Every founder program has a picture of its ideal applicant, and in most of them the picture is young, technical, fluent on stage and armed with a deck. The research on who actually builds durable, high-growth companies describes someone else: older than the stereotype, deeply experienced in the industry they are entering, part of a team rather than a lone genius, and distinguished less by the polish of the pitch than by how quickly they learn. This closing article of the series brings that evidence together — from the economics of founder age to the psychology of training — and translates it into the two decisions every program must get right: whom to select, and what to teach them.
The idea in brief. Founder readiness is measurable, and the things that predict it are not the things most programs select for. Age and industry experience predict high-growth founding more strongly than youth; prior founding success predicts later success; investors rank the team above the idea, and the evidence says they are right to at the point of selection. Judges’ scores of pitches predict outcomes weakly; observed behaviour — customer conversations completed, tasks done in a trial period — predicts better. Training changes outcomes when it changes behaviour: personal-initiative and error-management curricula outperform content-based business training. Programs should therefore select on evidence of readiness rather than promise, treat the intake as a work sample, and teach initiative as deliberately as they teach finance.
The myth of the twenty-something founder
The most consequential single finding in recent founder research concerns age. Pierre Azoulay, Benjamin Jones, J. Daniel Kim and Javier Miranda, using administrative data covering the full population of new firms in the United States, reported in American Economic Review: Insights in 2020 that the average age of founders of the fastest-growing new ventures — the top 0.1 per cent by growth — was about forty-five. Founders in their twenties had the lowest likelihood of building a top-growth firm; the likelihood rose steadily with age into the fifties, and a fifty-year-old founder was roughly 1.8 times as likely as a thirty-year-old to found a firm in the top growth tier. Prior experience in the specific industry was among the strongest predictors of success.
The finding runs against the entire visual culture of entrepreneurship programs, and it matters for governments in two ways. First, programs designed around students and recent graduates — which is most of them, because campuses are where the applicants are — are drawing from the population with the lowest yield of high-growth founders. Campus programs remain valuable for the reasons article 34 sets out, but a national founder pipeline that stops at graduation is missing its most productive segment. Second, the mid-career professional — the engineer with fifteen years in a sector, the manager who has run a business unit, the scientist who has spent a decade on a problem — is the highest-yield applicant a program can recruit, and almost no program is designed for them: they have jobs, families and no tolerance for a twelve-week residential bootcamp. Programs that want the vital few should build tracks that mid-career founders can actually join.
The Kauffman Foundation’s survey-based portrait of American founders, published under the title The Anatomy of an Entrepreneur and available through kauffman.org, reached compatible conclusions a decade earlier: the typical founder of a successful company was around forty at founding, had significant industry experience, and was motivated more by building something than by the absence of a job.
Jockey or horse: what investors bet on, and whether they are right
Venture investors are the most experienced selectors of founders in the world, and their behaviour is instructive even where it is not correct. Paul Gompers, Will Gornall, Steven Kaplan and Ilya Strebulaev’s survey of nearly nine hundred venture capitalists, published in the Journal of Financial Economics in 2020, found that the management team was the factor most often named as most important in deal selection — by close to half of respondents — well ahead of the product, the business model or the market. Investors, in the industry’s phrase, bet on the jockey.
Whether they are right is a separate question, and the best study of it gives a subtle answer. Steven Kaplan, Berk Sensoy and Per Strömberg’s 2009 paper in the Journal of Finance, following venture-backed firms from business plan to public listing, found that the firms’ core business lines were remarkably stable over time while their management teams changed substantially — which suggests that, over the life of a company, the horse matters at least as much as the jockey. The reconciliation is about timing. At the point of selection, the team is what can be observed; the idea will change, often beyond recognition, and the team’s ability to change it is precisely what a selector is trying to judge. Later, the business the team has found becomes the asset, and the team can be — and often is — replaced. For a program, the lesson is to select on the team’s capacity to learn and to execute, not on the idea’s apparent quality, and to expect the idea to be different by the end.
Experience, talent and persistence
Three further strands of research bear on selection. Paul Gompers, Anna Kovner, Josh Lerner and David Scharfstein’s 2010 study in the Journal of Financial Economics found that performance persists: founders who had previously taken a company to a successful outcome were markedly more likely to succeed again — a success rate of roughly thirty per cent, against about a fifth for first-time founders and little better for those whose previous venture had failed. Experience helps most when it was successful, which is a sobering finding for programs that assume failure alone teaches.
