What we look for in a co-founder pair

Co-founder pairs that fail mostly fail in predictable ways. The strong ones share a small set of properties that, in our experience reading hundreds of applications and watching dozens of companies up close, separate them durably from the weak ones. This essay is the list, with the failure modes attached.

Louis and I have been working together for ten years. We have made every mistake on this list, in some form, in our own partnership. The list is not theoretical. It is a working diagnostic that has helped us judge other pairs and, more importantly, repair the cracks in our own.

The five properties

One: complementary specialisation, single shared standard. The strong pairs we have seen specialise in different parts of the work. One person leads engineering, the other commercials. One inside the product, the other inside the market. The specialisation is the easy part. The single shared standard — the same definition of what "good" looks like in both halves of the business — is the hard part. A pair where the engineer thinks the commercial work is sloppy, or the commercial lead thinks the engineering is over-built, will eventually fracture. The shared standard is what holds the specialisation together.

We test for this by asking each co-founder, separately, to describe what excellent work looks like in the other person's domain. If the engineer can articulate what excellent sales looks like, and the commercial lead can articulate what excellent code looks like, the shared standard is there. If both can only describe their own domain with depth, the partnership has a structural weakness.

Two: durable trust over short-term agreement. Strong pairs disagree visibly. We have watched two co-founders argue, in a room with us present, about whether a deal should be £40,000 or £60,000, with both of them clearly aware that the other was unhappy with their position, and we have watched both of them say "okay, your call" five minutes later when one of them was responsible for owning the decision. That is the marker.

Weak pairs either avoid the disagreement entirely (which builds resentment) or handle the disagreement well and then carry the loss into the next decision (which slows everything down). The durable trust marker is being able to lose an argument cleanly and not bring it into the next one. We watch for this in the way teams describe their last disagreement during interview.

Three: separable equity, joint psychology. The cleanest pairs we have seen separate the legal of who owns what (equity, decision rights, vesting) from the psychological of how they operate (shared narrative, shared pride, shared shame). Pairs that conflate the two — "I have 51% so I have the final say on this product call" — are weaker than pairs that hold them separately — "I have the final say on this product call because we agreed I would, not because of the cap table."

The fastest way we have found to test this is asking how the equity split was decided. The strong pairs answer with the negotiation, not with the outcome: how they sat down, what they each argued for, what each was willing to concede. The weak pairs answer with a percentage and no story.

Four: asymmetric ambition along the same axis. The myth that good co-founders are "perfectly aligned" on ambition is misleading. The actual property is more subtle. The strong pairs are aligned on the axis of ambition — they want the same kind of outcome — but often asymmetric on the level. One wants the IPO; one wants the £50m acquisition. One wants to run the public company for a decade; one wants to start the next thing in five years. That is fine, as long as the axis is the same. The two of them are not pulling in different directions; they are pulling in the same direction at different intensities.

The pairs that fail are the ones with different axes of ambition. One wants the lifestyle business; the other wants the venture-backed rocket. One wants impact; the other wants money. These pairs can co-exist for two years and then violently disagree in year three over a single decision that surfaces the axis difference. Hard to repair from there.

Five: shared history of finishing something hard. This is the property we weight most heavily in the application read. Strong pairs have finished something together — a hackathon, a delivery under pressure, a school project that ran four times longer than planned, a side-business that lost money but shipped to twenty customers anyway. The shape of the thing matters less than the finishing.

Pairs that have only ever started things together — a Discord server, a half-built side project, three meetings about a future startup — have not been tested yet. We will not back a pair that has not been tested. We will encourage them to go and ship something small together before they apply. A four-week side project is enough. The output is not the point. The reps in the partnership are the point.

Three common failure modes

Now the inverse: the three failure modes we see most often in applications we end up rejecting on team grounds.

Failure mode one: the silent senior partner. One co-founder writes the application. The other is mentioned but never described. Their LinkedIn is linked. They will be at the interview. They have not, in any way visible to us, contributed to the thinking in the application. This is not always a sign of weakness — sometimes the silent partner is the technical engine and the writing partner is the commercial lead — but it is always a sign worth investigating. We probe hard at the interview by asking the silent partner to describe a recent decision they made independently.

Failure mode two: the freshly-matched pair. Two strong individuals who met three months ago at a hackathon, fell into easy collaboration, and decided to apply. The individuals are often genuinely impressive. The pair has not yet been tested. We almost always say no to these pairs, with the rejection letter explaining that the next thing to do is not to find a different accelerator, but to ship a small thing together for six months and reapply. About one in three come back as a tested pair, and most of those get a yes.

Failure mode three: the friendship pair without the partnership. Two people who are great friends but who have not yet figured out how to be partners. The conversational dynamics are easy. The professional ones are unbuilt. We watch in interview for the moment when one of them defers to the other on something they should have an opinion on — the silent yielding that suggests the relationship is structured around social comfort rather than around the company. This is a hard rejection because the friendship is genuine. It is also a structural weakness that we cannot fix in twelve weeks.

What this means if you are still finding your co-founder

If you are reading this and you are looking for a co-founder, the takeaway is find someone to ship a small thing with for six months, not someone to apply to an accelerator with next month. The ship-something-small is the partnership formation. Without it, the partnership does not exist.

We are aware this is annoying advice if you have just had a great two-hour coffee with someone who feels like the right co-founder. The two-hour coffee is real signal. It is not enough. Ship something. Then come back.

What this means for the partnership you are in

If you are reading this and you are already in a co-founder partnership, the diagnostic is a one-hour conversation. Three questions, asked seriously, with the answers honest:

1. What does excellent work look like in my domain — through your eyes?
2. What was the last thing we disagreed on, and how did we resolve it?
3. Do we want the same kind of outcome from this company in seven years?

The strong pairs can answer all three in detail. The weak pairs answer two of three. The weakest pairs answer one of three or none. The ones that answer none are the ones we usually meet six months later when one of them is looking for a new co-founder.

The diagnostic is not a death sentence. We have repaired our own partnership several times by sitting down with the three questions and being honest about the answers. The repair is cheaper than the breakup, dramatically.

Why this matters for the Moonlabs Incubator specifically

Operator-led incubators like ours are uniquely exposed to co-founder dynamics because we are in the room every week. A cracked partnership is something we will see, and something we will have to address. We do not pretend to be marriage counsellors; we do, however, pay attention to the signals, and we will be honest with the team when we see something that needs to be addressed.

The selection bias on our side is therefore strict. We back pairs that pass the five properties above, on first read. Pairs that pass three of five we will sometimes back with explicit conversation about the gaps. Pairs that pass fewer than three almost never get a yes.

If you are reading this thinking your partnership has gaps, the right move is not to hide them in the application. The right move is to write the application with the gap visible and our honest take on why you are addressing it. We have backed pairs that named their weakness clearly. We have rejected pairs that pretended the weakness was not there. The diagnostic-honest version of the application is, almost always, the one that converts.

James Freestone is co-founder of Moonlabs with Louis O'Connell-Bristow. The Moonlabs Incubator accepts both solo founders and pairs; pairs are evaluated jointly on the partnership properties described above.