MediumBusiness Success: Luck, Not Merit. | by Fayner Brack | MediumSitemapOpen in appSign up<br>Sign in

Medium Logo

Get app<br>Write

Search

Sign up<br>Sign in

Business Success: Luck, Not Merit.

Each new founder pays again to learn what the last founder already did. Some of them hit the jackpot.

Fayner Brack

5 min read·<br>May 28, 2026

Listen

Share

Top-down view of five identical cubicles in a row, each with a person at a laptop showing the same ERROR message, surrounded by crumpled paper.

Want to come back later? Save this to readplace.com.<br>Business Success: Luck, Not Merit. | Reader View<br>1 day ago

readplace.com

Most people think startup failure is a problem of execution. Of course, that's part of it. However, more often than not it’s also a problem of information.<br>A new founder with scarce financial capacity need to find the right product-market fit in order to achieve compunding distribution.<br>However, to get the right product-market fit, they need to test their hypothesis against real users to learn what works and what doesn't.<br>To get real users, they need to get product-market fit or nobody will be interested to try anything.<br>It's a chicken and egg problem. The loop closes before the experiment can even start.<br>Press enter or click to view image in full size

Founder’s Loop: Product-Market Fit needs users, users need tested hypotheses, tested hypotheses need Product-Market Fit.The founders who succeed are partly the ones who happened to know the answer (someone on the team or themselves had built something similar in another company), or they guessed, and guessed, and eventually the guess held up.<br>Those who have more money, either by coming from a wealthy family, by building it incrementally over decades or selling an asset like a house, have the ability to try more guesses. The more guesses they try increases their chances of winning. Those who are financially constrained have limited guesses and therefore less chances of winning.<br>Building a business is a game of chance. We overestimate merit and underestimate luck.

About 42% of failed startups in the CB Insights post-mortem analysis built something nobody wanted. The founders had no way to know what their customers would say without shipping. By the time they shipped, the money was running out.<br>The information that would have saved them already exists. It’s sitting inside the companies that already failed, and inside the ones that succeeded and decided not to publish. A failed venture is the only place the most valuable data lives: they spent the money, ran the experiment, and watched it fail.<br>The findings can be specific: pricing pages that destroyed conversion, onboarding flows that dropped half the users at step three, acquisition channels that looked great for a month and went flat, features people swore they wanted in interviews and ignored after signup. Those findings sit in someone’s head, or in a Slack archive that gets deleted six months after the company shuts down.<br>The knowledge that would help the next founder is the same knowledge the last founder paid to produce.

Universities used to be where this kind of knowledge accumulated. A researcher would publish, peers would replicate, the field would advance. The trouble is that universities now operate closer to being a businesses, far away from being recongised as a pillar of wisdom. They chase grants, file patents, and protect findings the same way a company would. Even setting that aside, a research team without a product on the market can’t generate the kind of data that matters for building one.<br>Customer behaviour shows up in production environments, not in clean experimental designs.<br>This is the biggest trap of the 21st century.<br>The institutions designed to share knowledge (universities) can’t produce the most useful kind. The institutions that produce the most useful knowledge (businesses) have no reason to share it.<br>Get Fayner Brack’s stories in your inbox

Join Medium for free to get updates from this writer.

Subscribe

Subscribe

Remember me for faster sign in

Not everyone retains knowledge, of course. In 2008 Netflix had a three-day database outage that prevented DVDs from shipping to customers. They started moving to AWS soon after. The migration took years and they hit reliability problems other companies hadn’t faced at that scale. They built a tool that deliberately killed instances in production. The point was to find weak spots in advance of real failures. They called it Chaos Monkey. They published the concept on their blog in 2011. They open-sourced the code in 2014.<br>The practice spread and other companies adopted chaos engineering without reinventing it.<br>That work was operational. Netflix wasn’t selling chaos engineering, and no part of it was patentable in any useful way. Sharing it cost them nothing competitive and saved other companies years of incident-driven learning.<br>Open source is one version of this pattern: a company solves a problem internally, decides what...