Charles Eesley and Edward Roberts, in a 2012 study in the Strategic Entrepreneurship Journal drawing on decades of MIT alumni data, found that both innate ability and prior founding experience predict venture performance, and that their relative importance shifts across a founder’s successive ventures — experience counts for more where a founder has less of it to draw on, and ability shows through as ventures accumulate. And William Kerr, Ramana Nanda and Matthew Rhodes-Kropf’s 2014 essay in the Journal of Economic Perspectives reframed entrepreneurship as experimentation: the value of a venture lies substantially in the information it produces about whether the idea works, which means the founders best suited to it are those who run cheap experiments quickly and read the results honestly. A program, in that framing, is a structured set of experiments, and readiness is the capacity to run them.
Selection science, borrowed from hiring
Article 22 in this series set out the century of evidence on what predicts job performance, drawing on Frank Schmidt and John Hunter’s 1998 synthesis in Psychological Bulletin: work samples and general cognitive ability predict best, structured interviews match them, unstructured interviews and years of experience predict poorly. Founder selection is a hiring decision under a different name, and the same hierarchy applies. The pitch competition is an unstructured interview with an audience. The pre-program task — produce evidence from ten customer conversations in two weeks; ship a prototype to five users — is a work sample. The evidence, and the experience of well-run programs, says the second predicts what the first only performs.
David McKenzie’s evaluation of Nigeria’s YouWiN! competition, discussed in article 31, supplied the field experiment: judges’ scores of business plans were only weakly related to what the businesses subsequently achieved, and random selection among applicants who had cleared a quality threshold performed about as well as expert ranking. The result is not that selection is pointless; it is that fine ranking by pitch is. Programs should clear a bar with structured criteria and observed behaviour, then allocate among qualifiers with humility — and, where they can, transparently, creating the comparison group their evaluation will later need.
A readiness rubric that follows the evidence looks like this:
| Dimension | What the evidence says | How to observe it | Weight |
|---|---|---|---|
| Commitment | Full-time founders convert; part-time ventures rarely do | Time allocated; resignation or leave plan; what has been given up | High |
| Domain experience | Industry experience strongly predicts high-growth founding | Years in the sector; roles held; problems solved from the inside | High |
| Customer evidence | Behaviour predicts better than plans | Conversations completed; pilots; letters of intent; pre-program task | High |
| Team completeness | Investors rank team first; complementary skills matter | Founders’ skill coverage; a named second founder or first hire | High |
| Learning speed | Experimentation capacity is the core founder trait | Changes made between application and interview; response to feedback in the trial task | High |
| Personal initiative | Trainable, and predictive of firm outcomes | Self-starting behaviour in the trial period; obstacles overcome without being asked | Medium |
| Prior founding | Success persists; failure alone teaches less than assumed | Outcomes of previous ventures, honestly described | Medium |
| Cognitive ability | Predicts performance across roles; modest incremental value here | Validated assessment where used; problem-solving in the trial task | Medium |
| Pitch quality | Weakly predictive | Note it; do not weight it | Low |
The rubric’s most important row is the last. Programs that weight pitch quality select for the applicants best at applying, who are not reliably the applicants best at building. Recording pitch quality without weighting it also produces, over cohorts, the program’s own evidence about whether it predicts anything.
Training that changes behaviour
Selection determines who enters; training determines what happens to them, and the evidence on training is unusually clear about what works.
The Togo experiment reported by Francisco Campos and colleagues in Science in 2017 — a randomised comparison of traditional business training with a psychology-based personal-initiative curriculum — found the behavioural program raised firm profits by roughly thirty per cent two years on, while the traditional curriculum’s smaller effect was not statistically distinguishable from zero. The personal-initiative training taught founders to be self-starting, future-oriented and persistent: to identify opportunities without being told, to plan for obstacles and to keep going through them. Dean Karlan and Martin Valdivia’s earlier experiment with microfinance clients in Peru, published in the Review of Economics and Statistics, had found what many studies since have confirmed — that content-based business training improves business practices and knowledge while leaving profits largely unchanged. Knowing is not doing; training that targets doing works.
