Proposals rarely fail because the science is weak. They fail because the Impact section was written last, by whoever had time, in the final week before the deadline. Evaluators notice this immediately, and it costs more points than almost anything else in the document.
It helps to picture the reading conditions. An expert evaluator is working through a batch of proposals, individually, against a scoring grid, often alongside a full-time job. They are not looking for reasons to reject you. They are looking for the sentence that lets them justify a score to the other evaluators in the consensus meeting. If that sentence is not in your proposal, they have to invent it - and they will not.
What is actually being scored
Under Horizon Europe, the three criteria are scored on a 0–5 scale, half points allowed. Individual criteria carry a threshold, and there is a combined threshold across the three. The exact thresholds and tie-breaking rules vary by action type and are stated in the call conditions - read them rather than assuming, because they change.
What matters more than the arithmetic is what each score band feels like from the evaluator's side:
| Score | What the evaluator is thinking |
|---|---|
| 5 | I could quote this section to explain why the project should be funded. |
| 4 | Convincing, with one or two gaps I would want addressed. |
| 3 | Plausible, but I am doing some of the work to believe it. |
| 2 | The ambition is stated, the pathway is not. |
| 1 | Generic. Could describe any project in this call. |
The gap between a 3 and a 4 is almost never ambition. It is specificity.
The three failures that cost the most
Impact stated as adjectives. "Significant improvements in efficiency" is not an impact claim, it is a hope. "Reduces planning time from six hours to under one for a network operator of this size, verified against operational data in WP5" is a claim an evaluator can score. If a number cannot be supported yet, say what will be measured, by whom, and against what baseline.
No pathway between result and outcome. Many proposals jump from "we will build a platform" to "European competitiveness will improve." The pathway in between - who adopts it, why they would, what has to be true for that to happen, what stands in the way - is the part being scored. Skipping it reads as though nobody has thought about the day after the project ends.
Exploitation written as an afterthought. Key exploitable results listed generically, with no owner, no route to market and no indication of what happens to the IP. This is the single most common weak spot, and it is also the cheapest to fix: it is a table, an owner per result, and two honest sentences each.
What a strong section does structurally
- States the target outcome first, in the terms the call itself uses. Evaluators are checking alignment with the destination and expected outcomes, and it is much easier to award that point when your language matches theirs.
- Quantifies the baseline before quantifying the improvement. A claimed 30% gain means nothing without the current figure.
- Names the barriers to adoption and what the project does about each. Acknowledging a barrier does not weaken the case; it is evidence you understand the market.
- Assigns every key result to a partner who has a reason to exploit it. A result owned by a university with no commercialisation route is a result that dies at the final review.
- Handles the mandatory sections - dissemination, communication, IP management - as substance, not as boilerplate. Evaluators have read the boilerplate hundreds of times.
The week-before test
If you have limited time, run this check. Take your Impact section, remove the project name and the technology, and read what is left. If it could describe any of the other proposals submitted to the same call, you are sitting at a 3.
Then take each claim and ask: what would have to be true for this to happen, and does the proposal say who makes it true? Every claim that fails that test is either cut or given an owner.
Neither of those requires new science. They require someone to treat Impact as a technical section with the same rigour as the methodology - which is exactly what it is.
When to walk away instead
Occasionally this exercise reveals that the proposal has no credible pathway at all: the results have no obvious buyer, the consortium has no route to market, and the call's expected outcomes were only ever a loose fit. That is worth knowing three weeks out rather than five months later.
A proposal costs weeks of senior time before it earns anything. The decision not to submit is a legitimate outcome of a good review, and it is usually the most valuable one on offer.