The reality of being a founder — what the data actually says

Most of what is published about being a founder is recruitment material — accurate enough on the survivors, silent on the rest. This is the picture you would see if the data were not filtered. Most fail. Mental-health conditions are more common in the population than in the comparison group. Both halves are well-replicated. With primary sources you can verify.

Most of what is published about being a founder is recruitment material. It is not dishonest in the legal sense; the people writing it generally believe what they are saying and most of them have lived a version of it. But what gets written, and what gets shared, and what gets surfaced by algorithms, is the part of the population who survived. The other part — most of the population — is not on stage. Survivor bias is the technical name. It means a reader who consults the available material to inform a decision sees a picture that looks better than the underlying data warrants, on every dimension that matters.

This piece is an attempt at the picture you would see if the data were not filtered. It is not a deterrent and it is not encouragement. It is what the published research reports about who founders are, what happens to them, and what the population-level outcomes look like. The publication's evidence-strength labels apply throughout: [STRONG] for findings replicated across multiple independent studies; [MODERATE] for credible evidence with limitations; [INTERPRETIVE] for claims that rest on judgment rather than measured data. Where the data is contested, both sides are stated.

Most fail

The exact failure rate depends on what you count and over what window, and a careful piece has to name the disagreement rather than pick a single number to sound louder.

The US Bureau of Labor Statistics, tracking all newly-formed private-sector businesses, reports that about one in five fail within the first year, around half fail within five years, and roughly two-thirds fail within ten years. [STRONG] These are not venture-backed startups specifically — they are all new businesses, including the high-survival categories (independent professional services, established trades) and the high-failure categories (restaurants, retail).

For venture-backed startups specifically, the failure rate is higher and the threshold for what counts as “not failing” is much harder. The often-quoted 90% failure rate for venture-backed startups rests on a definition where failure means anything short of a 10× return to investors over ten years. [MODERATE] By that definition, the figure is roughly accurate; by a less stringent definition (the company still exists in some form), the venture-backed failure rate is closer to 65–75%. Either way, the point that survives the definitional fight is straightforward: among venture-backed founders specifically, the modal outcome is failure, the median outcome is failure, and most of the population-level financial returns come from a small fraction of companies in any given fund. [STRONG] This is the power-law structure that the venture model is built around. The main analytical piece on venture capital on this publication treats the welfare and design implications in detail; here the point is just that the population-level distribution is what it is, and a prospective founder is selecting into it.

Three implications follow that the recruitment material does not always make obvious.

First. If most fail, then most of the people who tell themselves before starting that they will be the winner are wrong about themselves. They are not wrong to attempt the thing — the system needs the attempts — but the prediction they made about their personal outcome was incorrect. 33% of entrepreneurs in the Cooper, Woo, and Dunkelberg 1988 study estimated their personal probability of success at 100%, against base rates closer to 50% five-year survival. [STRONG] The over-confidence is a population-level fact, not a moral failing of any individual founder. It is a feature of how people who select into entrepreneurship think about themselves at the moment they make the decision.

Second. The financial outcome distribution is heavily right-skewed. The median founder, even of a venture-backed company that achieves an exit, earns less over the life of the company than they would have earned in salaried employment in the same role. [MODERATE] Hall and Woodward's 2010 paper on the burden of the non-diversifiable risk in venture entrepreneurship is the canonical citation here; their model shows that under reasonable assumptions about risk aversion, the typical founder is not financially better off for having attempted the thing. The mean is dragged up by the outliers; the median is below the salary alternative. The visible success stories in the press are drawn from the right tail of a distribution most of which sits to the left of the salary line.

Third. Failure is not evenly distributed. First-time founders fail at higher rates than repeat founders. Founders without prior industry experience fail at higher rates than founders with it. Solo founders fail at higher rates than founder teams. Founders without access to existing networks fail at higher rates than those with them. [STRONG] None of these is a moral statement; they are statistical regularities in the published data. A prospective first-time founder without industry experience and without a co-founder is selecting into the harder version of an already hard distribution.

The mental-health data

This is the part the popular discourse handles most badly, in both directions. It either romanticises mental-health conditions as founder superpowers, or it pathologises founders as a category of people the reader should pity or avoid being. The data supports neither framing cleanly. Here is what the research actually reports.

Freeman et al. 2019, “The prevalence and co-occurrence of psychiatric conditions among entrepreneurs and their families”, Small Business Economics 53(2), 323–342. A self-report survey of 242 entrepreneurs and 93 comparison participants. [STRONG] The headline findings:

• 72% of entrepreneurs in the sample reported a personal or family history of a mental-health condition. 49% reported a personal mental-health history; an additional 23% had family history without personal history. Only 24% of entrepreneurs in the sample were “asymptomatic with asymptomatic families”.
• Entrepreneurs reported significantly higher rates than the comparison group of: depression (30%), ADHD (29%), substance use disorders (12%), and bipolar disorder (11%).
• 32% of entrepreneurs reported two or more mental-health conditions; 18% reported three or more. Co-occurrence is the rule, not the exception.

