How can we ensure Horizon Europe and the ECF deliver?

Europe is about to commit €400 billion to the next Horizon Europe programme and the new European Competitiveness Fund (ECF). Much of the debate about this decision has focused on what the money should fund and who should decide. Both matter. Far less attention has gone to a third question, one that matters just as much for Europe's competitiveness: how can we make sure that the money actually delivers?

The answer is experimentation: testing different approaches while building in, from the start, the systems to learn whether or not they work. The Horizon Europe and ECF regulations, now being negotiated, should embed this explicitly, creating a legal basis for experimentation, alongside dedicated metascience units in the European Research Council, the European Innovation Council and the European Commission’s directorate general for research. In addition, it should include the flexibility to scale up what works and, crucially, ringfenced funding. 

Concretely, 3% of both budgets should be set aside for policy R&D that implements experimental interventions and rigorously learns what works. The EU asks its member states to invest 3% of GDP in R&D. The Commission should apply the same logic to itself.

Effective organisations, programmes and policies share two often-overlooked traits. The first is agility, the capacity to move fast and question inherited routines. The second is the ability to learn, using data and evidence to know what to scale up, what to change and what to stop. These can sound like opposites, but experimentation reconciles them. It does not mean improvising or launching another pilot. It means testing something new with the systems in place, from the outset, to learn whether it works. Experimenting is not the risky option; spending billions without knowing whether or not this will be effective is.

This is not theoretical. Tech companies run thousands of experiments a year to refine their products. Economists have won the Nobel Prize for bringing the same discipline to the fight against poverty, benefiting hundreds of millions of people. Governments across the OECD are launching experimental interventions and building their internal capabilities to do so.

Experimentation makes a difference. The UK Metascience Unit has trialled new peer-review approaches in its funding calls, cutting review time by 53-65% without loss of quality. A wave of small-scale trials has demonstrated more cost-effective approaches to support SME competitiveness. For instance, cheap group-based consulting for firms was expected to be the poor cousin of intensive one-to-one support, but instead overtook it, delivering far higher sales and survival a decade on at a much lower cost. This is a result almost no one would have predicted, and none could have seen without a trial.

The recurring lesson is humbling. Small design choices in how an intervention is run, not just how much money sits behind it, routinely determine whether it succeeds. The intuition about what works best is frequently wrong; we can only find out if we test it.

At the EU level, almost none of this happens. The Heitor Report urged the Commission to build this capacity into the next Horizon Europe, but the recommendation has not been taken up. Beyond a few small pilot calls, experimentation remains a rare exception.

The reasons are structural. Programmes are launched with absolute conviction that they will work, not designed for continuous testing and iteration. Proposing to test them can be misread as an admission of ignorance rather than a mark of rigour. A pervasive audit culture, the Court of Auditors included, favours a tidy, hidden failure over a well-run experiment that shows something does not work. 

There is rarely a mandate to experiment, budget set aside, the units or skills to do it well, or the IT and data systems to support it. Even randomisation, one of the most powerful tools for learning what works, is often blocked by legal services worried about equal treatment, even though allocating by lottery among proposals too close to call is no less fair than ranking them on differences too small to be meaningful, and often themselves the result of bias.

A chance to experiment

Without permission, capability, budget and mandate, the status quo wins by default.

This is what makes the current negotiations an opening. The bottlenecks are structural, but structures are exactly what the Horizon Europe and ECF regulations are now setting, and once set, they hold for seven years. None of this requires redesigning either instrument, only writing in a legal basis for experimentation, dedicated capability and ringfenced funding. 

Crucially, the 3% set-aside would not take a euro from any existing programme. Budget holders would remain in control, free to use it to test different ways of delivering their own programme's goals, adapt as circumstances evolve, and take calculated risks in the face of genuine uncertainty. 

The result would be more effective interventions and stronger evidence on how to allocate the remaining 97% of the budget, to the direct benefit of the scientists and innovators these programmes serve. Mandating a minimum level of investment in experimental interventions is the most effective lever to overcome institutional inertia, deliver the highest return on investment, and ensure the outcomes from this €400 billion match the ambition placed on them.

Horizon Europe and the ECF cannot be set in stone for seven years while the world moves this fast. They need agility and continuous learning, together. Our competitors are not waiting, and the countries that pull ahead will be those that move fast and learn fast, getting it right most of the time. Embedding experimentation in Horizon Europe and the ECF is how Europe does so. 

Albert Bravo-Biosca is the founding director of the Innovation Growth Lab at Nesta, a global policy lab for more impactful innovation and productivity policies, and a resident fellow at the Barcelona School of Economics.