The short version
- Name the decider before the discussion. An open question with no name against it stays open at the speed of whoever keeps raising it.
- Argue somewhere with a history. A decision only two people can see gets made a second time by whoever touches the work next.
- Split proposing from judging. Run them in one sitting and the confident half wins, the same failure we design around in content production.
- The same argument arriving twice is a request for a rule. Nine standing editorial rules exist on one of our accounts for exactly that reason.
A disagreement between founders ends when somebody’s name is on the decision and the conversation happens somewhere you can both search later. Without a named decider it reopens every few weeks. A private argument leaves two versions of what was agreed, and the one that ships is whichever of you builds first.
Why does the same founder disagreement keep coming back?
Because nobody said whose call it was. Both founders can veto and neither has to decide, so the question survives every conversation and returns in whatever week one of you has the energy for it.
The question that fixes it is the one that finds a blocked launch. A date moved twice on one engagement because two approvals nobody had listed sat with people who had never been in the room, and the kickoff had asked who is involved rather than who can stop this. Ask the second question about the decision itself.
Five ways a founder disagreement fails, and the mechanism against each:
| The failure | How it shows up | The mechanism |
|---|---|---|
| The argument happens in a private thread | Each of you remembers a different outcome | One shared channel, decisions written where anybody can search |
| Nobody said whose call it is | The question reopens every few weeks | A name against the decision before the discussion starts |
| Proposing and judging in one sitting | The more confident position wins | Two passes on separate days, propose then attack |
| The same argument arrives twice | A third one gets scheduled | Write the rule and stop arguing the instance |
| No date on the open question | We will come back to it | One owner and a date in the same sentence |
Where should the argument actually happen?
Somewhere with a history. We run one shared channel per relationship rather than scattered private threads, because a decision only two people can see is not searchable by whoever touches the work next, who then makes it again on their own judgement.
The exception is anything about a person, which belongs in a call and stays private afterwards. Everything that changes what gets built goes where the team can read it, including the reasoning. Neither of your memories is a record.
Why does the more confident position usually win?
Because proposing and judging happen in the same sitting. We split those deliberately in content production: generation is cheap and confident, verification is slow and adversarial, and running them together means the confident half wins.
The same split works between two founders. Put the option on the table with what it costs and what it wins, then stop. Attack it on another day, your own included. Whoever argues better in the room stops being the deciding variable.
What do you do when the same argument arrives twice?
Write the rule. Nine standing editorial rules exist on one of our accounts because the same notes were being given twice, and they were written at issue 20 of the newsletter rather than issue 1. A rule closes the whole class of argument.
Rules always get written later than they should be, and issue 20 is the normal order as well as the expensive one. The trigger is the second occurrence, because waiting for something undeniable means relitigating it in the meantime.
An argument that ends with both of you feeling heard and nothing written down has not ended.
What should a resolved disagreement leave behind?
An action item, not a feeling. We turn call recordings into action items rather than notes, because the transcript is raw material and the action item is the deliverable. The same conversion applies to an argument between founders.
The record carries four things: the decision, the name against it, the date, and what happens if nobody does anything. That last one decides the most, because the default in an unresolved disagreement is whatever the busier founder was already doing.
The two habits underneath this are covered in why decisions in a DM do not exist and disagreeing without damaging the relationship.
Frequently asked questions
How should co-founders handle a disagreement?
Name whose call it is before the discussion starts, hold the conversation somewhere both of you can search later, and separate proposing an option from attacking it by putting them on different days. Run those passes together and the outcome tracks whoever is more confident.
What do you do when co-founders cannot agree on a decision?
Give the decision a single owner and a date. An open question with no name against it stays open at the speed of whoever is willing to keep raising it.
Should founders disagree in front of the team?
The argument about the work belongs where the team can read the reasoning, because whoever implements the decision needs to know why it was made. Anything about a person stays in a call and then in private.
How do you stop the same argument from coming back?
Write a rule the second time it appears. Nine standing editorial rules exist on one of our accounts because the same notes were being given twice, and they were written at issue 20 of the newsletter rather than issue 1. A rule closes the class.
Where should founder decisions be recorded?
Somewhere with a history that survives people joining and leaving, in a form that carries the decision, the owner, the date and the default if nobody acts. A decision only two people can see gets remade by whoever picks the work up.
When is a co-founder disagreement a real problem?
When it recurs without a decider, or when it is only ever held privately. Both mean no mechanism is attached, so the outcome gets settled by energy and timing rather than by the argument.
Sources
- Chua Network delivery data across 8 client accounts (internal fact bank)
- Chua Network engagement records, anonymized (internal experience bank)
