
"90% of startups fail."
You've heard that line so many times it's basically elevator music. It's also almost useless for making real decisions about your company.
What matters is where they fail, when they fail, and which kinds of companies are dying vs surviving.
"90% fail" is like saying "most humans die." True, but not helpful.
If you want clarity as a founder, you need to think in survival curves, not a single big failure number. The game is a series of gates: each funding stage, each year of survival, each vertical has its own odds. Once you see those gates, the world stops feeling like a vague death sentence and starts looking like a map.
The 90% number mashes everything together: bootstrapped and venture-backed. SaaS, biotech, hardware, games, crypto. Dead-in-18-months and slow-profitables that never "exit."
Lumped together, it's noise.
The more useful way to think about it is conditional survival: "Given I made it to X, what are my odds of reaching Y?"
Here's what very rough venture-backed survival looks like (US/Europe, last decade-ish, all verticals mixed):
Chain those together and, yes, the overall odds are brutal. But your real question is not "what are my global odds?" It's: "Given I'm a funded seed SaaS company in 2024, how likely is it I hit Series A, and what has to be true for that to happen?"
That's a very different question than "do 90% fail."
Survival curves are not the same across industries. At all.
Some rough patterns (not gospel, but closer to reality than "90% fail"):
B2B SaaS (good gross margins, recurring) – More companies make it to seed/A, because the story is easy to underwrite. But many plateau after A/B if growth slows below ~2–3x year-over-year. A "serious" seed today might be $1–3M ARR within ~18–24 months, growing 10–20% MoM early on.
Consumer apps / marketplaces – Huge pre-seed/seed graveyard. Many never find repeatable acquisition. The ones that work can raise very fast because top-line user growth is explosive. You see more "zero or rocket ship" behavior vs. SaaS's steadier slope.
Fintech – Harder early (regulation, licenses, underwriting, trust). Surviving to Series B can be very valuable: infra, switching costs, and regulatory moats kick in.
Biotech / deep tech / hardware – Higher early failure from technical risk or funding gaps. But investors know this, so the conditional survival from, say, Series A to exit can actually be better than in generic SaaS. Once a drug hits Phase 2/3 or a hardware platform hits production, the odds profile changes a lot.
Crypto / frontier / "weird stuff" – Survival often depends more on cycles than on your own execution. Time-to-next-round is heavily tied to whether your narrative matches the current hype window.
So you don't want "startup failure odds." You want "survival curve for my specific type of company, in this fundraising climate."
Binary "failed/succeeded" hides the thing that actually kills you: running out of time.
Two companies can both be "alive" at 24 months and still be in totally different universes. Company A: $40k MRR, 8 months runway, growing 12% MoM, clean cap table. Company B: $5k MRR, 2 months runway, flat growth, messy cap table.
Same age. Completely different survival odds.
Better questions than "will I be in the 10%?":
Time to next gate – For your stage and vertical, how long does it usually take to hit the metrics for the next round? 12 months? 24?
Round-to-round bar – What are investors actually funding at that gate right now? (e.g., Seed → A: often $1–3M ARR, 3x+ YoY growth, or very strong usage/retention for non-revenue products.)
Runway vs learning speed – Are you iterating fast enough that each month of burn buys real information about the business?
You want a simple mental model: "Given my runway and current trajectory, how many realistic shots at the next milestone do I have?"
That's a more useful number than 90%.
If you raised $500k from angels and are comparing yourself to a startup that raised $8M seed from top-tier VCs, you'll feel like you're failing even if you're actually ahead for your lane.
Useful cohort filters:
Once you slice by those, survival rates look very different from the spooky 90%. Some cohorts are more like "50% die before seed, another 40% between seed and A, the rest battle it out over a decade."
Use "90% fail" as a mood, not a metric.
For decisions, build your own survival map:
Once you stop fearing a vague 90% and start optimizing for your own curve, fundraising and product decisions get much simpler: "What is the next gate, what's the bar, and what's the fastest way to hit it with the money and time I have?"
This is a functional model you can use to create your own formulas and project your potential business growth. Instructions on how to use it are on the front page.