The Freeman study has methodological limitations a reader should know about. The sample is non-random — recruited from MBA programs, psychology students, and entrepreneurs unaffiliated with a university. Self-reported diagnoses are not the same as clinically-confirmed diagnoses. The comparison group is small (n=93). The study is not population-representative; it is suggestive rather than definitive. [MODERATE] The pattern, however, has been replicated and refined in subsequent work, and a 2025 meta-analysis by Tran, Wiklund, Antshel, Jhawar, and Montgomery — synthesising 298 effect sizes from 47 studies — finds the relationship between ADHD symptoms and entrepreneurship to be one of the most consistently-replicated findings in the entrepreneurship-psychology literature. [STRONG]

What the meta-analysis adds is a more careful breakdown of which ADHD symptoms relate to which outcomes. The headline findings are subtler than “ADHD makes you a good founder”:

• Hyperactivity and impulsivity symptoms are positively associated with entrepreneurial attitudes and behaviours — the decision to start, the energy to pursue, the willingness to take risk. People with these traits are more likely to enter entrepreneurship.
• Hyperactivity and impulsivity are not significantly associated with post-launch outcomes. Entry rates and entry energy do not translate into measurable advantages in firm performance.
• Inattention symptoms are negatively associated with post-launch outcomes. The ability to sustain focus on the operational, financial, and managerial work of running the business once it exists matters; inattention is a real cost in that work.

The honest reading is that ADHD is correlated with entry into entrepreneurship but does not, on average, advantage the entrepreneur once they are inside. Some sub-traits help with starting; other sub-traits hinder running. The picture is mixed, not heroic. [STRONG]

On depression and burnout, the picture is also clearer than the popular narrative. The Cubbon, Darga, Wisnesky, Dennett, and Guptill 2021 scoping review of depression among entrepreneurs reports prevalence rates substantially above the general working population, with the elevated risk concentrated in stages of high financial stress, founder isolation, and proximity to failure. [STRONG] Suicide rates among entrepreneurs are higher than in the general working population, though the underlying data quality is mixed and methodologies vary. [MODERATE]

The Freeman, Cubbon, and meta-analysis findings together support a cautious summary statement: founders, as a population, have higher rates of mental-health conditions than the general working public. They also have higher rates of mental-health distress while in the founding role, particularly during periods of financial stress. The two facts are related but not identical: pre-existing conditions do not fully explain in-role distress, and in-role distress is not fully explained by pre-existing conditions. [STRONG]

Why the population looks this way

The data above admits two main causal stories, and reasonable researchers disagree about how much weight to give each. The honest version names both.

The selection story. People with certain traits — high tolerance for risk, novelty seeking, hyperactive cognitive style, dopaminergic temperament, willingness to act under uncertainty — are more likely to choose to start a company. Some of those traits overlap with the diagnostic criteria for ADHD, hypomania, and certain personality features. The population of entrepreneurs is therefore enriched for these conditions, not because entrepreneurship causes them, but because the selection filter favours the traits that correlate with them. [INTERPRETIVE]

The exposure story. Founding a company exposes a person to a sustained period of financial uncertainty, social isolation, identity fusion with the company, irregular sleep, irregular eating, and chronic stress. Each of these is, independently, a known precipitating factor for depression, anxiety, substance use, and burnout. People who would not have developed these conditions in salaried employment can develop them in the founding role. The population of active entrepreneurs is therefore further enriched, beyond the selection effect, by the exposure effect. [INTERPRETIVE]

Both stories are likely partly true. The published research does not yet definitively partition the variance between them. [INTERPRETIVE] What the research does support is that the prevalence numbers are real, that they have been replicated, and that a prospective founder considering whether to enter is selecting into a population whose mental-health profile is statistically distinct from the salaried alternative.

What this does and does not mean

It does not mean that an individual reader with ADHD, depression, or any other condition should not consider founding a company. The population-level statistics describe the population. They do not tell any specific person whether they will be the rare founder for whom the recruitment messaging is correct, or one of the many for whom it is not. The frames evaluate the system. Evaluating yourself is a different exercise this article cannot do for you.

It does not mean that the mental-health correlations are what defines a founder. Most founders are not, primarily, “a person with ADHD” or “a person with depression”; they are people doing a difficult job under conditions known to be stressful, some of whom have these conditions and many of whom do not. Reducing a person to a diagnosis is a category error the research does not endorse and this publication does not endorse.

It does not mean the entrepreneurial route is irrational. The system needs the attempts; the aggregate output the venture model produces depends on a population of individuals willing to try. The main VC piece on this publication argues, on balance, that the venture system produces net-positive outcomes at civilisational scale. None of that argument is undermined by the population-level fact that most individual founders do not benefit financially from the attempt.

