You and your co-founder want different companies
The split is already happening; the only open question is whether it happens on paper or through attrition. A four-week protocol, and the case for waiting instead.
$ eldr --explain cofounder-split
Two years ago you agreed on everything because there was nothing yet to disagree about. Now there is revenue, or nearly, and it turns out he wants a services business with eighty percent margins and you want a product, or you want to raise and he wants to stay lean, or he wants his brother-in-law running sales. The disagreement is not the problem. The problem is that neither of you has said the word "split" in a meeting, and both of you have said it to your partner at home.
The claim
A co-founder disagreement about the shape of the company does not resolve itself, and every month you avoid naming it converts equity into resentment at a fixed rate. The split is already in progress. What is undecided is only whether it happens in writing, while there is still something worth dividing, or through attrition — two people gradually doing less, blaming each other's pace, until the company is not worth splitting.
Falsifiable version: name one company you know where two founders wanted materially different businesses, said nothing structural for a year, and arrived somewhere good. I have not found one. If you have, the claim is weaker than I think.
The four-week protocol
This is a sequence, and the order is the part that matters. Every step is designed to be checkable, and none of them require you to have decided anything yet.
- Week one: separate written briefs. Each of you writes one page: what the company is in three years, what it sells, who it hires next, and what you personally want out of it. No meeting first — meetings average the two documents before they exist. Then swap. Some pairs find the gap is smaller than the silence made it feel; others find it larger and specific, which is also progress.
- Week two: the premortem, run jointly. It is eighteen months from now and the company failed. Each of you writes the story. Do this before any negotiation, because it is the last moment you are still on the same side of the table. Klein's version of this in corporate settings surfaced failure modes nobody would raise in a planning meeting, and a founder pair is the most conflict-avoidant planning meeting in existence.
- Week three: the terms, in the abstract. Not "you leave". Vesting as it actually stands today, who keeps the name, what happens to the customers, what a buyout would be priced from, what a clean handover looks like. Both of you should be able to state both sides of these terms. If you cannot state his side, you are not ready to negotiate.
- Week four: the position. MOVE, HOLD or SPLIT, with a date. A decision here is not "we break up" — it is one of three named outcomes with an owner and a review date.
The protocol is not clever. Its only property is that it is sequential and dated, which is exactly what a year of hallway conversations is not.
What men get wrong in week three
They negotiate against a person instead of against a document. The moment the terms are attached to him rather than to the situation, every clause reads as an accusation and the conversation becomes about whether you are a good friend. The document is the whole technology here. Write the terms as if a third founder, absent, were being dealt with fairly — then apply them to yourselves.
The second error is treating the cap table as the scoreboard. Equity is what you fight over when you have stopped being able to talk about the thing you actually disagree on, which is usually the shape of the work and who gets to decide. Two founders who resolve the decision rights rarely fight about the points. Two founders who resolve the points without touching decision rights fight again in the spring.
The case against this
A structural conversation started in the wrong month can destroy a company that would otherwise have grown out of the disagreement. Plenty of founder pairs disagree violently about direction in year two, hire a competent operator in year three, and stop caring because the business found its own shape. Forcing the split conversation before there is revenue to divide can cost you the option value of the company itself — and the man pushing for structure is often the one whose nerve failed first.
That objection is real and I have watched it be right. There is a narrower version of it that is even more uncomfortable: sometimes the "different companies" framing is cover for one founder's private wish to leave, and the protocol above hands him a respectable process for a decision he made in private months ago. If that is you, the honest move is to say so before week one, not to run four weeks of documentation that arrives at your predetermined answer while looking like fairness.
So the position is conditional, not universal. Run the protocol when there is an externally-enforced date in the next two quarters — a raise, a lease, a key hire, a customer contract that needs one signature. That date makes the conversation concrete and non-personal. With no such date, HOLD is defensible, and the read will say HOLD, and it will name the event that ends it.
Why the instrument is useful here specifically
Because this is the decision where both parties are least able to see their own file. Everything either of you says to a friend is pre-filtered for who comes out looking reasonable. A position generated from fourteen inputs has no stake in either of you looking reasonable, and it ships with the strongest available case against itself, which is the part you should read first and the part you will want to forward to him.
For entertainment and reflective purposes only. Not a substitute for professional financial, legal, medical or mental-health advice.