Complementary skills matter less than commitment, values, communication and how someone behaves when things go badly.
12 min read · Level 1
You will assess skills because skills are easy to assess, and skills are the least predictive thing on the list. Almost every co-founder breakup traces back to something nobody discussed: what outcome would make each of you happy, how you each behave in a bad month, what you would refuse to do to win. Those conversations feel presumptuous at the start and impossible in a crisis, which is exactly the wrong way round.
What actually matters, in rough order:
Commitment — how much risk are they genuinely taking? Values — what would each of you refuse to do? Communication — can you disagree productively? Response to stress — how do they behave in a bad month? Ambition alignment — is one of you building a $5M business and the other a $500M one? Complementary skills — useful, and the most overrated item on this list Time availability — full-time, or evenings?
Because skills are the easiest thing to assess and the least predictive. You can hire skills. You cannot hire someone out of a fundamental disagreement about what the company is for.
Almost every co-founder breakup traces back to one of the first five items — usually ambition alignment or how someone behaves under pressure — and almost every co-founder selection is made on the sixth.
Two technical founders, both excellent engineers, complementary specialisms. Perfect on paper.
Eighteen months in, an acquirer offers $8M. Founder A wants to take it — it is life-changing money and they are tired. Founder B is insulted by the offer and wants to raise and go for a hundred times that.
Neither is wrong. But they never discussed it, so the disagreement arrives at the highest-stakes moment possible, with lawyers in the room. The partnership does not survive it.
One conversation in month one — 'what outcome would make you happy? what would you turn down?' — would have surfaced this entirely. It costs an hour.
First-time founders often assess technical ability and personal liking, and skip everything else because the other conversations feel awkward or presumptuous.
The awkward conversations are the entire point. If you cannot discuss equity, ambition and failure with someone before starting, you certainly cannot discuss them during a crisis.
Are they leaving their job? When? What would make them stop? Vague commitment now becomes absent commitment later.
'Would you sell for $8M in two years?' A mismatch here is fatal and is invisible until it is urgent.
Disagree with them on purpose about something real. You are looking for whether they engage or avoid.
How they describe a failure tells you how they will describe yours — and how they behaved when it was happening.
What would you refuse to do to win? Ask for concrete examples, not principles.
Last, not first. Necessary but the least predictive of whether the partnership survives.
Before any commitment, and definitely before signing anything.
Do not run this as an interrogation with someone you have worked alongside for years — you already have the data. It matters most with someone you know socially but have never worked with.
A recurring pattern: two technically complementary founders receive an acquisition offer that is life-changing for one and insulting to the other. Neither is wrong. The disagreement is unresolvable because it is about what the company is for, and it surfaces for the first time with lawyers in the room.
Ask "would you sell for eight million in two years?" in month one. It takes a minute and it surfaces the only disagreement that cannot be negotiated later.
Self-reported conflict style is close to worthless — everyone believes they handle disagreement well. The founders who assess this properly deliberately disagree about something real during the evaluation period and watch whether the other person engages, avoids, or concedes to keep the peace.
You are not testing whether you get along. You are testing what happens when you do not, which is the condition you will spend years in.
Create a free account to read the rest and apply it to your own startup.