What the data does mean is more modest and more practical:

• If you are weighing whether to start a company, the base rates apply to the population you are joining. The base rates do not tell you your personal outcome, but they constrain the prior you should hold.
• If you are managing your expectations about your own mental health during the founding period, the published data suggests you should plan for the possibility of distress, isolation, and impaired sleep, and treat these as predictable consequences of the role rather than as personal failings.
• If you have a pre-existing mental-health condition and are considering founding, it is worth knowing that the population is enriched for people with similar conditions, that the in-role exposure tends to amplify rather than dampen these conditions, and that having a clinical relationship in place before the stressful period — not after distress emerges — is what the published guidance from Freeman, the World Economic Forum's 2019 entrepreneur mental-health initiative, and others tends to recommend. [INTERPRETIVE]
• If you are an investor, partner, or family member of a founder, the population data has implications for what kind of support is useful. Programmatic mental-health support is not a luxury or a perk; it is a response to a known feature of the population.

The three things the recruitment material omits

Most of the existing literature for prospective founders includes the success stories, the playbooks, the frameworks, and the post-hoc reasoning of the people who won. The three things it tends not to include, surfaced together for once:

One. Most of the people who decided to do this thing did not get the outcome they predicted for themselves at the start. [STRONG] This is true financially, true on the dimension of company survival, and true on the dimension of personal life satisfaction relative to counterfactual employment. Whether it is true for you depends on facts about you that no published study can know in advance.

Two. The population that selects into founding has higher rates of mental-health conditions than the population that does not, before they start; the population's rates of distress also rise during the founding period as the exposure accumulates. [STRONG] Both halves of this are true; neither half explains the other entirely. ADHD, depression, anxiety, substance use, and bipolar conditions are statistically more common in the population you are considering joining than in the population you are leaving.

Three. The system that surrounds founders — the accelerators, the investors, the press, the conferences, the LinkedIn ecosystem — is structurally unable to tell you the first two things plainly because its function is to keep the supply of new attempts flowing. The aggregate output the system produces depends on having a population of individuals who systematically over-estimate their personal success probability and under-estimate the personal cost of failing. The recruitment messaging produces those over-estimates, on purpose, regardless of whether any specific reader will actually be the winner. [INTERPRETIVE] Most of the messaging is not lying; it is selecting which truths to surface. The ones that suppress the supply of attempts are the ones that are not surfaced.

This article is an attempt to surface the three together, in the publication's own voice, with citations a reader can verify. It is not a deterrent and it is not an encouragement. It is the picture the data supports, presented without the rhetorical wrappers the system around founders usually applies. A reader making this decision should weigh the data against facts about themselves — the relevant industry experience, the financial runway, the support structure, the alternative they are forgoing, the prior mental-health profile, the support of the people closest to them, the answer to the honest question why are you doing this — and decide. The decision is theirs. The data is the data.

Primary sources

For the empirical claims in this piece, in citation order:

• Cooper, A. C., Woo, C. Y., & Dunkelberg, W. C. (1988). Entrepreneurs' perceived chances for success. Journal of Business Venturing, 3(2), 97–108.
• Hall, R. E., & Woodward, S. E. (2010). The burden of the non-diversifiable risk of entrepreneurship. American Economic Review, 100(3), 1163–1194.
• Freeman, M. A., Staudenmaier, P. J., Zisser, M. R., & Andresen, L. A. (2019). The prevalence and co-occurrence of psychiatric conditions among entrepreneurs and their families. Small Business Economics, 53(2), 323–342.
• Cubbon, L., Darga, K., Wisnesky, U. D., Dennett, L., & Guptill, C. (2021). Depression among entrepreneurs: a scoping review. Small Business Economics, 57(2), 781–805.
• Hatak, I., Chang, M., Harms, R., & Wiklund, J. (2021). ADHD symptoms, entrepreneurial passion, and entrepreneurial performance. Small Business Economics, 57(4), 1693–1713.
• Tran, M. H., Wiklund, J., Antshel, K., Jhawar, N., & Montgomery, C. (2025). Entrepreneurship and ADHD: A meta-analytical assessment of the state-of-the-art and suggestions for the future. Entrepreneurship Theory and Practice, in press.
• U.S. Bureau of Labor Statistics. Business Employment Dynamics, multiple years. Survival rates of new establishments.
• Wiklund, J., Yu, W., Tucker, R., & Marino, L. D. (2017). ADHD, impulsivity and entrepreneurship. Journal of Business Venturing, 32(6), 627–656.
• Yu, W., Wiklund, J., & Pérez-Luño, A. (2021). ADHD symptoms, entrepreneurial orientation (EO), and firm performance. Entrepreneurship Theory and Practice, 45(1), 92–117.

Reconfirm any specific figure with the primary source before relying on it. The publication's evidence-strength labels above flag where the evidence is strong and where it is more contested; readers making consequential decisions should look at the contested findings particularly carefully.