How to Set Up Decision Rights Between Cofounders
Most cofounder conflict is not about the decision. It is about who had the right to make it. That distinction matters because the two problems look identical from inside an argument and have completely different fixes: one needs a conversation about substance, the other needs a rule written down once. This page is the second thing — a practical method for deciding who decides, including the part almost every founding team skips, which is what happens when you genuinely cannot agree.
What’s inside
Why undefined decision rights cause so much conflict
Why the domain split stops working
Step one: the decision inventory
Step two: four roles per decision
Step three: the deadlock rule
How to write this in an hour
Keeping it alive
Where Blomma fits
Why undefined decision rights cause so much conflict
At eight people, every decision touches both of you and you simply talk. No rules needed, and writing any would have been a waste of time.
The problem arrives with scale, and it arrives disguised. Pricing becomes a product decision and a revenue decision at the same time. A senior hire is a function decision and a culture decision. There is no answer to who decides, so it gets decided by whoever pushes hardest, moves fastest, or raises it in front of the most people. Both of you now have direct evidence that the other overreaches.
Nothing about that requires either person to be difficult. It requires only an undefined boundary and enough decisions crossing it. But the experience is entirely personal, because what you observe is not a missing rule — it is your cofounder doing something that looks like disregard for you.
This is why decision rights are worth more than they appear. They do not make you agree. They convert arguments about authority into arguments about substance, and arguments about substance can actually be resolved.
Why the domain split stops working
Nearly every founding pair starts with a domain split: you take product, I take go-to-market. It is intuitive and it works until roughly thirty people, then degrades for three reasons.
The important decisions are the cross-domain ones.Pricing, positioning, roadmap sequencing, headcount allocation, which customers to say no to. A domain split explicitly does not cover the decisions that matter most, because those are precisely the ones that span domains.
Domains do not have edges.Is developer documentation product or marketing? Is a pricing page product, marketing, or sales? Every genuine question sits at a boundary, so the split resolves the easy cases and leaves the hard ones ambiguous.
Domains drift.The person who owned go-to-market at ten people may now own two functions and a fundraise. Titles stay; scope moves. Two years of drift produces an unwritten arrangement that neither of you could accurately describe.
The fix is not a better domain split. It is to stop organising by domain and organise by decision, because decisions are the unit where the conflict actually happens.
Step one: the decision inventory
List the decisions that matter. Not tasks — decisions, meaning moments where a choice is made that is costly to reverse.
Most founding teams land on ten to fifteen. A representative set: pricing and packaging; roadmap priority for the next quarter; whether to build, buy, or partner; senior hires; firing a senior person; compensation bands; spend above a threshold; taking on a customer that requires custom work; fundraising timing and terms; board composition; entering a new market or segment; publicly announcing something; and legal or regulatory positions.
Write yours from your actual history rather than a template. Two prompts get you most of the way: what were the last five things we argued about, and what are the five decisions coming in the next six months that we have not discussed? The first list tells you where the gaps are. The second tells you where the next argument is coming from.
Then add the item almost everyone omits: decisions about each other. Who decides if one of us should change roles? Who decides if one of us is underperforming? Leaving this undefined is why cofounder performance problems go unaddressed for years — there is literally no mechanism.
Step two: four roles per decision
For each decision, assign four roles. Keep it to four; more becomes a governance document nobody reads.
Decides.One name. Not “both of us” — that is the absence of a decision right, restated. If you genuinely cannot assign a single name, that decision belongs in the deadlock rule below, deliberately and explicitly.
Consulted.Who must be asked before the decision, with enough time to give a real view. This is the role that does most of the relationship work: being consulted properly is what makes not deciding tolerable. Most founder resentment about decisions is actually resentment about not having been consulted.
Informed.Who hears about it after, and how fast. Cheap to specify and it prevents the specific injury of learning something significant from a third party.
Threshold.What size or type escalates this to a joint decision. Spend under fifty thousand is one person’s call; above it is both. Thresholds are what make single-name ownership acceptable, because they bound the downside.
Work through the inventory assigning all four. Expect it to take an hour or two, expect two or three genuine disagreements, and treat those disagreements as the whole point — they are the arguments you were otherwise going to have later, under pressure, with a live decision attached.
Step three: the deadlock rule
Some decisions genuinely need both of you, and you will sometimes not agree. Deciding what happens then, while calm, is the highest-value thirty minutes in this exercise. Deciding it during a deadlock is another fight.
Several workable mechanisms:
Domain tiebreak.Whoever owns the domain closest to the decision has the final call, even on a shared decision. Simple and fast, and it means each of you loses some decisions you feel strongly about — which is the cost of having a rule at all.
Reversibility test.For reversible decisions, the person who will execute decides and you review the outcome. For irreversible ones, no decision without agreement, and if you cannot agree, the answer is no. This is a good default for most pairs.
Time-boxed escalation.If unresolved after two attempts, it goes to a named third party — a board member, an adviser, someone you both respect. Name the person now, not during the deadlock.
CEO decides.If one of you holds the title, say plainly that it carries the final call on shared decisions. This is honest and worth writing down, because the ambiguity of a CEO who does not actually decide is worse for both of you than either clear alternative. It also needs an explicit commitment about how the other founder’s dissent is heard — otherwise it reads as a demotion.
Whichever you pick, write the reasoning next to it. In eighteen months one of you will be on the wrong side of it, and the reasoning is what makes it survivable.
How to write this in an hour
The most common failure is treating this as a project. It is a single session and a one-page document.
Block two hours, away from the office, phones down. Bring the decision inventory drafted separately — each of you writes your own list first, then compare. The differences between your two lists are more informative than the overlap, because they show you where you have been operating on different assumptions.
Then work down the combined list assigning the four roles. Move fast, mark disagreements, and come back to them rather than getting stuck. Aim to leave with the whole list assigned and two or three items flagged as unresolved — that is a good outcome, not an incomplete one.
Write it up the same day in a shared doc, plainly. One line per decision. No preamble, no governance language. Then send it to your leadership team, because a document your executives have not seen does not change how decisions actually get made — people keep going to whichever founder they expect to get the answer they want.
And set a review date. Six months is about right at growth stage.
Keeping it alive
A decision-rights document that nobody references becomes archaeology within a quarter. Three habits keep it real.
Use it in the moment. When a decision comes up and the routing is unclear, open the doc in the room. That is not bureaucracy; it is the entire return on the exercise, and doing it visibly a few times teaches everyone the document is live.
Hold to it when you disagree with the outcome. The first time your cofounder makes a call inside their rights that you think is wrong is the test. Overriding it once costs you the whole document, because you have demonstrated that the real rule is that you decide.
Revisit on a schedule rather than in response to conflict. Roles that fit at forty people will not fit at a hundred. Founders who renegotiate every year rarely reach a crisis, because they never accumulate two years of drift.
Where Blomma fits
The hard part of this is not the template. It is that the conversation surfaces things neither of you has said — that you have felt cut out of decisions, that you think a domain has outgrown its owner, that the CEO title has never actually meant anything in practice. Those are the items that make founders quietly not schedule the session.
Blomma is an always-on AI career coach with no stake in your company. Use it to draft your own decision inventory honestly, including the items about each other that you would rather leave off. Use it to work out which deadlock rule you can actually live with when you are the one who loses. And use it to prepare the two or three moments in the session where you expect real disagreement — because those are where the value is and where founders tend to defer instead. When the situation warrants a human who has been in a founding pair, you can bring one in..
Decision rights do not remove disagreement between cofounders, and they are not meant to. They remove the second argument — the one about whether you had the right to decide at all — which is the one that damages the relationship rather than the company.