Contents08 sections
An accelerator opens its next call: "Apply with your team and your idea." Hundreds of applications come in. Motivated teams, good decks, lots of energy. The technology behind them is mostly what anyone can get.
Two streets away, at the research institute, there are patents nobody applies with. A process that has run stable in the lab for two years. The researcher who developed it wants to publish, not pitch.
The teams have no technology. The technology has no team. And the model waits for the two to find each other.
The assumption almost everything is built on
The example is made up, the pattern isn't. Almost all startup support starts with the founder. Accelerator, incubator, pitch competition: first comes a team with an idea, and everything else builds on it.
In 2023, Romme, Bell and Frericks took exactly this assumption apart in the Journal of Organization Design. Their finding: the entrepreneurship literature takes a lead founder and their team as the starting point. It underestimates where the technology comes from and treats an invention as if it were simply lying there for a founder to exploit. For software, that may hold. For deep tech, it doesn't.
Their alternative is a co-creation perspective. The founding team isn't the starting point there. It is something still to be formed, embedded in a constellation of institutes, inventors, talent and capital.
An accelerator turns itself around
The authors base this on a case they know from up close: HighTechXL in Eindhoven. From 2014 to 2018 it ran as a classic accelerator, with teams applying with their ideas. The record of those years: 64 ventures, 6 successful exits, 13 more still in the portfolio. The authors themselves put the success rate at around 10 percent.
In 2018 came the change of course, and the reasoning is remarkably honest. Many teams simply had no access to the frontier technology their value proposition would have needed. At the same time, research institutes were sitting on a large pool of breakthroughs that hardly found their way into application.
Since 2019, the builder has reversed the order. It sources patented technology from institutions such as CERN, ESA and TNO, evaluates it together with the inventors, and only then assembles the team. They call the format a reverse hackathon: teams don't look for a problem; a technology looks for its people.
What the numbers show, and what they don't
From 2019 to 2022, 26 ventures came out of this. In April 2023, 18 of them were active. Eight were stopped, and they share one thing: none of them could specify a convincing value proposition.
Honesty belongs here too. This is a single case study. Two of the three authors helped build the builder themselves. The new ventures are young, the old ones had years, and the authors write themselves that the success rate can only be assessed properly from 2028. Anyone setting "18 of 26 against 10 percent" is comparing apples with unripe pears.
A case study isn't proof. It's a pointer. And it matches what technology-transfer research has found for more than twenty years.
The invention is an embryo
In the late nineties, Jensen and Thursby asked 62 US research universities what they actually license. The answer is sobering. Around 48 percent of licensed inventions were a mere proof of concept without a prototype at the time of licensing, another 29 percent a lab-scale prototype. Only 12 percent were ready for commercial use.
And then the number that matters here: for 71 percent of the inventions, the universities said successful commercialisation requires the inventor's further cooperation.
That has two consequences. First: an invention isn't a finished product a founder just has to collect. Second: whoever takes the inventor out of the equation takes out the knowledge without which the technology won't grow up.
The inventor stays. But not alone.
If the inventor is indispensable, why not simply let them found the company? Because between the lab and a company lies work that has little to do with research: finding the market, selling, raising money, leading people.
Lundqvist studied 170 ventures from 16 Swedish university incubators, incorporated between 1995 and 2005. For 59 of them, the incubator helped bring an external entrepreneur on board, a so-called surrogate entrepreneur. These ventures grew significantly faster and earned more revenue than the rest. To be fair: this is an observational study. The incubators may well have picked the more promising cases for an external entrepreneur.
Back in 2001, Franklin, Wright and Lockett arrived at a matching picture in a survey of 57 UK universities. The universities with the most spinouts were the most open to external entrepreneurs. And as the best route, the authors named not replacement but combination: academic founder plus external entrepreneur.
To be fair: researchers aren't bad founders
It would be convenient to claim that scientists simply can't run companies. That isn't true, and the data say so clearly. Czarnitzki, Rammer and Toole compared academic spinoffs in knowledge-intensive industries in Germany with startups from industry. The academic ones grew employment 3.4 percentage points faster, and founders from research did slightly better still. And Zucker, Darby and Armstrong show for biotechnology: whether a firm's scientists co-publish with star scientists strongly predicts its success.
