Summary

  • In 30 proposal evaluation reports sampled by the European Court of Auditors, only three assessed expected turnover, employment and scale-up potential; another 27 gave impact scores with gaps in their explanation.
  • The Commission required final project reports and exploitation plans, yet did not systematically assess research outputs or follow those plans after closure. Six of seven compulsory indicators for the industrial-leadership priority lacked baselines.
  • The audit found real technical progress among visited SMEs, including 10 of 13 that moved from concept to validated innovation. Market uptake was a different test, and the programme did not reliably collect the evidence to judge it.

The paper trail stops where the commercial risk begins

The European Court of Auditors' October 2026 report turns a familiar grant ceremony inside out. A research team wins an award. It builds a prototype. It submits a final report with plans to disseminate and exploit the result. The administrator verifies closure. The public hears that innovation has been funded. Then the question that matters to a customer, employee or investor starts: what happened next?

For Horizon 2020's industrial-leadership priority, the auditors found no systematic Commission assessment of research outputs at closure and no structured follow-up on exploitation plans. They found no evidence of a Commission attempt to assess commercialisation and scale-up potential during or after the projects. The resulting gap reaches beyond retrospective reporting. Without knowing which finished research has a credible route to market, the funder lacks an evidence base for choosing targeted follow-on support.

The report is about a defined funding priority, not a verdict on every European research grant. Its finding also has a forward edge: the auditors say the proposed 2028–2034 Horizon Europe programme still envisages verification at project closure without systematic assessment of outputs or structured post-project follow-up. That proposal could change. A dated commitment to measure later commercial outcomes would change the judgment; the current audit records no such mechanism.

The impact score could not explain itself

The weak handoff begins before a project wins. Among 30 sampled proposal evaluation reports, expected turnover, employment and scale-up potential were assessed in only three. Fifteen more used turnover and staff counts in scoring without explaining how the scores were derived. The remaining 12 assigned an overall impact mark without a justification for each indicator or a calculation of the combined score. Those are sample counts; multiplying them into a programme-wide failure rate would repeat the very measurement error the audit exposes.

Selection rules shape who later gets to sell a result. Collaborative projects accounted for nearly 80% of Horizon 2020 funding. The auditors found that SMEs rarely coordinated those projects, limiting their influence over direction, intellectual property and commercial use. An SME can be invited into a consortium, appear in the participation count and still lack the position to decide what becomes of the work. The distinction between a seat and control is not academic when the grant ends and a product needs an owner.

The record is not a story of empty laboratories. In the SME Instrument group the auditors visited, 10 of 13 mono-beneficiary SMEs moved technology from concept to validated innovation. Three successfully commercialised outputs and had received coaching. The same small visit sample showed two firms with zero growth and no route to more finance, while seven reached the commercialisation stage without market uptake. Technical transfer can succeed while sales stall. The Commission's task was to learn where that fork occurs, not to treat a prototype or a coaching session as a sale.

A dashboard without the afterlife

Six of the seven compulsory performance indicators for industrial leadership had no baseline. The Commission collected information during projects, not after they ended, and did not gather the participant economic-performance data needed to test whether support produced lasting value. The auditors also found that macroeconomic models used in ex-post evaluation assumed complete commercial success for supported innovations despite the absence of post-funding outcome information. A model can estimate GDP to 2040; it cannot replace the missing company-level record beneath its assumptions.

This creates a tidy institutional incentive. The administrator can count awards, participants, reports, publications and patents while the difficult denominator—firms whose results reached paying users—falls outside the reporting window. That does not make each grant a failure. It makes the success claim harder to audit than the expenditure. The original ECA report recommends identifying finished projects with strong commercial potential and directing later-stage support to the companies that drove them. Its target date is 2027; a target is not implementation.

The next useful public receipt would follow a defined cohort beyond closure: what was actually produced, who controlled the exploitable result, which firms received later finance, which reached market, which stopped and why. Keep selection scores beside the outcomes so a good result does not retrospectively excuse a poor selection process. Brussels can close a file with a signature. It can establish competitiveness only by following the invention out of the file.

Sources: ECA publication and report; original special report 24/2026 PDF.