Should I start a company? Self-assessment and opportunity cost

Core idea#

"Should I start a company?" is really three separate questions, and most researchers blur them together:

1. Is there a company here? Can the technology become a product someone will pay for, at a scale investors care about? (Playbooks 1.07, 1.08, 1.19 test this.)
2. Am I the right person to lead it, now? Do I have, or can I quickly assemble, the skills, team and staying power? (This playbook and 1.09.)
3. Can I afford to? What do I give up in money, career position and time, and can my household carry that for five to ten years? (This playbook and 1.05.)

A "yes" to the first question doesn't oblige you to say yes to the other two. Plenty of good technologies are better licensed to an existing company, or led by someone else while the inventor stays as an adviser or CSO (playbook 2.12). The aim here is to make the decision deliberately, with numbers, rather than drifting into it or out of it.

Who it's for: PhD students, postdocs, faculty and industry researchers who have a technology or a problem in mind and are weighing whether to build a company around it. It is most useful 6 to 18 months before you'd incorporate.

What the data says#

Age and experience help, not hurt#

A US Census study of 2.7 million founders who hired at least one employee between 2007 and 2014 found that the mean founding age of the fastest-growing 1-in-1,000 new firms was 45.0, and the pattern held in high-tech sectors and for successful exits (Azoulay, Jones, Kim and Miranda, AER: Insights 2020). A 50-year-old founder was about 1.8 times as likely as a 30-year-old to build one of those top firms, and prior work in the same industry more than doubled the odds (Jones and Kim). Founders in biotech and energy were older on average, at around 47 (Weiss Ratings, summarising the MIT study).

For a researcher, this matters in two ways. A postdoc of 32 isn't "late". And the years spent in a field are an asset, if they're paired with exposure to the customer side.

The median founder earns less#

Barton Hamilton's well-known study found that most self-employed people earn less initially and see slower earnings growth than in paid work, and that the gap isn't explained by lower-ability people choosing self-employment (Hamilton, JPE 2000). Median self- employment earnings were around 35% below median wages; only around the 75th percentile did the order reverse (summary in a later paper). A Belgian study found that people who returned to employment after running a business earned about 27% less than matched peers, and the gap persisted five years later (Belgian administrative data).

These studies cover all self-employment, not venture-backed deeptech. But they set a sober baseline: the typical outcome is financially worse, and the upside is concentrated in a small share of companies.

Deeptech takes longer#

• A 2025 analysis estimated average time to exit at about 5.2 years for SaaS, six to eight years for hardware, and over 11 years for biotech platforms (RETVRN Research via e27).
• Indian investors told the Financial Express in November 2025 that deeptech exit horizons they used to plan at 10 to 12 years are shortening to six to nine (Financial Express).

Plan your personal commitment on a seven to ten year horizon, not three to five.

What this means for the decision#

The data argues against two common views. "I'm too old or too academic to found" is not supported. "Founding is a good financial bet for me" is only true if you have reason to think you're in the upper tail, or if the non-financial returns (the mission, the autonomy, the learning) matter a lot to you. Be honest about which of these you're relying on.

A self-assessment you can actually use#

Personality quizzes ("are you a risk-taker?") predict very little. What helps is answering concrete questions with evidence, then scoring yourself honestly. Do this on paper, then ask one person who knows you well and one who has founded a company to score you too. Where your score and theirs differ by two or more points, that's the conversation to have.

The five questions#

Score each from 1 (no evidence) to 5 (strong evidence).

How to read the total#

• 20 to 25: The case is strong. The question is timing and team, not whether.

• 14 to 19: Typical for first-time researcher-founders. Find the low scores and work out what would raise each one in the next six months: customer interviews for pull, a business co-founder for range, a savings plan for room.

• Below 14: Not a "no", but not yet. Consider a lower-commitment route first: a fellowship (1.10), an incubator place while still employed (1.11), an I-Corps cohort (1.19), or licensing the technology and staying on as an adviser.

Two scores matter more than the total. A 1 or 2 on pull means you have a technology looking for a problem; fix that before anything else (1.07). A 1 or 2 on room means the risk falls on people other than you; fix that before you quit.

Worked example#

Priya is a fourth-year postdoc in battery materials. Her scores:

• Pull 3. Two cell makers have tested her samples; neither has talked price.
• Edge 4. A patent filed with her institute; a process nobody else has scaled past grams. Range 2. She has never sold or hired.
• Stamina 4. A PhD through a lab move and a failed first project.
• Room 3. Her partner earns; savings cover 12 months.

