Harvard Studied Hundreds of Startups and Found Three Things Most Founders Refuse to Accept:
Professor Tom Eisenmann interviewed 470 entrepreneurs and spent two decades studying and teaching entrepreneurship before understanding the numbers behind the reasons why most startups fail:
Knowing what kills startups and being willing to accept that it’s happening to you are two completely different things. Harvard Business School professor Tom Eisenmann spent years interviewing 470 entrepreneurs and two decades teaching entrepreneurship before writing his landmark book “Why Startups Fail”, and what he found wasn’t a single cause of failure or a simple checklist of mistakes, but a set of patterns that are recurring, predictable, almost universal that show up at specific stages of a startup’s life with enough consistency to be mapped in advance. Three of those patterns are the ones that kill the most companies, and the most dangerous thing about all three of them is that they look like success when they’re happening.
Here’s what it looks like:
Stage 01 — Validation: The 42% Problem
CB Insights analysis of startup post-mortems consistently finds that 42% of startup failures come from building products that nobody actually wants. Surprising enough, it wasn’t from running out of money, bad timing, or from aggressive competition, but from spending months or years building something in isolation and then discovering at launch that the market had no appetite for it. Nearly half of all startup deaths are self-inflicted wounds that happen before a single customer ever pays for anything, and the cause in almost every case is the same: a founder who was so convinced their idea was right that they skipped the only step that could have told them otherwise.
HBS Professor Christina Wallace frames the core principle plainly: “During the early days of a startup’s life, it must iterate on all parts of its business model. The name of the game is to fail quickly and continue to ideate until you find something that works.” That sounds obvious in a lecture hall but it feels completely different when you’re six months into building something you believe in and someone suggests you stop and go talk to 20 strangers before writing another line of code. Most founders don’t stop. Most founders ship first and learn second, which is exactly backwards from the process that actually produces companies people want to pay for.
The fix isn’t complicated: talk to 20 strangers who have the problem you’re trying to solve, before you build anything, and listen to what they actually say rather than what you hope they’re saying. The founders who find product market fit quickly are almost always the ones who spent the most time in front of customers before they ever touched a keyboard, because the real metric for PMF isn’t a revenue number or a retention rate, it’s whether someone’s pupils dilate when you show them the product, whether they say “where have you been all my life,” whether they try to give you money before you’ve asked for it. That’s the signal you’re looking for, and you can only find it by getting in front of the people who have the problem.
Stage 02 — Early Traction: The Speed Trap
The Speed Trap is what Harvard’s Tom Eisenmann calls the number one killer of late-stage startups, and the cruel irony of it is that it only happens to companies that have already succeeded. You’ve found your early customers, the product is working, revenue is coming in, and then the panic sets in because now you feel the pressure to scale, and scaling feels like the responsible thing to do with the momentum you’ve built. So you hire too fast, and spend on ads before you understand what’s actually driving your growth. You add headcount before you’ve locked down one repeatable, predictable way to acquire a customer and the weight of all that premature scaling starts to slow down the very thing that made you successful in the first place.
Eisenmann documented this pattern through dozens of cases, including Fab.com: a flash-sale site that launched in 2011 with instant viral success, raised $320 million in venture capital off the back of that momentum, expanded into new product categories and international markets before its core model was truly understood, and then collapsed under the weight of the infrastructure it had built to support growth it was never actually generating. Fab sold $115 million in merchandise in 2012 and was out of business a few years later. The product hadn’t failed, the market hadn’t disappeared: the company had scaled past its own understanding of what was working and by the time the founders realized what was happening, the burn rate had made a course correction almost impossible.
As Eisenmann put it directly: “Hypergrowth can attract new rivals, result in unexpected market saturation, and reveal inadequacies in leadership. A larger workforce requires structure and processes which conflict with the entrepreneurial spirit of the startup’s early culture.” The answer isn’t to grow slowly. It’s to stay small long enough to find one repeatable way to get a customer before you try to do it at scale. The founders who navigate the Speed Trap successfully are the ones who resist the pressure to hire and spend and expand until they can draw a straight line from a specific input to a specific customer acquisition, every single time, with predictable economics. Until that line exists, every dollar you spend on growth is a dollar you’re spending before you understand what growth actually costs you.
Follow us on Instagram- @FoundedCEO
Stage 03 — Scaling: The Leadership Failure Nobody Talks About
The third pattern is the one that’s hardest to accept because it requires a founder to look in the mirror and acknowledge something deeply uncomfortable: 65% of startup failures at the scaling stage happen not because of the market, not because of competition, but because of leadership. The product works, the revenue is real, the team is talented, and then something shifts. The founder loses the obsession with the customer that built the company, the internal complexity of managing a larger organization starts consuming the attention that used to go to product and users, and the company slowly stops being the thing that earned its customers’ loyalty in the first place.
HBS Senior Lecturer Jeffrey Rayport, whose research on the scaling phase has become one of the most cited frameworks in entrepreneurship, found that even startups with glowing reviews and skyrocketing sales can fail at this stage because they cannot sustain profitability at scale , and the primary reason is almost always a leadership failure rather than a market failure. The founder who was the best possible person to find product market fit is not always the best possible person to run a 200-person organization, and the companies that make the transition successfully are almost always the ones where the founder recognized that distinction early enough to build the team around it rather than resist it.
Eisenmann identifies this as the “Help Wanted” pattern where the startup has outgrown the capabilities of its founding team and the gap between what the company needs and what leadership can provide starts to compound silently, showing up first in retention metrics, then in revenue growth, then in the kind of organizational dysfunction that makes it into the press right before everything falls apart. The companies that avoid it are the ones where the founder stays pathologically close to the customer even as the organization scales around them: not by personally doing every customer call forever, but by building systems that ensure the voice of the customer never stops being the primary input to every major decision the company makes. Don’t let the organizational complexity that comes with scale become the thing that disconnects you from the reason your customers chose you in the first place.
Why It Matters: These three patterns of building without validation, scaling before you understand what’s working, and losing the customer obsession that built the company, account for the overwhelming majority of startup failures across every industry, every stage, and every funding level. Harvard’s research makes clear that two-thirds of startups never show a positive return, and the founders inside that two-thirds aren’t there because their ideas were bad or their markets were wrong, but they’re there because they moved through these three stages without the self-awareness to recognize the pattern while it was happening to them.The founders who build the companies that last are the ones who end up on the lists we covered when A16Z released where startup dollars are actually going and when YC and A16Z shared their startup wish lists for 2026.
The Takeaway: Harvard spent years studying hundreds of startups and found the same three failure patterns showing up at the same three stages with the same predictable consequences. 42% of startups die at validation because they never talked to a customer. The Speed Trap kills companies at early traction because founders scale before they understand what’s working, and 65% of scaling failures come not from the market but from leadership losing the obsession with the customer that built the company in the first place. The long term successful founders who build companies that last are almost never the ones with the best ideas but the ones who talked to customers before they built, found one repeatable acquisition channel before they hired, and stayed obsessed with the customer long after the market validated that they were right. A prime example of this is Jeff Bezos obsessing over customer satisfaction when building Amazon, and Steve Jobs while building Apple to a $3 Trillion empire. That’s the exact pattern Harvard found, that’s the pattern the data confirms, and that’s the pattern most founders will read, nod along with, and then not follow, which is exactly why only 1% of startups ever become unicorns.