The implications for curriculum design are direct, and they echo across this series:
- Organise the program around actions — customer interviews, prototypes, pricing tests, first sales — with content delivered in service of the next action.
- Teach personal initiative explicitly: goal-setting, obstacle planning, error management, persistence, as a curriculum with exercises, not as a motivational session.
- Use the cohort as an accountability engine: weekly commitments made in front of peers and reviewed the following week.
- Design for learning speed: short experiment cycles, honest read-outs, and explicit permission — expectation, even — to change the idea.
- Treat quitting as a legitimate outcome: as article 31 discussed, accelerated founders who learn quickly that an idea will not work and stop are the program working, not failing. Curriculum should make that decision respectable and fast.
Designing for the founders the evidence favours
If the highest-yield founders are mid-career, experienced and time-constrained, and if readiness is best observed in behaviour, then the standard program format — a residential cohort of recent graduates selected by pitch — is designed for the wrong people in the wrong way. Programs that take the evidence seriously make four changes.
- A mid-career track. Evening and weekend delivery, remote sprints between in-person intensives, and a curriculum that assumes domain expertise and teaches venture-building rather than the reverse. Recruitment through employers, professional bodies and the alumni of campus and skills programs (articles 34 and 38), not only through open calls.
- Intake as a work sample. A two-to-four-week trial period between selection and admission in which applicants complete real tasks — customer conversations, a prototype test, a co-founder conversation — and are admitted on what they did.
- Team brokering. Active matching of technical founders with commercial ones and domain experts with builders, because the complete team predicts and the lone founder rarely completes.
- Cohort tracking that tests the rubric. Every readiness dimension scored at intake and correlated with outcomes at 12, 24 and 36 months, so that the program learns, cohort by cohort, which of its own criteria predict — the closed loop that turns a program into an evidence base (article 32).
HexGn’s Startup Ready program, described in article 31, was rebuilt around these principles: a readiness rubric at intake with a trial task, a mid-career route alongside the graduate one, personal-initiative modules in the curriculum, and outcome tracking that scores the rubric against results. The design is not proprietary; it is what the evidence recommends, and any program can adopt it.
Adapting readiness to the corridor
The evidence reviewed here is mostly American and European, and the series is explicit that its transfer to India and the Gulf is an assumption to be tested rather than a fact to be assumed. Two adaptations are already clear from program experience on both sides. In India, the mid-career founder the evidence favours is abundant — a large cohort of engineers and managers with fifteen or twenty years in technology, pharmaceuticals, manufacturing and financial services, many of them inside the capability centres this series has described — but is rarely reached by open calls, which recruit the young and the unemployed; recruitment through employers, alumni networks and professional bodies is the design change that finds them. In the Gulf, the readiness dimension that needs the most deliberate support is commitment: leaving a secure public-sector role to found a company is a larger step than the rubric’s “full-time” row implies, and programs that succeed pair the intake with the wage-support and enterprise instruments that national labour funds provide, so that commitment is made affordable rather than merely required. In both settings, the trial task translates without change — customer conversations and prototypes look the same in Pune and Manama — and it remains the single most transferable selection tool the evidence offers.
A composite case: the program that selected for polish
A composite from several engagements; details altered.
A national founder program had run for four years with a selection process built on a written application, a pitch to a panel of prominent judges and a public final. Its alumni were articulate, its finals were well attended, and its 24-month outcomes — when finally measured — were poor: a small share of ventures operating, fewer with revenue, and a pattern the data made obvious once it was looked for. The ventures that had done best were disproportionately founded by older applicants with industry experience who had scored in the middle of the pitch rankings; the ventures that had done worst were disproportionately the finalists.
The redesign retired the final as a selection device and made it a celebration. Intake moved to a structured rubric, with domain experience, commitment and customer evidence weighted, and a three-week trial task between shortlisting and admission. A mid-career track was added with evening delivery, recruited through professional associations and two large employers whose alumni had founded the program’s best companies. The curriculum acquired a personal-initiative strand and weekly cohort accountability. Two cohorts later, the program’s 24-month operating rate had risen substantially, the mid-career track accounted for a disproportionate share of ventures with revenue, and the program’s intake scores had begun to predict its outcomes — which was, its director observed, the first time the program had known what it was looking for.
