Former colleagues first, then communities and events. The best predictor is having worked together before.
10 min read · Level 1
The strongest predictor of a co-founder relationship working is having worked together before, and it is not close. Everything else is inference from a small sample taken while both of you were on best behaviour. Meeting someone at an event and committing three weeks later is not a search, it is a coin flip with your company on it — which is why the trial project exists and why skipping it is the expensive decision.
In rough order of how well they work:
Former colleagues — you have already seen them work Your professional network — one degree away, with references you trust Founder communities and accelerators — people who have already chosen this life University connections — especially for younger founders Events and hackathons — you see them build something under pressure Co-founder matching platforms — real, but the highest variance
The pattern: the higher on this list, the more evidence you already have.
Because a co-founder decision is a bet on how someone behaves over years, and the only reliable evidence is having seen them work.
Meeting someone at an event and committing three weeks later is how most bad partnerships form. It is not that strangers cannot work — it is that you have no data, and enthusiasm feels like data.
Best case. You worked with them for two years. You have seen them under a deadline, seen them wrong, seen them handle a difficult client. You are not guessing.
Common case. You meet at a founder event. You get on well, the skills fit, you are both excited. This is a first date, not evidence. Do a trial project before committing anything.
Worst case. A friend you have never worked with, who is enthusiastic and available. Social compatibility says almost nothing about working compatibility — and if it fails you lose the company and the friendship.
The fix in every case except the first is the same: work together on something real before you commit equity.
First-time founders often treat looking for a co-founder as a search that must conclude quickly, because building alone feels like stalling. It is closer to hiring for the most senior role in the company — and taking three months is entirely reasonable.
The other mistake: not telling people you are looking. Most good matches come through someone who knew both of you and thought of it.
List everyone you would work with again. That list is your first and best source.
Not 'looking for a co-founder' but 'looking for someone who has built payments infrastructure and wants to work on logistics'.
Accelerator programmes, founder communities, industry meetups. People there have already accepted the risk.
Same discipline as hiring. Talking to one person and committing is not a search.
Non-negotiable with anyone you have not worked with. The next lesson covers how.
Once you have decided you want one and know what you are looking for.
Do not pause building while you search. Keep making progress — it also makes you far more attractive to a good co-founder than a search does.
The dominant pattern across well-known companies is co-founders who were previously classmates, colleagues or long-standing friends who had worked together. Accelerators run matching programmes precisely because forming a pair from strangers is known to be hard, not because it is the normal route.
Start your search with everyone you have worked with and would work with again. That list is short, and it is far better evidence than any interview.
Founders frequently pause progress to search for a co-founder, which makes the search harder — a company with momentum attracts far better partners than a founder with a plan. The people worth having are choosing between options, and visible traction is the strongest thing you can offer them.
Do not stop shipping to go looking. Progress is both the reason a good co-founder joins and the thing that makes waiting for one survivable.
Create a free account to read the rest and apply it to your own startup.