A deadlock clause answers one question: when the founders cannot agree on a decision the company needs, how does the company still move?
That is all it does, and companies die for lack of it. A 50/50 team, or any team whose rules require unanimity on big calls, carries a structural stall point: the first genuine disagreement between equals freezes everything downstream of it. Shipping waits, hiring waits, the round waits. I watched deadlocks kill companies that had revenue, users and momentum, because the one thing they could not produce was a decision.
A deadlock clause is not for deciding who wins. It is for making sure the company does not lose.
The main designs, from lightest to heaviest
Final say by domain. The cleanest for two founders: product and technology decisions land with one founder, commercial and financial with the other, each after mandatory consultation. Most "deadlocks" are really jurisdiction disputes; this design dissolves them in advance by giving every decision a home.
The swing vote. A trusted third party, named in advance, who breaks defined deadlocks: an advisor both founders respect, later an independent board member. Powerful, with two conditions written down: what counts as a deadlock (a defined class of decisions, a cooling-off period first) and what the third party decides on (the specific question, not the company's direction).
The escalation ladder. For teams who want process before verdicts: a mandatory cooling-off period, then a structured founders' session, then mediation, and only at the ladder's end, a decisive mechanism. The ladder's value is that most disputes die on its lower rungs; its risk is a ladder so long it becomes the deadlock.
Buy-sell mechanisms, the heavy machinery. The shotgun clause and its variants: one founder names a price, the other must either sell at it or buy at it. Elegant on paper, brutal in practice, and dangerous where cash positions are unequal, because the richer founder can name a price the poorer one cannot match. In my files, shotgun clauses were rarely fired and often brandished. Treat them as the last rung of a ladder, never the first response, and price-protect them if you adopt one.
Choosing yours, before you need it
Three questions decide the right design. How many founders? (Two equals usually want domains plus a ladder; three can often outvote a deadlock away, if voting rules are written.) Is there a person you both genuinely trust? (If yes, the swing vote is available; if no, that is worth knowing too.) And what class of decisions deserves protection? Not everything: reserve the mechanism for the calls that can kill the company, and let daily life run on speed and forgiveness.
Then write it while you agree on things. A deadlock mechanism negotiated during a deadlock is itself deadlocked; the entire value of the clause is that it predates the fight. Before incorporation, it governs how the founding team decides; at incorporation it carries into the shareholders' agreement, where investors will be genuinely relieved to find it already exists.