Why Do Startups Fail? 9 Real Reasons, Backed by Data

How many startups actually fail — the real numbers

The honest answer is "it depends which population you're counting." U.S. Bureau of Labor Statistics data on new businesses puts first-year closure at roughly one in five (around 20–21%), climbing to just under half by year five and about two-thirds by year ten. That's every new registered business — coffee shops, agencies, and consultancies included — not specifically venture-scale startups.

For the startups this blog is actually about — funded, growth-oriented, product-led companies — the picture is worse. Harvard Business School research led by Shikhar Ghosh, tracking roughly 2,000 VC-backed U.S. startups, found that about 75% never return investor capital, and a large share of those liquidate everything. That's the population behind the oft-repeated "90% of startups fail" claim: it's approximately true for venture-scale, high-growth attempts, and roughly false for the broader small-business population BLS tracks.

Where the "90% of startups fail" number actually comes from

"90% of startups fail" gets repeated so often it reads as an official statistic, but there's no single government dataset behind it — it's a rounded-up summary of exactly the gap described above: almost all small businesses survive in some form for years, while almost none of the ambitious, funded startups this figure is usually invoked about turn into the outcome their investors underwrote. Both things can be true at once because they're describing different populations.

That distinction matters for how you read your own risk. If you're bootstrapping a services business, the BLS numbers are closer to your actual odds. If you've taken venture money and are underwriting a 10x outcome, the HBS numbers — and the roughly 25% base rate of even partial investor payback — are the more honest reference class.

The reasons behind the reasons: what actually kills companies

CB Insights has run one of the longest-standing postmortem projects in the space, coding the stated reasons founders give when they shut a company down. Across its various report cycles, "no market need" — building something nobody was willing to pay for — has consistently ranked as the single most commonly cited reason, showing up in roughly four postmortems out of ten. "Ran out of cash / failed to raise more" is usually the next most common, cited in something like three postmortems out of ten.

Read literally, that ranking makes it sound like founders die of two separate diseases: a product problem and a finance problem. In practice they're usually the same disease at different stages. A company that built the right thing rarely runs out of cash mysteriously — it either finds a market and the cash follows, or it doesn't, and the cash clock just becomes the visible deadline for a decision that was already made months earlier.

Running out of cash is usually the symptom, not the cause

Founders talk about failure the way a death certificate talks about cause of death — the last thing that happened, not the disease that produced it. "We ran out of runway" is almost always technically true and almost never the useful answer, because the real question is why the metrics that should have refilled the tank — retention, expansion revenue, a fundable growth curve — never showed up in time.

That's why runway discipline and PMF discipline are the same conversation, not two separate ones. Extending runway without fixing whatever is keeping product-market fit out of reach doesn't save a company — it postpones the same ending by however many months of cash you bought yourself, unless something about the underlying trajectory actually changes in that window.

The 9 ways a startup actually dies

Real postmortems and structured failure taxonomies both converge on the same short list once you strip out the specifics of any one company's story. Founder Runway's simulation models that same list as nine distinct failure paths, because "ran out of cash" is the only one most founders plan against — the other eight end just as many runs, in the game and outside it.

The failure modes founders don't see coming

Ask a founder mid-raise to list ways their company could die and most lists start and end with "we run out of money." It's the visible one — there's a bank balance you can watch tick down. The other eight are quieter, and that's exactly why they're the ones that catch people off guard.

Founder burnout rarely announces itself as a single event; it's a string of decisions made on too little sleep and too much adrenaline, each individually defensible, that compound into a founder who can no longer make the next hard call well. A cap table doesn't collapse in one meeting either — it erodes one option-pool top-up and one uncapped bridge note at a time, until a founder who started with a healthy ownership stake discovers, two rounds later, that they no longer control the outcome they're supposedly building toward.

Investor trust works the same way in reverse: it's not damaged by one bad quarter, it's damaged by one founder who reported the bad quarter as a good one. All three are survivable in isolation and often fatal in combination — which is exactly how they show up in a run, and in a cap table.

The signals that show up before the ending does

Every one of the nine failure modes above has a leading indicator, and none of them is "the bank balance hit zero." Burn multiple climbing while growth stays flat is the earliest tell that cash is being spent faster than it's buying anything durable. A PMF signal that plateaus for two consecutive review periods, rather than continuing to climb, is worth treating as a stop sign, not a plateau to push through with more spend.

Investor trust and founder energy are the two metrics people track the least and regret not tracking the most — by the time either one is visibly low, it's usually been eroding for a while. The founders who catch their own failure mode early are rarely the ones with better instincts; they're the ones who made a habit of checking the quiet metrics on a schedule, not just the loud one.

Practice the decisions, not just the postmortem

Reading about why startups fail is useful the way reading a competitor's fundraising deck is useful — it tells you the shape of the risk without making you feel any of the pressure that produces bad decisions in real time. The nine failure modes above are the ones Founder Runway's 20-turn runs are built around, precisely because they're the real list, not a game-design invention.