EU Funding Proposals: Where Evaluators Take Points
A mature EU funding proposal can still lose points for a reason its authors consider minor: a claimed market need with no evidence, a work package that does not produce the stated result, or a partner described as excellent but assigned no meaningful task. Evaluators do not score the months spent building the consortium or the difficulty of the science. They score the document submitted against the published form.
That distinction becomes unforgiving near deadline. By then, most teams are refining language, collecting partner sign-off and repairing formatting. The more valuable question is usually harsher: if an evaluator had to justify a lower score against this criterion, which passage would they cite?
EU funding is assessed through the call’s form
Admissibility, eligibility and award criteria are separate gates. Missing a required annex, exceeding a page limit or failing an eligibility condition can prevent substantive evaluation altogether. A technically eligible proposal then has to earn its score under the award criteria and sub-criteria set out for that call and action type.
For many Horizon Europe actions, the familiar headings are Excellence, Impact, and Quality and Efficiency of the Implementation. But familiar headings do not make a universal checklist. The sub-criteria, thresholds, weighting and score bands depend on the action and the published evaluation form. Research and Innovation Actions, Innovation Actions, Coordination and Support Actions, Digital Europe calls and Erasmus+ applications do not ask precisely the same questions, and they should not be reviewed as though they do.
The official scale commonly runs from 0 to 5, often in half-point increments, with a threshold at criterion level and sometimes an overall threshold. A proposal can be above the overall threshold yet fail because one criterion is below its minimum. Where weighting applies, a weakness in a highly weighted criterion has a different practical cost from the same weakness elsewhere. Check the current call documents rather than relying on a score convention from the previous programme or a previous work programme.
A score is not a line edit. It is an evaluator’s judgement against score-band descriptors. “Good” can be a costly word if the form reserves the top band for a proposal that addresses the criterion convincingly and has no, or only minor, shortcomings. An otherwise strong draft with a significant weakness does not become top-scoring because its objectives are ambitious.
Where EU funding proposals lose points
Excellence: ambition without a testable route
Excellence is often weakened by a mismatch between the problem, objectives, methodology and expected outcomes. The opening pages may establish that a challenge matters, but the methodology never explains how the proposed approach will answer the stated research or innovation question.
Evaluators look for more than an attractive technical narrative. They test whether the objectives are specific, measurable where appropriate and achievable within the action. They compare claimed novelty with the state of the art, then look for a credible route through technical uncertainty. A table of risks does not repair a methodology that assumes every dependency will work.
Common deductions arise where a proposal calls an approach “beyond the state of the art” without identifying the relevant baseline, names a data source without demonstrating access, or presents validation as a final task rather than a designed method. For interdisciplinary work, the failure point is often at the boundary: one discipline supplies data, another develops a model, but no work package states who will integrate, test and interpret the combined result.
Impact: a credible destination, not a list of benefits
Impact sections tend to attract broad claims: competitiveness, resilience, jobs, standards, inclusion, emissions reductions. The issue is rarely that these outcomes are undesirable. It is that the proposal has not established the chain between its outputs and those outcomes.
An evaluator should be able to trace that chain. What will be produced? Who will use it? What changes in their practice, product, policy or capability? What must be true for that change to occur? Which barriers are within the consortium’s control, and which depend on regulation, procurement, adoption or follow-on investment?
Dissemination, exploitation and communication should serve that logic. A webinar programme and conference attendance are activities, not an exploitation pathway. A target group is not a user case. A KPI is not credible simply because it has a number beside it. If the proposal promises to reach 500 organisations, it should show the channel, ownership, timing, capacity and basis for the estimate.
For calls requiring contribution to expected outcomes and impacts, map each claimed contribution explicitly. Do not force every outcome into every work package. That creates repetition and makes attribution less believable. It is better to show a narrow, evidenced contribution than to promise system-wide transformation with no mechanism.
Implementation: the document must operate as a plan
Implementation is where a polished narrative often exposes its internal contradictions. Evaluators cross-check the work plan against objectives, deliverables, milestones, resources, governance and the role of every beneficiary or associated partner.
A work package title is not a work plan. Tasks need owners, inputs, dependencies, outputs and a plausible sequence. Deliverables should be useful evidence of progress, not labels such as “report” or “platform” with no acceptance condition. Milestones need decision value: what will the consortium do differently if the milestone is missed?
Resource tables are particularly revealing. A partner assigned responsibility for a critical technical component but given negligible person-months invites questions about capacity. The reverse also applies: large allocations with vague tasks look like budget distribution rather than justified effort. Explain the relationship between effort and work, especially where subcontracting, financial support to third parties, access to infrastructure, ethics procedures or external dependencies are material.
Governance deserves the same scrutiny. “The coordinator will manage the project” is not a management structure. The evaluator needs to see decision rights, escalation routes, quality control, risk ownership and how scientific, technical, commercial and stakeholder decisions are joined up. In a multi-country consortium, this is not administration. It is delivery risk.
Read the draft as six different evaluators would
An internal review frequently fails because the same people who made the draft coherent are asked whether it is convincing. They already know what an abbreviated sentence means, why a partner is trusted and how a missing dependency will be managed. An evaluator knows none of this unless the proposal establishes it.
A disciplined pre-submission reading separates the seats. One reading tests Excellence against the relevant sub-criteria. One tests Impact and the evidence behind every claimed pathway. One tests Implementation, resources, risks and governance. A consortium-capacity reading checks whether the named organisations and people can credibly deliver their allocated roles. A red-team reading hunts claims that the document does not support. A sector reading asks whether the narrative reflects the technical, policy or market conditions of the call.
Those readings should begin independently. If every reviewer sees the first reviewer’s conclusions, the team risks manufacturing consensus. Disagreement is useful evidence. When two readers differ by more than a point on a criterion, inspect the cited passages and establish whether the difference comes from an ambiguous claim, a genuine interpretation issue or an unstated assumption. That is adjudication, not averaging.
The useful output is not merely a predicted score. It is a criterion-by-criterion record of the weakness, the proposal text on which the finding rests, the applicable marking logic and the revision that would remove or reduce the concern. A finding without a passage cannot be checked. A score without a rationale cannot be prioritised.
Fix score risk in the right order
Do not begin with prose polish. First repair anything that could affect admissibility or eligibility. Then deal with threshold risk and significant weaknesses in the highest-weighted criteria. After that, resolve cross-document contradictions: an impact claim unsupported by work packages, a deliverable absent from the resource plan, or a risk register that omits the project’s central dependency.
Only then should the team improve clarity, shorten repetition and standardise terminology. These edits matter, but they should not displace a missing exploitation route or an implausible implementation plan. A concise proposal with a defensible intervention logic is safer than an elegant document that asks evaluators to infer its case.
BidShark applies this separation through independent automated readings against the hand-transcribed evaluation form for the relevant call type, with adjudication where criterion scores diverge and quoted evidence for each finding. It does not replace the real evaluation panel or guarantee an outcome. Its purpose is narrower: expose the weaknesses still present in a finished draft while the consortium can still correct them.
A submission deadline forecloses revision, not doubt. Before the proposal is sent, make each important claim survive the question an evaluator is entitled to ask: where, exactly, is the evidence?