Looks Like a Unicorn

👋  Hi, welcome to Signal Lab where the team at Innate Language Processing digs deep into language & behavior, the limits of technologies, and the connection between what people say and do. For venture investors and startups.

Most funds begin by assessing Product, Market, and Team when evaluating companies for investment.

At early stage, market quantification is easily done but can be inapplicable (unicorns are often category-defining), and product is harder to quantify (e.g., traction, growth, differentiation come later).

So, many early stage investors “bet on the jockey”, an approach that’s not without its own problems.

VCs may eliminate “jockeys” who do not fit a particular profile of success. This profile is usually drawn from their personal experience of unicorn founders, which — even among blue chip firms — will be extremely limited. The pool of unicorns (~1500 in the past 25 years) is already small; drawing generalizations from <1% of that population is spurious, at best.

Today’s note focuses on a few of the founder demographics that VCs believe are correlated with success.

"If you look at the most successful technology companies, they were started by people who were very young. They didn't know what was impossible." — Peter Thiel (Founders Fund)

Despite the stereotype, only 4% of unicorn founders are college dropouts. Some data suggests that AI unicorn founders (in particular) do average under 30. But unicorn founder average age only dropped under 30 for the first time in 2024, and only for AI companies (nb. sample size of AI companies founded and reaching a 1B+ valuation within 24 months is — as you’d imagine — very very small, making this result pretty meaningless).

Stanford professor Ilya Strebulaev’s Unicorn Report instead finds the average age of a unicorn founder to be 35 years.

More data on age here. And, a MIT Sloan study looking at the ages of founders of the fastest growing startups and startups with “upper-tail” exits (IPO or huge acquisition).

“Most VCs around me believe that pedigree is most important in deciding what founding teams to back.” — Elizabeth Yin (Hustle Fund)

Pedigree is a particularly long-standing false correlation with success in venture capital. One example of faith in pedigree gone wrong (there are many) is BitClout, an a16z-backed startup whose founder, Nader Al-Naji, was later charged with fraud by the Justice Department as his company failed. When Fortune interviewed several VCs after the indictment, they said they had invested because they were been impressed with Al-Naji’s Ivy League pedigree.

One study found that 85% of unicorn founders did not attend the top 5 universities. Half had no postgraduate degree.

"We look for founders who have a 'secret'—something they know about a market that no one else does because they've lived in it" — Roelof Botha (Sequoia).

Ben Horowitz famously coined the term “earned secret” to describe the “unfair advantage” that a successful founder has, gained over time and close contact with the problem their startup will solve.

But aside from exceptions of healthcare and biotech startups, the data does not confirm this idea. According to some studies, 60% of unicorn founders had no prior sector experience.

Many unicorn founders had never worked for anyone before. Nearly 70% of unicorn founders had less than ten years’ experience in any work at all.

Many GPs consider disagreement in the Investment Committee to be a bad sign for a startup, with some firms even requiring partners’ unanimous agreement in funding decisions. However, a 2024 paper from MIT Sloan looked at over 2500 startups in competitions and found that the more disagreement among decision-makers about the startup, the more likely that startup to ultimately succeed.

Even after the author controlled for average score, competition, vintage, and patents, the relationship remained strong.

Disagreement was more common in more distinct startups. In fact, and importantly, for less distinct startups, that predictive power from disagreement basically disappeared, suggesting disagreement is only meaningful in startups with bold, new ideas.

👉 External validation?

Many of these indicators are sourced in circular reasoning rather than causality: founders with the “right” signals get funded, models learn these signals predict funding, investors double down on these signals as predictors of success.

For instance, many more unicorns are minted in the Valley than any other US ecosystem, combined. Is this because Valley founders have more talent, or more access? Likewise, a 2024 analysis found 40% of unicorn founders were serial entrepreneurs, and Ali Tamaseb (DCVC) gathered 30,000 data points on startups to confirm that repeat founders have greater success, eventually. Are repeat founders better able to build a successful company, or better able to raise funding?

Because these patterns were drawn from the structure of past capital allocation, it’s hard to disentangle structural advantage from true performance indicators.

…And when the data is more subjective in nature than this (e.g., unicorn founder personality, or leadership traits correlated with startup success), it becomes even more complicated.

And that’s what we’ll tackle next.

See you in two weeks.