What could go wrong
- Selecting for the stage. Pitch competitions as selection. Antidote: rubric plus trial task; pitch recorded, not weighted.
- Youth as a proxy for potential. Antidote: mid-career tracks and recruitment through employers.
- Content-only training. Antidote: action-based curriculum with an explicit initiative strand.
- Lone-founder intake. Antidote: team brokering as a program function.
- Rubric never tested. Antidote: cohort tracking that correlates intake scores with outcomes.
- Quitting punished. Founders who stop are counted as failures and hidden. Antidote: fast, respectable exits reported as outcomes.
- Fine ranking overconfidence. Antidote: threshold plus transparent allocation among qualifiers.
Questions program directors actually ask
“Should we prefer experienced founders?” Weight domain experience heavily; it is among the strongest predictors. Do not require prior founding — most successful founders are first-timers — but score its outcomes honestly where it exists.
“What do we do with brilliant students?” Give them the campus programs article 34 describes, connect them to experienced co-founders and mid-career mentors, and expect their best ventures to come years after graduation.
“Can we assess founders with tests?” Partly. Validated cognitive and personality assessments add modest signal; the trial task adds more. Use the assessment science (article 22) as one input, never as the decision.
“Is training worth it if selection matters so much?” Yes, if it changes behaviour. Personal-initiative training has the strongest evidence of any curriculum component; content-only training has the weakest.
“How do we know our selection works?” Score every applicant on the rubric, admit transparently among qualifiers, track everyone — including those turned away — and correlate. In two cohorts you will know which of your criteria predict.
Methodology & data notes
Findings from the academic literature are summarised approximately; readers should consult the papers for effect sizes, samples and boundary conditions. The founder-age findings are drawn from United States administrative data and their transfer to India and the Gulf is an assumption the series makes explicit; the direction — that experience predicts — is consistent across the studies cited. The venture-investor survey figures are approximate shares of respondents naming a factor most important. The readiness rubric is a design proposal grounded in the evidence, not a validated instrument; programs adopting it should test it against their own outcomes as the text recommends. The composite case combines several engagements with details altered. This article closes the Government & Ecosystems series; companion articles cover program design (article 31), measurement (article 32), research and campus pipelines (articles 33 and 34), investment promotion (article 35), the corridor (article 36), incubator economics (article 37), skills (article 38) and agency governance (article 39).
References & further reading
- Azoulay, P., Jones, B., Kim, J.D. & Miranda, J. (2020) — Age and High-Growth Entrepreneurship, American Economic Review: Insights
- Gompers, P., Gornall, W., Kaplan, S. & Strebulaev, I. (2020) — How Do Venture Capitalists Make Decisions?, Journal of Financial Economics
- Kaplan, S., Sensoy, B. & Strömberg, P. (2009) — Should Investors Bet on the Jockey or the Horse? Evidence from the Evolution of Firms from Early Business Plans to Public Companies, Journal of Finance
- Gompers, P., Kovner, A., Lerner, J. & Scharfstein, D. (2010) — Performance persistence in entrepreneurship, Journal of Financial Economics
- Eesley, C. & Roberts, E. (2012) — Are You Experienced or Are You Talented? When Does Innate Talent versus Experience Explain Entrepreneurial Performance?, Strategic Entrepreneurship Journal
- Kerr, W., Nanda, R. & Rhodes-Kropf, M. (2014) — Entrepreneurship as Experimentation, Journal of Economic Perspectives
- Schmidt, F. & Hunter, J. (1998) — The Validity and Utility of Selection Methods in Personnel Psychology, Psychological Bulletin
- McKenzie, D. (2017) — Identifying and Spurring High-Growth Entrepreneurship: Experimental Evidence from a Business Plan Competition, American Economic Review
- Campos, F. et al. (2017) — Teaching personal initiative beats traditional training in boosting small business in West Africa, Science
- Karlan, D. & Valdivia, M. (2011) — Teaching Entrepreneurship: Impact of Business Training on Microfinance Clients and Institutions, Review of Economics and Statistics
- Ewing Marion Kauffman Foundation — The Anatomy of an Entrepreneur research series
HexGn’s founder programs select on evidence of readiness rather than promise — a rubric, a trial task and a mid-career track — and train the behaviours the research says matter. Designed for governments and agencies; delivered end to end; measured at 24 months.