Set decision rights before you need them. Give each domain a single owner who decides in their lane, reserve a short list of company-defining calls for consensus, and name one cofounder — usually the CEO — as the tiebreaker for urgent deadlocks. Write the agreement down and keep it current. The goal is not to remove disagreement. It's to know who owns the call when you diverge.
Why "we decide together" breaks down
Most cofounders start with consensus. You had to agree to start the company, your skills overlap, and there's no real hierarchy yet. So you decide everything together, and for a while it works.
Then it doesn't. As the company grows, your roles diverge and the volume of decisions climbs. Consensus on everything turns into endless debate and paralysis, and the founding team gets stuck relitigating calls that should already be made. Ed Batista, who coaches founding teams, describes three modes cofounders actually rely on — consensus, functional expertise, and hierarchical authority — and warns that leaning too hard on any single one is the failure (Batista).
The fix is not a personality change. It's a structure. Decide how you'll decide, and do it while you still agree.
The three modes, and when each one wins
Every cofounder decision runs through one of three modes. Naming them out loud is half the work.
- Consensus. Everyone has to agree. Right for a small set of high-stakes, hard-to-reverse calls — raising money, changing the direction of the company, adding or removing a founder, equity, big hires and fires. Wrong for anything routine.
- Functional expertise. The person who owns the domain decides. The one with the relevant knowledge and responsibility makes the call and keeps the others informed. This is where most decisions should live once your roles are actually divided.
- Hierarchical authority. One person — typically the CEO — breaks the tie when the first two modes don't resolve it. Batista's caution: the leader has to influence rather than command, and the others have to learn to disagree and commit.
The mistake isn't picking the wrong mode once. It's never deciding which mode applies to which decision. So decide that first.
A decision-rights framework you can copy
Borrow from the tools bigger companies use. The two most common are RAPID, built by Bain, and DACI, which came out of Intuit.
RAPID assigns five roles to any important decision: Recommend proposes the path, Agree holds a bounded veto (legal, risk, compliance), Perform executes, Input supplies facts with no approval right, and Decide is the single person who commits the company (Umbrex on RAPID). DACI is simpler — Driver, Approver, Contributor, Informed — and keeps exactly one Approver for speed and accountability (Umbrex on DACI). For a two- or three-person founding team, DACI's speed usually beats RAPID's formality. Reach for RAPID when a call carries real governance or risk weight.
You don't need the full apparatus. You need one thing: for each class of decision, a single name in the Decide seat. Here's a starting map two cofounders can adapt in an afternoon.
| Decision type | Owner (decides) | Mode | Who must weigh in |
|---|---|---|---|
| Product roadmap, engineering architecture | Technical cofounder | Functional expertise | CEO gives input |
| Go-to-market, pricing, brand | CEO / commercial cofounder | Functional expertise | Tech cofounder gives input |
| Hiring within a function | Function owner | Functional expertise | Other cofounder informed |
| Fundraising, board, cap table | Both | Consensus | Unanimous |
| Company direction, pivot | Both | Consensus | Unanimous |
| Adding or removing a founder | Both | Consensus | Unanimous |
| Day-to-day operating deadlock | CEO | Hierarchical tiebreak | Loser disagrees and commits |
Two rules make the table work. First, the owner consults but does not need permission — input is not a veto. Second, the consensus list stays short. If everything is on it, you've rebuilt the paralysis you were trying to escape.
Name the tiebreaker before the tie
The hardest conversation is the one about deadlock, and the worst time to have it is mid-deadlock. Agree the tiebreak rule while things are calm.
For a 50/50 partnership, "we'll just work it out" is not a rule — it's a coin flip you haven't named. Better options, in rough order of how much structure they add:
- Domain owner decides. Most deadlocks dissolve because the call clearly sits in one person's lane. Default here first.
- CEO breaks operating ties. For genuinely shared operating calls, the CEO gets the final say — and the other founder commits fully once the call is made.
- Consensus with a cooling-off clock. For the short list of company-defining decisions, require agreement, but cap the debate. If you can't align in a set window, the status quo holds and you revisit on a schedule.
Whatever you choose, the CEO-versus-CTO split should make most of this obvious — clear roles mean fewer decisions land in the contested middle in the first place.
Write it down, then keep it current
A decision-rights framework in your heads is not an agreement. It's two different memories waiting to conflict. Document the agreements: which decisions belong to whom, what needs consensus, how ties break. Capture the actual calls you make in your Commitments so there's one record instead of two accounts.
Then keep it current. Your framework from month one won't fit month eighteen — roles diverge further, the stakes rise, and decisions that were once shared should move to a single owner. Revisit the map on a cadence. When it drifts out of date, you feel it as friction long before you name it as a decision-rights problem.
Decision rights are not about control. They're about speed and trust — knowing who owns the call, so you can move fast on the ninety percent and slow down on purpose for the ten that matter. When you disagree about the decision itself rather than about who gets to make it, that's a good sign. And when a call surfaces a deeper rift, handle it as conflict to resolve, not a rule to rewrite.
Decide how you'll decide. Write it down. Keep it current. That's the operating system a partnership runs on.
Frequently asked questions
- Who gets the final say between cofounders?
- It depends on the decision. Give each domain an owner who decides in their lane. Reserve a small set of company-defining calls for consensus. When cofounders deadlock on something urgent, name one person — usually the CEO — as the tiebreaker in advance.
- Should cofounders decide everything together?
- No. Deciding everything by consensus is the fastest way to stall a company. Consensus is right for a short list of high-stakes, hard-to-reverse decisions. Most calls should belong to a single owner who consults the other.
- What is a decision-rights framework?
- A decision-rights framework names, ahead of time, who recommends, who must agree, who provides input, and who decides for a given class of decision. RAPID and DACI are two common models. The point is to remove ambiguity about who owns the call.
- How do cofounders resolve a deadlock?
- Agree on a tiebreak rule before you hit one. Common patterns: the domain owner decides, the CEO breaks ties on operating calls, and a defined list of decisions requires unanimous agreement. Write the rule down and keep it current.
- Does the CEO cofounder overrule the others?
- Only on the decisions you've agreed the CEO owns. Hierarchical authority works when the leader influences rather than commands and the others practice disagree-and-commit. It fails when the CEO reaches for it on every call.
