Y Combinator is the most prestigious accelerator in the world. Airbnb, Stripe, Dropbox — the list of companies that passed through its programme reads like a history of the modern internet. By any measure, YC is the gold standard.
Here is how they select startups.
93% of applicants never make it past step one. Let that sink in.
A thought experiment
You hire a builder to renovate your kitchen. You reject them because their van isn't clean. That's exactly what a 1-minute video proves — that someone can produce a 1-minute video. A polished pitch deck proves someone can design slides. Neither proves they can build a company.
Now name one other high-performance selection process that works this way.
Football academies assess players over years. They watch how a player responds under pressure, in different positions, against different opponents, at different stages of a long development arc. Olympic selection runs for entire cycles — four years of qualifying events, trials, and performance data before a team is finalised. Special forces selection takes months and is deliberately designed to surface how a person behaves when exhausted, uncertain, and under sustained pressure — not how they perform on a good day with a rehearsed script.
In every high-performance domain, there is a strong, deliberate correlation between selection rigour and cohort quality. The selection process is designed around what actually predicts performance. Every single one.
What the results say
YC's cohorts are not failures. But look at the aggregate outcomes honestly, even after the most rigorous selection process in the startup world:
That last figure is the most revealing. Half of accepted companies had been rejected before — turned away when they were ready, accepted later when they had learned to play the application game. The process selected for application skill, then got lucky when some of those applicants turned out to also be good founders.
A football manager with these numbers would not be called a gold standard. They would be looking for a new job.
A constraint dressed as a methodology
This is not a criticism of Y Combinator specifically. YC reviews over 30,000 applications per batch. They cannot spend 40 hours per company — the economics are simply impossible. So they built a process around what they could afford: a short application, a short video, a short interview. And because YC's results are better than most alternatives, the ecosystem came to regard their constraints as wisdom.
"We can spot talent in ten minutes" is not a confident claim. It is a resource limitation wearing a suit.
The proof is not in the theory — it is in the results. Forty percent failure rates, despite the world's most competitive application filter, suggest that the filter is selecting on the wrong signal.
What a different approach looks like
What if every applicant — all 30,000 — received a structured assessment across sixteen dimensions, with every claim graded on evidence? Not on presentation skill. On what has actually been built and proven. A report with the depth and detail that would take an analyst over 40 hours to produce.
To be clear: the founder still does the work. They provide the input, they validate the claims, they demonstrate what they have built. No different from a player showing up and performing on the pitch. The difference is that the scout no longer has to guess from a 1-minute highlight reel.
The assessment generates the depth. The founder earns the signal. The investor reads the evidence.
Sceptical? Good. Go see what the output actually looks like. thestartupmentor.ai