Y Combinator's Most Successful Partners

There's a number on Garry Tan's profile page at Y Combinator that I keep rereading. The companies he's worked with are worth a combined two hundred and twenty-six billion dollars.

One guy.

One partner. And right under it: he has read more than six thousand YC applications.

Six thousand? That's a full-time job just reading.

Which raises the question I want to answer today. These partners are treated like oracles... but is it the picking that makes them great, or is something else going on?

Okay, hold on. Before the mystery. What even is a YC partner? Like a VC in a quarter-zip?

Not quite. Here's the rule that defines the whole place: every YC general partner founded a YC-funded company before. Every single one.

So no MBAs.

Paul Graham said it out loud back in twenty eleven. He told TechCrunch that most VCs are MBAs, whereas YC partners are mostly entrepreneurs, and that they're, quote, looking for people like us.

That's either beautifully honest or a little bit terrifying.

Look at the actual roster and it's wild. Tom Blomfield founded GoCardless, then co-founded Monzo, which raised five hundred million pounds and reached ten percent of the UK population. He was awarded an OBE in twenty nineteen.

An actual OBE. From the Queen.

Diana Hu sold Escher Reality to Niantic and shipped the augmented reality tech for Pokémon GO to over a hundred million users. She grew up in Chile. Tyler Bosmeny sold Clever for five hundred million dollars; it's used by sixty percent of American students.

So they've all been on the other side of the table.

And then David Lieb, who built Bump, an app a hundred and fifty million people used to swap contacts by literally bumping phones. Google bought it. He then co-founded Google Photos, which has over a billion users.

Bump. I had Bump. I bumped phones with people.

Everyone did. And Garry Tan's path is the strangest. Before Posterous, he was an early designer and engineering manager at Palantir. He designed the company logo.

Wait, the Palantir logo guy runs Y Combinator now?

He's president and CEO. He was a partner from twenty eleven to twenty fifteen, left, co-founded Initialized Capital, and came back. In between he built Bookface and the Demo Day website, the actual plumbing founders use.

Okay but here's my skeptical question. Two hundred and twenty-six billion. Is that him being brilliant, or him standing near Airbnb?

That is exactly the right question. Because there's an analysis of the YC portfolio that found four companies account for more than eighty-five percent of YC's returns to date. Airbnb, Coinbase, Reddit and Instacart.

Four. Out of how many?

Roughly five thousand in that study. One dataset counts nearly six thousand companies across fifty batches, and only twenty-three have gone public. That's about four-tenths of one percent.

So the entire mythology rests on four names.

And YC itself admits the picking is imperfect. Graham told that same TechCrunch audience that they keep track of the successful companies they rejected. Robert Morris, the MIT professor and YC co-founder, was a notoriously harsh grader. He gave one application a C, which sank it... and the company went on to succeed.

What did they do about it?

Now they double-check every application Morris gives a low grade to. That's the system. A literal patch for one partner's taste.

They version-controlled a human being.

So here's the payoff to your question. The partners who succeed aren't mystics. The edge is mostly in what they look at. Graham said that with their three biggest companies, Airbnb, Dropbox and Stripe, the decision was largely about the co-founders and how well they got along. In an interview that lasts eight to ten minutes.

Ten minutes to judge a relationship?

He said one guy literally rolled his eyes while his co-founder was pitching. Surefire tell.

They rejected him over an eye-roll?

They bet against him. And remember Graham told Drew Houston to go get a co-founder when he applied in two thousand seven. Houston recruited Arash Ferdowsi in two weeks. That's Dropbox.

So the famous advice was... find a person you don't hate.

And the numbers say the model works, even if nobody can pick the winner. Around four and a half percent of YC companies become unicorns, versus about two and a half percent for other venture-backed seed startups. More than half are still alive after ten years, against a thirty percent average.

So they're not better at prophecy. They're better at survival.

That's the whole thing. Aaron Epstein, another general partner, sold Creative Market to Autodesk, then spun it back out as an independent startup in twenty seventeen. His page says he's read more than eight thousand applications and his companies are worth fifty-seven billion.

Eight thousand. These people have read more startup pitches than most humans have read sentences.

And that's the lesson you can actually use. Nobody at YC knows which one is Airbnb. What they have is volume, a system that catches their own blind spots, and a stubborn focus on the two or three people in the room rather than the slide deck.

Judge the band, not the song.

Right. Because the idea will change six times before it works. The people in the room are the only part of the bet that stays.