A mean of 45.0 is reported for a narrow top-growth group (Azoulay et al., 2020). That is a group statistic, not a personal forecast. Read in that spirit, it can correct an impression. Read carelessly, it becomes a rule about your life.
What does the research actually report?
The study covers United States employer firms from 2007 to 2014, drawn from administrative records on about 2.7 million founders, and it excludes sole proprietors and businesses without employees (Azoulay et al., 2020). Age at startup was measured at the year of a firm's first hire. Mean founder age was 41.9, and 45.0 for the top 0.1 percent by five-year employment growth. Other outcomes, including sales and exits, were also examined.
What does the study count as a startup and as success?
"Success" is not one thing, so a founder-age figure only makes sense once you fix the outcome it was drawn on. A company can be counted a success by job growth, by profit, by the income its owner takes out, or by the flexibility it buys. Those are different yardsticks, and they pull in different directions. A firm can add staff quickly and still pay its owner modestly. Another can stay small, pay its owner well, and grant real freedom over the calendar.
This is why selecting the outcome first is necessary rather than optional. Until you name it, no average can be read correctly, because it is answering a question you have not yet asked. Two further points follow from the full measurement in the paragraph above. First, an average tied to an upper tail is not usable as a personal yardstick. Second, a wider group average is a broader and more ordinary reference point, and the two answer different questions.
Why is a mean not the typical or best age?
Because a mean is an arithmetic total divided by a count. It is one measure of central tendency, and it is not meaningless, but on its own it does not tell you the median or how spread out the group is. Consider two invented sets of four ages, offered purely to show the arithmetic. They are not founders.
| Set | Ages | Sum | Mean | Median |
|---|---|---|---|---|
| A | 30, 40, 50, 60 | 180 | 45 | 45 |
| B | 20, 20, 60, 80 | 180 | 45 | 40 |
In Set A, the sum is 180, divided by four gives a mean of 45, and the two middle values are 40 and 50, so the median is also 45. In Set B, the same sum and the same mean, but the two middle values are 20 and 60, so the median is 40, and the spread runs from 20 to 80.
Two things follow. First, nobody in either set is 45. The mean is a calculation, not a member of the group. Second, a mean can stay in the same place while the distribution underneath it changes completely. A headline that reports only the mean has told you nothing about spread or shape. This is why a mean is worth reading carefully, not dismissing.
Does industry experience explain the whole result?
Prior industry experience predicted stronger growth. That association is not proof that the full age pattern is explained. The paper also examines age-conditioned rates. A mean alone is not your personal odds (Azoulay et al., 2020).
A timing distinction helps here, offered with invented figures. Imagine a fictional founder who had the first idea at 45, made the first hire at 45, and reached public recognition at 58. Each of those ages answers a different question.
How should you check a founder-age headline?
Four questions, in order. These come from general evidence-reading practice, not from the study itself.
- Who was measured, and when? Identify the population, the country, the years, and the point at which age was captured.
- What outcome was counted? Employment growth, profit, owner income, and flexibility are different results. A claim about one does not transfer to another.
- What comparison is being made? A figure for an upper growth tail and a figure for a full cohort answer different questions.
- What follows? Ask whether the evidence supports a description of a group, a prediction about a person, or a causal claim. Group averages support the first.
| Claim you read | What to check before repeating it |
|---|---|
| "Successful founders are 45." | Which outcome, which sample, and which part of the distribution? |
| "Older founders always win." | Is this a group tendency described as a rule for individuals? |
| "Experience is the real cause." | Does the study claim cause, or only a predictor? |
| "The average is meaningless." | Or is it simply one measure among several? |
A worked example. Suppose you read a headline stating that the average successful founder started at 45. Applying the questions: the figure describes a particular sample in particular years, so it is not a statement about your market. The conclusion that follows is descriptive, not predictive. So the headline alone cannot tell you whether your venture will work. Evidence about your own situation, such as customer conversations, a competitor review, and a cost model, addresses a different question. Both kinds of evidence have limits. The skill is noticing which kind answers which question.
What can this number help you decide?
It can prompt a useful check on a stereotype. If you assumed fast-growing companies are mostly founded by people in their twenties, a figure in the forties may make you revisit that impression. But a mean alone cannot establish the most common or youngest age, or the age of the majority. That needs the distribution, not one summary number.
It cannot tell you your own odds, and it cannot tell you whether to start now or wait. An average conditional on an outcome is not the probability of that outcome for any individual. Questions worth asking before you act include: what would have to be true in your market, what evidence would change your mind, and what small test could produce that evidence cheaply.
The practical move is not to wait until a particular age, and not to launch on the strength of one statistic. It is to read the full measurement, name the questions a headline left out, and then gather evidence that speaks to your own situation. If that process is the part you want help with, you can work with me, and the broader framework lives under success psychology.
Key takeaways
- A mean is a total divided by a count: one summary of a group, not a description of any individual in it.
- A mean alone does not determine the median or the spread. Two very different groups can share one mean.
- The median of an even set is the mean of the two middle values.
- An average conditional on an outcome is not the probability of that outcome for you.
- When you read a founder-age claim, ask who was measured, what outcome was counted, and whether the conclusion is descriptive or predictive.
Frequently asked questions
References
Azoulay, P., Jones, B. F., Kim, J. D., & Miranda, J. (2020). Age and high-growth entrepreneurship. American Economic Review: Insights, 2(1), 65-82.
This article is for informational and educational purposes only and does not constitute financial, legal, tax, medical, or professional advice. Individual results vary.