Survivorship bias occurs when we draw conclusions from a sample that systematically excludes failures — because failures are no longer visible, no longer counted, or no longer in a position to report their outcomes. The result is a distorted picture of reality, in which the world seems more conducive to success than it actually is.

One of the most striking historical illustrations comes from World War II. The statistician Abraham Wald was asked by the US military to recommend where to add armour to bomber aircraft. Analysts had studied the bullet hole patterns on planes that returned from missions and proposed reinforcing the areas with the most damage. Wald pointed out the error: the planes being studied were the ones that had survived. The holes showed where a plane could be hit and still return. The places with no holes on the survivors were precisely the places that, when hit on other planes, had caused them not to return at all. The right conclusion was the opposite of the intuitive one.

Why Survivorship Bias Is Invisible

The bias is unusually hard to correct because the missing data is, by definition, absent. You cannot easily survey the failed startups, the rejected manuscripts, the investors who went bankrupt, or the soldiers who did not come home. The evidence available to us is structurally skewed toward cases that survived long enough to be recorded.

This makes survivorship bias particularly insidious in domains where we try to learn from examples. Business schools study successful companies. Investors study successful fund managers. Self-help literature is written by people who succeeded and attributes that success to specific habits or mindsets. None of these sources can tell you about the equally motivated, equally disciplined people who had the same habits and failed — because those people are not writing bestsellers.

Where Survivorship Bias Appears

• Entrepreneurship: "Most successful founders dropped out of university" is a classic survivorship observation. For every dropout who founded a billion-dollar company, thousands dropped out and failed. The dropouts who succeed are visible; the ones who did not are not writing keynote speeches.
• Investment advice: A fund manager with a ten-year track record of outperformance is remarkable — but only if you account for all the funds that began the decade and were quietly liquidated after underperforming. The surviving funds look far better as a group than all the funds that started.
• Historical narratives: Histories tend to be written about states, institutions, and people that persisted. The civilisations, companies, and individuals that vanished leave fewer records. Our view of the past is filtered by what survived to be studied.
• Product design: Customer feedback is collected from people who are still using a product. People who abandoned it for a competitor are not in the feedback channel. The result is a systematically incomplete picture of what is going wrong.

How to Correct for Survivorship Bias

• Ask what the denominator is. "Ten out of ten successful founders I know had X habit" tells you nothing without knowing how many founders had X habit and failed. Always ask about the total population, not just the visible successes.
• Seek out failure data deliberately. Look for post-mortems, failure analyses, and studies of unsuccessful cases. They are harder to find but far more instructive for avoiding the same outcomes.
• Be suspicious of inspirational case studies. A single vivid success story is not evidence of a general principle. It is evidence that this specific person, in this specific context, with these specific conditions, succeeded. Context and luck are hard to replicate.
• Consider the graveyard. When you see a pattern in successful examples, ask what the equivalent graveyard of failed examples looks like. If you cannot see the graveyard, be cautious about the lesson.

Practise Recognising It

Survivorship bias scenarios are among the most enjoyable to work through because the "invisible" missing data feels so obvious once pointed out. Encounter them in the Dojo and explore how they connect to Confirmation Bias and the Texas Sharpshooter fallacy in the Library.