Cofounder breakups are like divorces in at least one precise way: the event itself does not determine the damage. The handling does. I spent years on the destructive kind, and along the way I got to study, by contrast, the teams that separated and kept building. The difference between the two groups was never the severity of the disagreement. It was method.
Here is the method, distilled from the wreckage.
What clean breakups share
Four things, consistently.
Speed. Once the decision is real, clean teams execute in weeks. Destructive teams let it stretch across quarters, and the stretch is where the poison brews: ambiguity for the team, selective memory on both sides, leverage accumulating for whoever benefits from delay.
One honest conversation before any negotiation. Not about terms; about truth. What worked, what did not, what each person actually wants now. Teams that skipped straight to numbers negotiated angry; teams that had the human conversation first negotiated tired but fair. The order matters more than the content.
Dignity as a term of the deal. The departing founder's story ("what do we tell the team, the users, the investors") is negotiated explicitly, and generously. It costs the company nothing to let someone leave as "a founder who built the first version and chose a different path," and it costs everything to make an enemy with equity. In my files, the vindictive announcements were reliably followed, within the year, by the vindictive lawsuits.
Paper, immediately. Every agreed point written and signed while goodwill exists: the equity outcome, the IP confirmation, the accounts handover, the announcement language, the mutual releases. Clean teams paper the exit in days. The others "agree on principles" verbally and discover, months later, that principles have two versions.
The checklist of what must be settled
Whatever your situation, the exit has to answer six questions, and an unanswered one is a future dispute: Equity: what the departing founder keeps (vested, or negotiated), what returns, at what price, paid how. IP: written confirmation that everything built for the project belongs to the project, including the pieces on personal machines and accounts. Access: domain, repositories, cloud, banking, socials, transferred and passwords rotated, the same week. Obligations: confidentiality confirmed, non-solicitation if agreed, and clarity on any non-compete's real scope. The story: the exact announcement, internal and external, agreed word for word. The release: mutual, so the exit actually ends the matter instead of pausing it.
Six questions, one sitting, one signature each. That is the entire difference between an exit and a dormant lawsuit.
With an agreement, this is administration. Without one, it is diplomacy.
If your team signed a founder agreement, the breakup mostly executes it: vesting says what is earned, leaver provisions say how this departure is classed, the buyback formula prices what returns. Emotions still exist, but the numbers are not negotiable at knifepoint, because they were negotiated at friendship. Cofounder breakups are not won; they are either administered or litigated, and the agreement is what makes administration possible.
If nothing was signed, everything above is still achievable; it just requires both people to behave well simultaneously, which is exactly what a breakup makes hard. Pre-incorporation, formalize the exit deal in a signed separation agreement now, since intended equity is still just promises to reallocate. Post-incorporation, expect a negotiation, keep it fair and fast for the reasons above, and involve counsel early if real value or real anger is on the table.
For the founders reading this before any breakup
If you are here preventively: this entire note is the argument for the founder agreement you have not signed yet. Every clean breakup I described was clean because the hard questions had answers before anyone was angry. You cannot negotiate the fire escape during the fire. Sign while you like each other; that is the whole trick.