How Horizon Europe Proposals Are Scored: Thresholds, the ESR, and What 3/3/3 Means

Every Horizon Europe applicant eventually holds the same two-page document: the Evaluation Summary Report, or ESR, with three scores and three blocks of comments. Most read it once, in the first hour after the result letter, and never decode it. That is a waste, because the ESR is written to a published standard by experts who were briefed with a published slide deck, and once you know the rules the report tells you precisely why the proposal landed where it did and what a resubmission has to change. This page sets out those rules from the Commission’s own evaluator briefing and then reads two large samples of real ESRs against them.

The evaluation in four steps

After the admissibility and eligibility check, a proposal passes through three expert stages. In the individual evaluation, each expert, usually three or more, reads the proposal alone and completes an Individual Evaluation Report with a comment and a score for each criterion. The briefing tells them to evaluate the proposal as submitted, not on its potential, to reflect any shortcoming beyond minor ones and clerical errors in a lower score, to make scores match comments, and to explain shortcomings without making recommendations. In the consensus step the same experts meet, moderated by Commission or agency staff, and agree comments before agreeing scores; for full proposals they are told not to converge immediately on the average, while for first-stage proposals the average is the starting point. A rapporteur drafts the consensus report. In the panel review a wider panel checks consistency across proposals, resolves minority views, cross-reads proposals with equal scores, endorses final scores and comments, and recommends a ranked list. The briefing is explicit that the consensus report often remains unchanged at panel stage, so in most cases the ESR you receive is the consensus report verbatim.

The scale, in the words experts are given

Scores are awarded per criterion, not per aspect within a criterion, in steps of 0.5 from 0 to 5, for a maximum of 15. The briefing asks experts to use the whole range and defines each integer as follows.

Minor shortcoming, shortcoming and significant weakness are defined terms in the evaluator briefing, not loose description, and they are the key to reading a comment. Each one tells you how far the score moved and whether the proposal is still fundable.

The practical consequence is that a single significant weakness on any criterion is fatal on its own, while a page of minor shortcomings is not. Count the defect words in your ESR before counting the comments.

Thresholds, weighting and the 3/3/3 trap

The default threshold is 3 for each criterion and 10 for the total, unless the work programme says otherwise, and a proposal must pass both to be considered for funding. The arithmetic has a consequence that surprises many first-time applicants: three scores of exactly 3 pass every individual threshold, total 9, and are rejected. “Good” everywhere is not fundable. In practice a fundable full proposal needs at least one 4 and no score below 3, and in competitive topics the funding line sits well above 10.

Passing is not being funded: ranking and ties

Once thresholds are met, proposals are ranked by total score and funded in order until the topic budget is exhausted. Those above threshold but outside the budget are placed on a reserve list, and in calls that provide it, receive a Seal of Excellence that some national and regional funders honour. Because many proposals cluster at 13 to 14.5, the tie-break order matters more than applicants expect. For each group of proposals with the same total, the panel applies the following sequence.

1. Proposals addressing aspects of the call not otherwise covered by higher-ranked proposals are prioritised.
2. Within that group, priority goes to the higher Excellence score, then Impact. For Innovation Actions the order is Impact first, then Excellence.
3. If still tied, the gender balance among the personnel named as primarily responsible in the researchers table.
4. Then geographical diversity: the number of Member States or Associated Countries in the proposal not otherwise funded higher up the list, and if equal, budget.
5. Finally, other factors related to the call or to Horizon Europe generally, such as portfolio synergies or SME involvement.

Two practical consequences follow. For a RIA, an extra half point on Excellence is worth more than the same half point on Implementation, because it breaks ties. And the researchers table in Part A is not decorative: it can decide a tie at step 3.

What the ESR contains, and what it is not allowed to contain

The ESR arrives with the evaluation result letter through the Funding & Tenders Portal; the programme’s service standard is to inform applicants within five months of the deadline and to sign grants within eight. The report has a main part with comments and a score for each of the three criteria, plus any remarks on scope, on activities excluded from funding, and on operational capacity. The briefing sets a quality standard that experts are reminded applicants will read and can use in an evaluation review. The report must reflect strengths and weaknesses fairly and give reasons for the scores; it must address all aspects listed under each criterion and only those; it must not contain comparative statements about other proposals, because each proposal is judged on its own merit; it must not penalise the same weakness under two criteria; and it must not contain recommendations or suggestions for improvement. Comments must be precise, verified, and consistent as a whole with the meaning of the score awarded.

Three further rules are useful to applicants. For second-stage proposals, inconsistencies between the stage-1 and stage-2 ESR must be specifically justified. For resubmissions declared in the form and made within two years under comparable conditions, comments and scores that differ significantly from the previous ESR must also be justified. And a proposal in a blind first stage that directly identifies an applicant in Part B or the abstract is declared inadmissible before any of this happens; the rules and a checklist are in the two-stage and blind calls guide.

What real ESRs say: two analyses

Two National Contact Point teams have read large samples of ESRs and coded the comments. The Horizon Academy project, led by the Slovak NCP centre CVTI SR, analysed 130 reports from December 2021 to March 2024, taking the five highest- and five lowest-scored ESRs per call across every Pillar II cluster and the Widening component, for RIA, IA and CSA actions. The French Health NCP analysed 189 reports from the 2021 Health cluster RIA topics with a total score of 11 or more, out of 501 proposals submitted, 207 of which, or 41 percent, fell below the overall threshold. The recurring comments line up closely with the aspects in the template, which is the point: experts are answering the template’s questions, and the comments record which ones went unanswered.

Two findings from the samples are worth stating on their own. The Horizon Academy authors note that where a proposal is written understandably, makes a good impression and covers almost all requested aspects, experts are tolerant of minor shortcomings and will give the top score; the losses come from aspects that were simply absent. The French team found little difference between funded and unfunded proposals in thenumber of negative comments on the most commented aspects; what differed was whether those comments used the vocabulary of shortcomings or of significant weaknesses. Read your ESR for that vocabulary, not for the count.

Reading your own ESR: the resubmission decision

Put the three scores next to the thresholds and the funding line for your topic, which the result letter or the NCP can tell you, and classify the outcome. The bands below are a working heuristic built on the descriptor definitions and the two ESR samples, not an official rule.

Whichever band you are in, the rewrite should start from the ESR comment, then the corresponding template question in the Part B template walkthrough, and only then the old text. If the comments point at implementation tables, the work package and Gantt worked example shows what a fully specified plan looks like. For the broader strategy of when a resubmission is worth the effort, see The Resubmission Renaissance.