Startups rarely die for lack of ideas. They die because they built something not enough people wanted to pay for, and found out too late, with the runway almost gone.
CB Insights has tracked this pattern for years. In its analysis published in March 2026, the firm examined 431 VC-backed startups that shut down since 2023. "Ran out of capital" shows up in 70% of cases, but the report itself notes that this is almost always the final cause of death, not the root problem. The causes that explain why the money dried up come next: poor product-market fit (43%), bad timing or macro conditions (29%) and unsustainable unit economics (19%).
The earlier version of the study, based on 101 post-mortems written by founders themselves, reached a similar conclusion by a different route: building a solution looking for a problem, not serving a real market need, was the most cited reason, in 42% of cases. Different sample, method and period. Same message.
Money doesn't buy market fit
The 431 companies in the 2026 analysis raised a combined US$17.5 billion before shutting down, with a median of US$11 million per company. At the top of the list are Olive, a healthcare AI automation company, and Convoy, a digital freight brokerage. Each raised close to US$1 billion, both reached valuations of around US$4 billion during the pandemic boom, and they shut down within two weeks of each other in October 2023.
Neither lacked ideas, capital or talent. The report also shows that poor fit isn't only an early-stage problem: two-thirds of the product-market fit failures were early-stage companies that never found a market, but 20 companies at Series B or later also cited it as a primary cause. Growth doesn't prove fit. Sometimes it only delays the discovery.
Where validation usually breaks down
Conversations that only confirm what you wanted to hear
"Would you use this?" is the most common validation interview question and one of the least useful. People who like the founder say yes to be polite. Rob Fitzpatrick wrote a whole book on this problem, The Mom Test, built on one central idea: ask about the customer's life and what they did in the past, not about your idea and what they would do in the future.
In practice, replace "would you pay for a tool that automates bank reconciliation?" with "how did your last month-end close go? How long did it take? What have you already tried to fix it? What did that cost?". If the person never tried to solve it, never spent money on the problem and can't remember the last time it hurt, the pain probably isn't big enough to turn into a purchase.
Validating the idea and forgetting the business model
People liking the product doesn't mean the math works. Unsustainable unit economics shows up in 19% of the shutdowns in CB Insights' analysis. It's the product customers love but that costs more to acquire, deliver or support than they pay. Before scaling, you need to know what it costs to win a customer, how much they pay over time and what it costs to serve them. A rough spreadsheet estimate already rules out many models that don't add up.
Growing before measuring retention
Top-of-funnel growth hides a retention problem for a while. If customers arrive and leave, acquisition cost keeps rising until the day the next round doesn't come.
One simple way to gauge fit is the test Sean Ellis proposed: ask users how they would feel if they could no longer use the product. According to First Round Review, Ellis compared answers across nearly a hundred startups and found a cutoff: companies with strong traction almost always had more than 40% of users answering "very disappointed", while those struggling to grow almost always fell below it. The same article tells how Superhuman measured 22% in the summer of 2017, segmented its users, rebuilt its roadmap around what the most engaged ones valued, and reached 58% within three quarters.
The test isn't a law. It's a cheap thermometer you can run with your first users, and it beats the founder's gut feeling.
A six-week validation plan
If you don't have a product yet, or you have an MVP with little traction, this sequence works as a starting point:
- Week 1: write down the assumptions that would kill the business. Who has the problem, what it costs them today, what they would pay, which channel reaches them. Each assumption needs a signal that confirms or refutes it.
- Weeks 2 and 3: talk to potential customers about their past. Use Mom Test-style questions. Record what they did, not what they said they would do.
- Week 4: ask for a real commitment. A pre-sale, a letter of intent, a paid pilot, a meeting with whoever signs the contract. Compliments aren't commitment; time, reputation or money are.
- Week 5: deliver the value by hand if you have to. Before building the software, solve the problem manually for three or four customers and see if they come back.
- Week 6: decide. Based on the signals, continue, adjust the audience or the problem, or stop. Write down why.
Signals that count and signals that mislead
They count: a customer who pays before the product is ready, a customer who comes back without a reminder, a customer who refers someone without being asked, someone who reshuffles their calendar to talk to you about the problem.
They mislead: compliments at events, likes on a launch post, a waitlist with no commitment, "send me a proposal and I'll take a look", investors excited about the market, and paid growth that stops when the ad budget runs out.
It applies to large companies too
Corporate innovation repeats the same mistake under another name. The new digital product, the internal app, the customer platform: many start from one executive's conviction and get approved without anyone asking users what they did the last time they had that problem. Validation costs weeks. A product nobody uses costs the year's budget.
The Sean Ellis test has an internal version too: ask pilot users how they would feel about going back to the old process. If most say they wouldn't mind, the pilot hasn't solved anything important for them yet, however many daily logins the adoption dashboard shows. Mandatory usage isn't fit.
Next step
Write down today the three assumptions that, if wrong, would close your company or your project. Next to each, note what evidence you have now and what evidence you would need. If the "still missing" column is fuller than the "already known" one, your next task is validation, not development.
If you want someone experienced working through these decisions with you every week, that's what Advisor As a Service is for. Once the assumptions are validated and it's time to build, MVP As a Service delivers only what's needed to prove traction and hold up in conversations with investors.