Not one reports how many people it saved from starting the wrong one.
That asymmetry is not an accounting quirk. It shapes what gets taught, who gets admitted, which formats survive the next funding round, and what participants are implicitly told success looks like. If a programme’s headline metric is ventures founded, the programme will eventually optimise for ventures founded - including the ones that should not exist.
What “ventures founded” does to program design
Follow the incentive through a programme and the distortions show up in specific places.
It pushes toward incorporation. A registered company is countable in a way that “three months of serious customer discovery that ended in a well-evidenced no” is not. So participants are nudged toward registering, sometimes months before there is anything worth registering, which converts a cheap learning exercise into an expensive commitment with legal and tax obligations attached.
It produces pitch theatre. Demo day exists partly because it makes the output visible to funders. The skill it rewards - presenting a venture as more certain than it is - is close to the opposite of the skill early-stage founders actually need, which is holding uncertainty accurately in their own heads.
It selects idea-first. If the KPI is companies, the rational admissions strategy is to admit people who already have something resembling a company. Which is exactly the doorway problem from Article 6, arriving through the reporting system rather than through the application form.
And it punishes honesty. A participant who concludes “I have now seen this up close and it is not for me” has arguably received the highest-value outcome the programme can deliver. In the reporting, they appear as a non-outcome. In some funding logics, as a failure.
The metric also hides most of what programmes actually produce. The participant who joined someone else’s team. The one who became an unusually effective early employee at a startup because they understood what the founders were dealing with. The one who came back three years later as a mentor. The one who founded something in year five, long after the reporting period closed. None of these appear in “ventures founded in the funding year.”
The research field mirrors the programme problem, which should be a warning rather than a comfort. Studies use intention as a proxy for entrepreneurial behaviour because intention can be measured with a questionnaire and behaviour requires longitudinal tracking. Programmes count incorporations for the same reason: it is the number you can get, not the number you need.
We measure entrepreneurship education by startup count for the same reason we measure schools by test scores.
What to measure instead: agency, readiness, and informed decisions
The alternative starts with naming the actual product.
Entrepreneurial agency is the capacity to recognise that you can actively influence your environment by taking initiative and creating solutions. It is the difference between “someone should solve this” and “what could I do about this?” It transfers into founding, into employment, into research, into public administration, into any context where someone has to decide whether a situation is fixed or changeable. It is plausibly the most valuable thing entrepreneurship education produces, and almost nobody reports it.
Founder readiness is the intermediate outcome that sits between agency and a company: can this person identify a problem, talk to customers, test assumptions, work in a team, make decisions with incomplete information, learn from evidence, and take ownership? That is measurable through evidence of activity, and it is measurable during the programme rather than five years later.
An informed decision is a legitimate outcome in its own right. The purpose of entrepreneurship education is not necessarily to turn every participant into a founder. It is to give people enough knowledge and experience to make an informed decision about entrepreneurship - which means treating it as a genuine career alternative alongside employment, freelancing, academia and corporate careers, rather than as a calling that only the insufficiently brave decline.
Practically, a reporting set built on that logic could include: customer conversations conducted; assumptions explicitly tested and killed; prototypes put in front of real users; first revenue events; participants who joined other teams; participants who return as mentors; ventures started three to five years after the programme; and yes, documented informed decisions not to start.
One caution, because this argument can be over-rotated. Failure should not be romanticised. The objective is never to fail - it is to test assumptions cheaply enough that failure produces useful information before too many resources are committed. Hypothesis, experiment, evidence, learning, adaptation. A programme that celebrates failure as such has just replaced one bad metric with a worse one.
The policy environment is beginning to catch up. The Nationales MINT Forum’s 2026 position paper on entrepreneurial culture explicitly calls for a nationwide monitoring system for entrepreneurship competences and effectiveness, and puts the underlying diagnosis bluntly: “Fehlervermeidung, Leistungsbewertung und Fachlogik stehen oft vor kreativem Problemlösen, Experimentieren, Selbstwirksamkeitserfahrungen und eigenverantwortlichem Handeln.” Error avoidance, performance grading and subject logic ahead of creative problem-solving, experimentation, self-efficacy and independent action. That is a description of an assessment system, and assessment systems are built out of metrics.
For anyone who needs a ready-made structure, EntreComp - the European Entrepreneurship Competence Framework published by the Joint Research Centre - already defines 15 competences across three areas (ideas and opportunities, resources, into action), broken into 60 thematic threads with eight proficiency levels. The instrument exists. What is missing is the willingness to report against it instead of against incorporations.
None of these measures fit neatly into a one-year funding report. Which links straight back to the funding structure in Article 5, and is precisely why the two problems have to be solved together.
A programme that only counts companies will eventually only produce companies. And not good ones.
