B2B or Consumer? What 1,200 Y Combinator Companies Actually Did

Every founder who has ever sat across from an accelerator partner has heard some version of the same advice: build for businesses, not consumers. B2B is stickier. Enterprise revenue is more predictable. Consumer is a lottery. I gave that advice myself, more than once, back when I was writing angel cheques on gut and pattern-matching — and I gave it the way most people do, as a thing everyone knows rather than a thing anyone had checked.

So we checked it.

What happens to B2B companies versus consumer companies?

We built an outcome graph of more than 6,000 Y Combinator companies — every batch since 2005 — and linked each one to what actually happened next: still operating, acquired, went public, or dead. To ask a fair question of it, you have to look at companies old enough to have an answer. A company from the 2024 batch is mostly still "active" because it hasn't had time to be anything else. So take the 2010–2019 batches, where the story has largely finished playing out, and split them into the two camps the advice is about.

About 1,200 of those companies were tagged B2B or consumer. Here is what became of them:

• B2B (762 companies): ~27% dead, ~32% acquired or public, ~40% still operating.
• Consumer (445 companies): ~43% dead, ~24% acquired or public, ~34% still operating.

The advice holds. A consumer company in this set was roughly 1.6 times more likely to be dead than a B2B company from the same vintage. That is not a rounding difference or a story you can wave away with a couple of counter-examples. Across more than a thousand companies, the enterprise bet is the lower-mortality bet, and the gap is wide.

Which startup sectors die the most?

The B2B-versus-consumer line is really a proxy for something more granular. When you break the same batches into finer categories, the death rates sort themselves into a league table, and it is remarkably orderly:

• Social: ~58% dead
• Consumer: ~43% dead
• Retail: ~40% dead
• B2B: ~27% dead
• Healthcare: ~27% dead
• Infrastructure: ~25% dead
• Fintech: ~21% dead
• Education: ~19% dead
• Recruiting & talent: ~15% dead

Read from the top, it's the consumer-facing world — the things a person opens on their phone at night. Read from the bottom, it's the unglamorous machinery a business pays for on a contract. A social company from this era died nearly four times as often as a recruiting company. Nobody stands on a demo-day stage dreaming of building payroll software. The payroll software is what's still alive.

There's a clean mechanical reason underneath it, and it's the thing the advice is gesturing at without saying: a business that pays you for a contract has told you, in the only language that counts, that the problem is real. A consumer who downloads your app for free has told you almost nothing. Enterprise revenue is a lagging confirmation of demand. Consumer traction is a leading guess at it. Guesses are wrong more often. That is the whole difference, and it shows up as a mortality curve.

So consumer is the worse bet?

Here's where I have to argue against the piece I've just written, because if you stop at the death table you'll draw exactly the wrong conclusion.

Survival is not the same instrument as returns. They measure different things, and the advice quietly collapses them into one.

Look at what actually generates Y Combinator's value — not its survival rate, its outcomes. The names that dominate the ledger are Airbnb, DoorDash, Coinbase, Instacart. A tiny handful of companies account for the overwhelming majority of the public-market value the whole programme has ever produced, and they are, almost to a one, consumer businesses. The category that dies the most also houses the biggest winners. Both facts are true at the same time, and they don't cancel out — they describe the shape of the bet.

This is the part the advice hides. When a partner tells you "go B2B, it's safer," the word doing the work is safer — and safer is a real, measurable, defensible claim. B2B lowers your odds of dying. What it does not do is raise your odds of building something that returns a fund. Those are two different questions. A B2B-heavy portfolio has fewer zeros and fewer moonshots. A consumer-heavy one has more of both. Neither is the smart choice in the abstract; the smart choice depends entirely on which game you signed up to play.

What should a founder actually do with this?

Not choose your market off a death-rate table. That would be its own kind of mistake — building payroll software you don't care about because the spreadsheet said it survives. The best companies in this dataset weren't founded by people optimising a percentage. They were founded by people who couldn't not build the thing.

What the data earns you is honesty about the trade you're making. If you're building consumer, you are playing a game with a higher failure rate and a higher ceiling, and you should raise, spend, and set your own expectations like someone who knows that — more shots, faster reads on whether it's working, no pretending the enterprise playbook applies to you. If you're building B2B, you have a structurally kinder path to survival, and your risk is the quieter one: becoming a company that never dies and never matters. Different failure mode. Same requirement — know which one is stalking you.

The advice was never wrong. It was just answering one question — how do I lower my odds of dying — and letting founders hear it as the answer to a different one. The number tells you which game the market rewards for staying alive. It cannot tell you which game is yours.

A note on what this data can and can't say

Intellectual honesty demands the caveats, because they're load-bearing. Every company here was accepted into Y Combinator — a programme that admits roughly one applicant in a hundred. So these are not base rates for startups in general; they're base rates for startups that already cleared an extreme filter. Selection is doing quiet work in every figure above, and it compresses the range: the very worst ideas never made it into the sample at all. "Dead," "acquired," and "active" are status snapshots drawn from public records, not audited financials. And we deliberately used older batches precisely because the recent ones haven't resolved — which means the picture is honest about the past and silent about whether the same sector odds hold in an AI-reshaped 2026. Probably not exactly. That's the next thing to check.

Frequently asked questions

Is B2B or B2C better for a startup?

On survival, B2B is measurably better: across Y Combinator's 2010–2019 batches, B2B companies died at roughly 27% versus roughly 43% for consumer companies. But on the size of the winners, consumer dominates — the largest outcomes in Y Combinator's history (Airbnb, DoorDash, Coinbase, Instacart) are consumer businesses. B2B lowers your odds of dying; consumer raises your odds of an outsized outcome. They are different bets, not better-and-worse versions of the same bet.

Which startup sectors have the highest failure rate?

In this dataset of accelerator-backed companies, the consumer-facing categories fail most: social (~58% dead), consumer (~43%), and retail (~40%) sit at the top of the death table. The lowest failure rates were in recruiting and talent (~15%), education (~19%), and fintech (~21%).

Why do consumer startups fail more often?

Enterprise revenue is a lagging confirmation that a problem is real — a business signing a contract has demonstrated demand in the only language that counts. Consumer traction, especially free usage, is a leading guess at demand, and guesses are wrong more often. That difference in signal quality shows up, at scale, as a higher mortality rate for consumer companies.

Do these numbers apply to all startups?

No. Every company in this analysis was admitted to Y Combinator, which accepts around 1% of applicants. The figures are base rates for accelerator-selected startups, not for startups in general — selection compresses the range and removes the weakest ideas before they enter the sample. Treat them as directional, not as universal odds.

These figures come from NUVC's accelerator outcome graph — more than 6,000 Y Combinator companies and every Startmate cohort, each linked to what happened next. It's the same outcome data we use to keep NuScore, our AI pitch-deck score, honest: a score is only as good as the record of what actually happened to the companies it once judged. If you want to see how your own startup reads against that record, that's what we built NUVC to do.