That's 16. Her plan: run 30 customer interviews in three months to test pull; look for a cofounder with commercial experience in batteries; build savings to 18 months; apply to a fellowship that pays a stipend. She re-scores in six months, and decides then.

Working out the opportunity cost#

Opportunity cost is what you give up by choosing this path over the next-best one. For researchers it has three parts, and the third is the one people forget.

1. Money#

Add up, year by year for the next seven years:

• The salary you'd earn on your next-best path (a faculty post, an industry R&D job, staying a postdoc), including pension contributions and benefits.
• The founder salary you'd realistically pay yourself. Pre-seed founders often take little or nothing in the first year, then something below market after the seed round.

The difference, summed, is your cash cost . Investors rarely see it, but you and your family will.

2. Career position#

• Academic track. Leaving a tenure-track or permanent post can be hard to reverse. Ask now: does your institution offer sabbatical or entrepreneurial leave (playbook 1.03)? Will a lien on your post or a dual role be allowed?
• Publication record. A two to three year gap in publications can close academic doors. If staying academically viable matters to you, plan for it (shared lab work, co-authored papers, an adjunct role).
• Industry track. Founding usually raises your value to industry employers, even if the company fails, especially if you've hired, raised money or shipped a product.

3. The option you're not taking#

The right comparison isn't "founding versus doing nothing". It's founding now versus:

• founding in two years with more savings, more customer evidence and a co-founder;
• licensing to a company and taking royalties or equity while keeping your job;
• joining an existing startup in your field as an early technical hire, getting most of the learning with less of the risk.

A simple model#

An illustrative example only; put your own figures in.

Now ask: what chance of what outcome would make up for 300? If your founder stake after dilution might be 10% of a company sold for 300 times your salary, that's a payout of 30 times salary (3,000 in these units) in the good case. You'd need about a 1-in-10 chance of reaching that outcome just to break even on the cash cost alone. For most companies, the odds are lower than that, which is why the non-financial reasons need to be real ones.

This isn't an argument against founding. It is an argument for going in knowing the price, and for reducing it where you can: fellowships and grants that pay a stipend, leave rather than resignation, a part-time start while the technology is de-risked.

How the decision differs in India, the US and the UK#

Policies and schemes change; check your institution's current rules and the scheme websites.

Scripts#

Asking a founder to score you#

"I'm deciding whether to spin out a company from my research in the next year. Would you look at five questions and score me 1 to 5 on each? I'd rather hear the low scores now, from you, than from investors later."

Talking to your partner or family#

"Here's what it would cost us: about [X] less income a year for the next three years, and maybe longer. Here's how much savings we'd keep untouched no matter what, and here's the date we'd review it. If by then I don't have [funding / a paying customer / a co-founder], I go back to a salaried job."

A pre-agreed review date and a "stop" condition make the risk bounded, which is easier for everyone to agree to.

Talking to your PI or head of department#

"I'd like to explore spinning out the [technology]. I want to do it in a way that's good for the lab and the institute. Can we talk about the options: leave, a dual role, or my staying on and someone else leading? And who in the tech transfer office should I speak to first?"

Common mistakes#

• Deciding on the technology alone. A great result in the lab is the start of the question, not the answer.
• Comparing founding to nothing. Compare it to your realistic next-best path and to founding later.
• Ignoring the household. The cost falls on others too. Agree the plan, the runway and the stop condition with them in advance.
• Quitting before checking leave options. Many institutions offer sabbaticals, leave or dual roles; resigning first can throw these away.
• Treating "not yet" as "no". A score of 14 is a list of things to fix, not a verdict.
• Believing you're too old or too academic. The evidence points the other way, as long as you pair research depth with customer exposure.

Related playbooks#

• 1.03 Leave, sabbatical or dual role
• 1.05 Personal runway: savings and the quit date
• 1.07 Tech-push vs market-pull
• 1.09 Founder-market fit
• 1.17 Talking to family
• 1.18 Solo founder or team?

• 2.12 Professor as co-founder, adviser or CSO

Sources#

• Azoulay, Jones, Kim and Miranda, "Age and High-Growth Entrepreneurship", American Economic Review: Insights 2(1), 2020 (abstract; MIT News).
• Jones and Kim, "Most successful entrepreneurs are older than you think".
• Hamilton, "Does Entrepreneurship Pay?", Journal of Political Economy 2000 (abstract).
• Persistent earnings penalty for former entrepreneurs, Belgian administrative data.
• RETVRN Research, State of Exits 2025, via e27.
• Financial Express , "Deeptech's long road to exits getting shorter", 24 November 2025.

Last checked November 2026.