We Analyzed 100 Failed Ideas: What They All Skipped

The study

Over several months, we analyzed post-mortems, founder interviews, and documented failure cases from 100 startups and business ideas that didn't survive past 18 months.

The ideas ranged from pre-seed consumer apps to bootstrapped B2B tools to solo-founder side projects. The markets ranged from fintech to healthcare to edtech to consumer goods. The teams ranged from solo founders to 4-person founding teams with previous exits.

One pattern appeared in 78 of the 100 cases. Not execution problems. Not market timing. Not competition.

It was simpler than any of those, and sadder: the founders never validated whether the problem they were solving was real.

The failure taxonomy

Before we get to the dominant pattern, it's worth understanding the full picture of what we found. Failures fell into six main categories:

Category	Cases (of 100)	% of total
Problem wasn't real or painful enough	78	78%
Right problem, wrong solution design	34	34%
Right solution, wrong audience	28	28%
Ran out of money before finding fit	41	41%
Team breakdown	19	19%
Competitive pressure	12	12%

Note: categories overlap — many failures had multiple contributing factors. But the first category is the one that dominated everything else.

What "problem wasn't real" actually means

When we say 78% of failed ideas had a problem that "wasn't real or painful enough," we don't mean the founders imagined the problem entirely. In most cases, the problem existed — it just wasn't painful enough, frequent enough, or prioritized enough for the people experiencing it to change their behavior and pay for a solution.

The distinction matters:

• Real but not painful: "Yes, this annoys me occasionally" vs. "This costs us €5,000 a month"
• Painful but infrequent: A problem that matters when it happens but only happens twice a year
• Painful but low priority: A real problem that falls below the threshold of "something we'd spend money to fix right now"

All three of these look like valid problems in founder conversations. All three lead to builds that nobody buys.

The pattern in the post-mortems

The most striking thing in the failure post-mortems wasn't what the founders did wrong — it was what they skipped. In 78 cases, the founder had done one or more of the following before building:

• Asked friends and colleagues (people with social incentive to be supportive)
• Run a survey (which generated encouraging opinions, not behavioral evidence)
• Looked at existing products in the space (which confirmed demand at a category level, not for their specific approach)
• Built an MVP to "see if anyone would use it" (without conversations to understand whether they cared)

What they almost never did: have direct, structured conversations with 15-20 strangers in their target market about the problem — before writing any code.

The two-week test

Here's what's frustrating about this pattern: in almost every case, proper validation would have been possible in two weeks. Not all founders would have found clear "no" signals — some ideas had real promise that needed refinement. But the ones that were fundamentally flawed showed warning signs that experienced interviewers would have caught in 10-15 conversations.

The warning signs that reliably precede failure:

• Nobody can name a time they experienced the problem in the last month
• Everyone acknowledges the problem but nobody is currently trying to solve it (no workarounds)
• The most enthusiastic responses come from "that sounds interesting for someone else"
• No one immediately says "I need this now" or "I've been waiting for something like this"

What the successful founders did differently

Of the 22 ideas that didn't fit the dominant failure pattern, we found consistent differences:

1. They talked to strangers. Not just people who knew them. Specifically people who had no social reason to be encouraging.
2. They asked about problems, not solutions. The conversations were about how people currently experienced pain — not about whether they'd like the proposed fix.
3. They looked for current workarounds. Before building, they asked: "Are people already trying to solve this with a bad solution?" If yes, strong signal. If no, weak signal.
4. They set a threshold before starting. "If fewer than 10 of 15 people have experienced this problem in the last two weeks, we don't build." Setting the bar in advance prevents retrofitting the evidence to justify the work.

The cost of skipping validation

The median time wasted by the 78 ideas that skipped proper validation: 7 months of building before the founder concluded it wasn't working. At an opportunity cost of roughly €8,000-15,000 per month (conservative estimate of founder time + infrastructure + marketing), that's €56,000-105,000 per failed idea in the dataset.

The cost of doing proper validation: 2-3 weeks of conversations, typically 15-20 hours of founder time total.

The ROI calculation is simple. The behavior change is hard. But the tools available in 2026 — AI-powered contact discovery, automated Mom Test outreach, structured reply tracking — have reduced the friction enough that there's no longer a good excuse to skip it.

One last finding

The most painful cases in our dataset weren't the founders who built nothing and gave up. They were the founders who built for a year, found early traction, raised money, grew a team — and then discovered 18 months in that their core problem wasn't real enough to support a sustainable business.

Everything looked like it was working. The problem was real for a few customers. But it wasn't painful enough, frequent enough, or prioritized enough for a market to exist around it.

Better validation wouldn't have caught all of these. But it would have caught most.