How grant evaluators actually score proposals, and what that means for how you write

Applicants write proposals in a state of deep familiarity. You know why the methodology is right, which partner brings which capability, what the numbers in the budget represent. An evaluator arrives with none of that, a fixed set of criteria, and a limited amount of time in which to form a judgement they will have to defend to colleagues. Everything that follows comes from that asymmetry.

What happens after you press submit

In most competitive programmes, a proposal passes through four stages: eligibility and admissibility screening, independent individual evaluation, a consensus step, and ranking against the available budget. The details differ between funders, but the shape holds.

1. Screening

Before quality is discussed at all, the application is checked against formal requirements. Ineligible applicants, missing mandatory content, exceeded limits and incomplete packages are removed here. No amount of project quality survives this stage, which is why the pre-submission checklist matters more than it feels like it should.

2. Individual evaluation

Several evaluators, usually working remotely and independently, read the proposal against the criteria, award a score for each, and write the strengths and weaknesses that justify those scores. They do not discuss the proposal with each other at this point, and they cannot ask you anything.

3. Consensus

The evaluators then reconcile their assessments, often with a moderator, and agree a single score per criterion and a joint written report. Scores move in both directions here. A weakness that two or three readers noticed independently is very hard to talk down.

4. Ranking and budget

Proposals above the thresholds are ranked by score, and the call budget is allocated down the list until it is gone. Strong proposals are rejected at this stage every round. The details of thresholds and weighting are covered in our guide to Horizon Europe evaluation criteria.

How an evaluator actually reads

Not linearly. An evaluator holds a criterion in mind and moves through the document hunting for the evidence it requires. Four practical consequences follow:

• Findability beats elegance. A plain sentence under a heading that matches the criterion scores better than a beautiful paragraph three sections away.
• Search costs count against you. If the evidence takes effort to locate, it is often recorded as absent. Not out of malice, but because the reader has four more proposals waiting.
• Tables and headings do heavy work. A criteria-to-section mapping table near the start is one of the highest return additions to any long application.
• The first page sets an expectation. A clear, specific summary earns the benefit of the doubt through the middle sections. A vague one does the opposite.

The language of evaluation reports

Evaluation comments are written in a restrained, repeatable register. Learning to hear it lets you predict your own report before it is written. The cost column below is indicative, since scoring is the panel's judgement and no rulebook prices a weakness, but the direction is consistent:

Notice that none of these comments concern the merit of the project. They concern what the document did or did not prove. That distinction is the whole of grant evaluation.

What makes a score drop

Scores fall for a small number of recurring reasons, and they are the same across programmes:

1. A criterion aspect that is never directly answered. It becomes a named shortcoming in the report, and it is the most preventable loss on this list.
2. Claims without evidence. Impact figures with no baseline, capacity claims with no track record.
3. Internal inconsistency. Effort in one table contradicting the budget in another. Inconsistency reads as carelessness and infects the reader's view of everything else.
4. Generic content. Text that could have been written for any call in the programme signals a recycled application.
5. Imbalance. Twelve pages on methodology and half a page on a weighted Impact criterion.

Our guide to why grant proposals get rejected works through each of these with the fix.

Writing for the way it will be read

• Use the call's vocabulary in your headings. If the criterion is called "Quality and efficiency of the implementation", do not head your section "Project management approach".
• Answer first, elaborate second. Put the direct answer in the opening sentence of the section, then justify it.
• Give every claim a number, a source or a name. Anything unsupported will be read as unsupported.
• Add a criteria mapping table. Tell the evaluator where each criterion is answered. It costs you a page and saves them the search.
• Allocate space by weight, not by interest. Count the pages you spend on each criterion and compare them with the points on offer.

Rehearsing the evaluation before you submit

You cannot un-know your own project, which is why self-review has limits. Two things get close to a real evaluator's view:

The cold reader. Give the criteria and the proposal to a colleague with no involvement in it, and say nothing else. No context, no "the idea is that". Ask them to score each criterion and name the weaknesses. The gap between their reading and yours is the gap the panel will find.

The evaluator simulation. This is what the Evalora Engine was built to do. It extracts the criteria and weights from your call document, scores the draft criterion by criterion, flags unmet requirements, ranks improvements by the points they recover, and produces a predicted score with reviewer-style comments per criterion. You get the report before the deadline instead of three months after it, and every run is saved so you can see whether your edits moved the score.

A general chat assistant cannot do this reliably, because it holds no fixed rubric and produces a different judgement on each run. That difference is explained in what ChatGPT, Claude and Gemini miss on grant proposals.

Frequently asked questions