reports problems only once they're unfixable is optimising for image over the company.
6. Contracts treated as a formality
Reluctance to write down anything you've already agreed to verbally is information.
Workload, pace and ambition
1. Effort asymmetry nobody will name
One person is doing 70% and both know it. Unnamed, it becomes contempt within a quarter.
2. Different definitions of urgency
Not laziness — genuinely different internal clocks. Fine if explicit, deadly if assumed.
3. Feedback lands as an attack
If "this landing page isn't working" is heard as "you're not good", you can't ship together.
4. The relationship only survives on wins
Plenty of teams look great while the graph goes up. Test them on the flat month.
5. No shared picture of the outcome
One wants a lifestyle business, one wants a decacorn. Both are legitimate; together they aren't.
How to actually test for these
Reading a list doesn't help much — you'll both nod along and recognise each other in the flattering entries. What works is a structured situation with private answers and a simultaneous reveal, so nobody mirrors the room. That's what this site is: two short games (a scenario trek and a shared lemonade-stand company) that put the six areas above under mild pressure and then hand you the disagreements, ranked by distance.
It's free, needs no accounts, runs on your own phones, and takes about ten minutes.
Frequently asked
- What is the biggest co-founder red flag?
- Contempt during disagreement, combined with no attempt to repair afterwards. Nearly every other red flag is survivable if the pair can fight and then reconnect; contempt without repair predicts breakdown better than any skills gap or equity dispute.
- How do I know if a co-founder is right for me?
- Stop interviewing and put them under mild, real pressure: split a small budget, make a decision without full information, disagree about something that matters, and see whether they repair, share credit, and tell you bad news early. Behaviour under trade-offs predicts far better than a compatibility conversation.
- Are co-founder personality differences a bad sign?
- No. Differences in style, pace, and risk appetite are usually an asset. The problem is undiscussed differences: two founders who assume they share a definition of fairness, urgency, or decision rights and only find out during a crisis.
- How can I tell if a co-founder is hiding their real views?
- You can't detect lying reliably, and no test can. What you can do is look for inconsistency across contexts — the same question asked in different framings, private answers revealed simultaneously so nobody mirrors, and situations with a small real cost attached. Impression management leaks under those conditions.
- When should co-founders split up?
- When repair attempts have stopped working, not when there's conflict. Persistent contempt, decisions consistently made without you, and an unwillingness to write down agreements are all stronger split signals than any single argument.