Both support the thesis rather than refute it. They show how much the researchers' knowledge is worth. They don't show that the same person should also take on sales, fundraising and leadership. And they only measure the spinouts that exist. The many inventions that never crossed the threshold appear in no statistic.
A study that looks uncomfortable at first fits in here. Ewens and Marx studied VC-backed US startups in which the founder was replaced as CEO. Looked at naively, these firms do worse. Estimated causally, the change improves performance. The lesson isn't that founders should leave. The lesson is that the role has to fit the phase, and that it pays to divide roles right early instead of later under pressure.
What this means in practice
Vohora, Wright and Lockett studied where spinouts get stuck. They name four critical junctures: opportunity recognition, entrepreneurial commitment, credibility and sustainability. None of them is a research question. And at none of them does waiting for the right team help.
That gives an order which, in my experience, works:
First: evaluate the technology first, together with the inventor. Maturity, IP, cost curve. Not behind closed doors, but with the person who knows it best.
Second: settle the value proposition before the team. The eight stopped ventures in Eindhoven didn't fail on the team. They failed because nobody could say what a customer pays for. That question belongs before incorporation.
Third: build the team around the gaps. Who's missing follows from the technology and the market, not from the list of applicants. Almost always it's someone who can sell and raise money.
Fourth: give the inventor a real role. As CTO, as chief scientist, with equity and with time. 71 percent of inventions need them. An honorary seat on the advisory board won't do.
Don't ask who wants to found a company. Ask what this technology needs to become one. Then build it the team.
Sources
- Romme, Bell, Frericks, "Designing a deep-tech venture builder to address grand challenges and overcome the valley of death", Journal of Organization Design 12(4), 217–237 (2023), open access: entrepreneurship literature starts from a lead founder and underestimates technology sourcing, co-creation perspective; HighTechXL 2014–2018 64 ventures, 6 exits, 13 in portfolio, success rate around 10 %; change of course 2018, technology from CERN, ESA, TNO and others since 2019; 26 ventures 2019–2022, 18 active in April 2023, all 8 stopped ventures without a convincing value proposition; single case, two authors involved in building it, valid evaluation only from 2028 — doi.org, manuscript (TU Eindhoven)
- Jensen, Thursby, "Proofs and Prototypes for Sale: The Licensing of University Inventions", American Economic Review 91(1), 240–259 (2001); figures from NBER Working Paper 6698 (1998): survey of 62 US research universities, 48 % proof of concept without prototype, 29 % lab-scale prototype, 12 % ready for commercial use, 71 % require the inventor's cooperation for success — doi.org, nber.org
- Lundqvist, "The importance of surrogate entrepreneurship for incubated Swedish technology ventures", Technovation 34(2), 93–100 (2014): 170 ventures from 16 incubators (1995–2005), 59 with a surrogate entrepreneur, significantly higher growth and revenue — doi.org
- Franklin, Wright, Lockett, "Academic and surrogate entrepreneurs in university spin-out companies", Journal of Technology Transfer 26, 127–141 (2001): survey of 57 UK universities, high-spinout universities more open to surrogate entrepreneurs, combination of academic and surrogate entrepreneur as the best approach — doi.org
- Czarnitzki, Rammer, Toole, "University spin-offs and the 'performance premium'", Small Business Economics 43(2), 309–326 (2014): academic spinoffs in Germany with 3.4 percentage points higher employment growth than industry startups — doi.org, ZEW Discussion Paper 13-004
- Zucker, Darby, Armstrong, "Commercializing Knowledge: University Science, Knowledge Capture, and Firm Performance in Biotechnology", Management Science 48(1), 138–153 (2002): co-authorship with star scientists strongly predicts firm success — nber.org
- Ewens, Marx, "Founder Replacement and Startup Performance", Review of Financial Studies 31(4), 1532–1565 (2018): estimated causally, replacing the founder-CEO improves performance; the naive correlation is negative — doi.org
- Vohora, Wright, Lockett, "Critical junctures in the development of university high-tech spinout companies", Research Policy 33(1), 147–175 (2004): four critical junctures — opportunity recognition, entrepreneurial commitment, credibility, sustainability — doi.org
