Why we say no

We turn down roughly nine in ten applications. Most accelerators would call that a problem. We think it is the most important number we report, because the alternative — filling a cohort because we have to — is what produces the kind of average outcome we are not willing to deliver.

This essay is the actual breakdown of why we say no, in honest categories. We are publishing it because the reasons we reject are also the reasons most accelerators reject, and almost nobody writes them down. Founders who have been turned down once or twice get more value from honest categories than from a polite form letter.

Category one: we do not believe the buyer exists yet (≈30% of nos)

The single most common reason we say no is that the buyer described in the application is aspirational rather than current. The founder believes a future buyer will exist for a future version of the product, and is asking us to underwrite that future.

We are operators. We do not underwrite futures. We underwrite traction against present buyers. Even at the seed stage, even at the idea stage, we want evidence that someone in 2026, today, would pay for a version of the thing being described.

The version of this rejection we send most often: "Your application describes the buyer 18 months from now. We can only fund the company that exists today. Come back when you have spoken to fifteen present-day buyers and can tell us what at least three of them said yes to."

About half the time, those applicants come back six months later with a sharper company, and we say yes. The application that gets the second-time yes is usually the same person, with the same idea, executed differently.

Category two: the founder cannot describe the company in operator language (≈20% of nos)

A surprising amount of weight in our reading goes to the language of the application. Specifically: can the founder talk about the company the way an operator would, or do they talk about it the way a consultant would.

The operator version: "We charge £1,400 a month per conveyancing firm. Three firms have signed pilots. The cost of running the inference is £190 a month per firm. Gross margin is 86%."

The consultant version: "We are revolutionising the legal-tech space with an AI-first platform that empowers conveyancing professionals to deliver higher-quality outcomes for their clients."

Both might describe the same business. Only one of them is run by a person who could survive a Tuesday-morning team stand-up. We back the first kind because the second kind cannot operate the company at the scale the application is asking us to underwrite.

This is a particularly hard rejection to receive, because the founder genuinely does not know they are writing in the second mode. The fix is reading more applications from operators they admire, and forcing themselves to drop every adjective that does not survive the "and so what" test.

Category three: the team has not yet been tested (≈15% of nos)

For team applications — usually two co-founders — we look hard at whether the team has survived a hard week together. Building a company is many consecutive hard weeks. A team that has only had the easy ones in front of them does not yet know whether the partnership will hold under pressure.

We are not looking for past co-founded ventures specifically. We are looking for any evidence of high-stakes collaborative work. A co-founder pair that has built a Year 12 maths competition team together has shown more useful signal than a pair who matched on a hackathon two months before applying.

The version of this rejection: "We like both of you separately. We do not yet know whether you survive each other. Come back with a story of a hard week — a delivery you stayed up for, a disagreement you worked through. Not a curated one. A real one."

Category four: the idea is fashionable in the wrong way (≈10% of nos)

There are categories of company that are fashionable in 2026 in ways that we think are durably wrong. AI-coded T-shirts. AI girlfriends. AI Twitter ghostwriters. Things that ride a current model capability up the curve and have no defensible position when the next model lands.

We say no to these. Politely, often, but firmly. The honest version: "Your company is built around a model capability that will be commoditised within twelve months. We do not believe there is a durable business here. We may be wrong; we have been wrong before. But we are not going to deploy a year of operator time finding out."

The hard sub-version of this rejection is the adjacent application — an AI-flavoured business that we genuinely cannot tell whether it is durable or fashionable. We default to no on these, with a longer letter explaining why, because the cost of saying yes wrongly is larger than the cost of saying no wrongly. Some of those founders will come back having sharpened the durability case. Some will, with hindsight, look at us with regret. Both outcomes are okay.

Category five: the founder is solving the wrong personal problem (≈10% of nos)

A subset of applications are genuinely well-described, but the founder appears to be applying for the wrong reason for them personally. Most commonly: a founder who clearly needs a job, not a company. A founder for whom the equity timeline does not match their personal financial runway. A founder who is leaving a stable role for reasons we suspect are about the current role rather than about the new company.

These rejections are the hardest to write because the founder is usually capable. We just do not think the company-building path will be the right one for them in the next three years. The version we send: "You are clearly competent. Building this company will take four to seven years. We do not see, in the application, the version of you that survives year three on the runway you are describing. Talk to us about the role market first."

A meaningful fraction of these founders come back two years later in a much stronger position, having built independent runway in a senior employment role. We back several of them at that point. The no was right; the eventual yes is right.

Category six: timing is not personal, it is product (≈10% of nos)

Sometimes the company is good, the founder is right, the buyer exists — and the timing of the market is the rejection. We have said no to companies that, two years later, would be obvious yeses, because in the year of the application the underlying tooling, regulation, or buyer readiness had not arrived.

We are usually honest about this. "This is a 2028 company. The model capability you are relying on will not exist at production cost for another two years. The thirteen-month runway you are proposing does not bridge that gap."

The other version: "The regulatory window for this is closing. You are right about the product. You are eight months late to ship before the rule change makes the business uneconomic."

These rejections are not satisfying to write. We send them with as much detail as we can, because the founder deserves to understand the timing argument even if they disagree with our reading of it.

Category seven: we are wrong (≈5% of nos)

We do not pretend our judgment is perfect. Some of the founders we have turned down have gone on to build excellent companies. The version of this we are most proud of is when we have actively introduced founders we have rejected to other investors who would back them, because we believed in the founder even though we did not believe in the fit for the Moonlabs cohort.

This is sometimes called the "second-best yes" — the founder who is not for you, but who deserves to win, and to whom you owe a useful introduction. We try to do this on every no we send. It is the cheapest costless thing we can offer, and the founders we make introductions for almost always remember.

What this means if you have been turned down

If we have said no to your application, three honest things to do.

Read the rejection letter slowly, twice. We try to write them in the same operator-direct voice as the rest of our work. They contain real information about what to change. They are not boilerplate.

Treat the no as data, not as judgment. A no from Moonlabs is not a comprehensive verdict on your company. It is one cohort-fit decision by two people on a Tuesday afternoon. It is one data point, weighted by the things we are good at evaluating and biased against the things we are not.

Come back when the rejection reason has changed. The founders we say yes to second-time-around are almost always the ones who showed us, in the second application, that they understood the first rejection. "You asked me in February to come back with fifteen buyer conversations. Here are the notes from twenty-three." That is the version of an application that almost always converts.

Why we write this down

We write the rejection categories down because we think the honest version of "why we say no" is more useful than any amount of polished marketing about who we are. The selection is the product. The selection has to be honest to work. The honest selection is also the kindest one — to the founders we say no to, and the founders we say yes to.

If you are about to apply, the takeaway is not to game the categories. The takeaway is to write the application that would survive each category honestly, and trust that the application is read carefully. Both halves matter.

Louis O'Connell-Bristow is co-founder of Moonlabs. He reads every application personally alongside James Freestone. We respond to every applicant within a week.