← shouldwebecofounders

guide

Co-founder conflict: how to fix it, and when to split

Co-founder conflict is normal; the failure mode is unrepaired conflict. Fix it by separating the recurring pattern from the current argument: name the loop you are in (usually one pursuing and one withdrawing), define who decides what in writing, replace contempt with a specific complaint, and make repair attempts explicit rather than hoping the mood passes. Split when repair attempts have stopped working, not because you fight.

last updated 1 September 2026

Conflict is not the problem. Unrepaired conflict is.

Gottman's relationship research is the cleanest analogy for founding teams: pairs who argue often do fine, provided they reach back afterwards. What predicts breakdown is contempt — mockery, eye-rolls, explaining someone to themselves — and the absence of repair attempts. If your last ten disagreements ended with someone leaving the room and nothing being said the next day, that is the finding you are living in.

A repair attempt can be small and even clumsy: a joke, a coffee, “I was sharp earlier”. It only counts when the other person can accept it, which is why one founder's repairs sometimes read to the other as sweeping things under the rug.

Find the loop before you argue the incident

Nearly every long-running founder fight is one pattern wearing different costumes. The most common is pursue-withdraw: one person escalates to get engagement, the other goes quiet to lower the temperature, and each move justifies the other. Neither person is being unreasonable on their own terms, which is why both feel misread.

The practical fix is a pause with a fixed return: the withdrawing founder gets an hour, and the conversation resumes at a stated time. Space without a return time reads as abandonment; pressure without space reads as an attack.

Most ‘personality clashes’ are undefined authority

When the same argument keeps recurring, look for a missing decision. Who decides hires? Who owns pricing? What spend needs both signatures? Who breaks a deadlock, and in which domains? Ambiguity does not stay neutral — it defaults to whoever is more comfortable taking ground, and the other founder banks resentment they will not describe as resentment.

Write it down, including the boring parts: vesting, cliff, what happens if someone leaves in month six. Reluctance to write down something already agreed verbally is data.

Fairness is where two reasonable people mean opposite things

One founder means equal shares; the other means shares proportional to contribution. Both say “fair”. Add unequal personal runway — one can go two years unpaid, one can go four months — and you have a company whose risk appetite is being set silently by whoever is closest to broke.

Three signals it is time to split

  • Repair attempts no longer land. You still apologise, and nothing changes.
  • Decisions keep being made without you, after you have named it more than once.
  • Agreements will not be written down, or written agreements are treated as flexible.

None of those is “we fought again”. Frequent, resolvable conflict is a working founding team; polite avoidance with a stack of unspoken grievances is the one that ends badly.

A ten-minute way to surface this

If you are already tense, an open conversation tends to reproduce the loop rather than expose it. Structured scenarios with private answers and a simultaneous reveal remove the mirroring, and give you a ranked list of where you are actually far apart — repair, conflict style, pressure, power, money and blind spots. Free, no accounts, on your own phones.

Frequently asked

How do you resolve co-founder conflict?
Deal with the pattern before the incident. Agree a time to talk that is not immediately after the fight, state a specific complaint instead of a character judgement, get explicit about which decisions belong to whom, and make a repair attempt that the other person can actually accept. Recurring conflict about the same theme is nearly always undefined decision rights or unspoken fairness expectations, not a personality problem.
Why do co-founders fall out?
The common causes are decisions made without the other founder, equity and workload perceived as unfair, one person absorbing all external relationships and information, differing ambitions discovered late, and contempt entering disagreements. Most of these are visible early but nobody names them while things are going well.
What is the pursue-withdraw pattern in a founding team?
One founder escalates to get engagement while the other goes quiet to reduce heat. Each behaviour triggers the other, so it intensifies under pressure. The fix is not for the pursuer to push harder or the withdrawer to disappear better — it is agreeing a pause with a fixed return time, so the withdrawer gets space and the pursuer gets certainty.
Should co-founders use a mediator or coach?
It helps when both people still want the company to work and the arguments have become repetitive. It does not help when one founder has already decided to leave, or when the real issue is an unwritten agreement about equity or authority — that needs a document, not a conversation.
How do you know when to break up with a co-founder?
Look for three things: repair attempts that no longer land, a pattern of decisions being made without you, and refusal to write down what has been agreed. Those predict a bad ending far better than the frequency or volume of arguments.
How do we prevent co-founder conflict before it starts?
Put the six pressure areas — repair, conflict style, response to pressure, decision rights, fairness about money, and blind spots — under mild pressure early and in writing. Structured scenarios with private answers surface disagreements while they are still cheap.