Notes on EU evaluation

How to Audit Proposal Logic Before Submission

← All articles

A proposal can contain credible science, experienced partners and a substantial budget, then lose points because its argument does not survive an evaluator’s first reading. The usual failure is not poor writing. It is broken logic: a need is asserted but not evidenced; a work package produces an output that does not lead to the claimed outcome; a KPI measures activity rather than impact. Knowing how to audit proposal logic before submission means testing these connections under the same constraints that govern the real evaluation.

For Horizon Europe, Digital Europe and Erasmus+ applicants, this is not a stylistic exercise. Once submitted, the draft is fixed. Evaluators do not reconstruct the case the consortium intended to make. They score the case that is set out against the published award criteria, sub-criteria, thresholds and weighting for the relevant type of action.

Start with the evaluation form, not the narrative

Before reviewing logic, separate three questions that are too often conflated. Admissibility concerns formal conditions such as completeness, page limits and submission requirements. Eligibility concerns whether the applicant, consortium, activity and costs meet the call conditions. Award criteria concern quality and are where Excellence, Impact and Quality and Efficiency of Implementation are marked.

A logic audit is primarily an award-criteria exercise. It should use the evaluation form published for the specific call and type of action, rather than a remembered template from an earlier proposal. Horizon Europe Research and Innovation Actions, Innovation Actions and Coordination and Support Actions can apply different emphasis and sub-criteria. Digital Europe and Erasmus+ instruments have their own forms and scoring rules. Do not assume that a familiar section heading means the evaluator will assess it in the same way.

Set up the audit around the exact criterion text, score bands and thresholds. Where the form applies half-point scoring, do not review only for a binary pass or fail. Ask what distinguishes a credible 3.5 from a 4.5. A proposal may clear an individual threshold yet remain uncompetitive once weighting and ranking are applied.

How to audit proposal logic criterion by criterion

Read the document in the order an evaluator will assess it, not in the order it was written. The task is to follow each claim backwards to its evidence and forwards to its consequence. If either direction fails, the argument is incomplete.

1. Test the problem-to-objective chain

Start in Excellence. Each objective should arise from a defined problem, gap or opportunity that matters to the call. The draft must establish more than that a challenge exists. It must show why existing practice, research or provision is insufficient, why the proposed approach is appropriate, and why the action is necessary now.

A frequent weakness appears where a proposal moves directly from a broad policy problem to a technical solution. For example, citing a general skills shortage does not itself justify a particular training platform, pilot model or data architecture. An evaluator needs to see the intervening reasoning: the affected groups, the specific bottleneck, the evidence for it, and the mechanism by which the intervention addresses it.

Check every objective for precision. “Improve uptake”, “support interoperability” and “strengthen capacity” can be legitimate ambitions, but they are not auditable objectives unless the proposal identifies a baseline, target group, intended change and means of verification. If an objective cannot be expressed in a form that a later KPI can test, it is unlikely to anchor the work plan coherently.

2. Test methodology against the claimed advance

Methodology should be proportionate to the promise made in Excellence. This is where optimistic internal assumptions most often become visible. A consortium may claim a novel approach but describe familiar methods. It may propose a demonstration while implying market validation. It may present stakeholder engagement as co-creation without specifying who holds decision rights or how input changes the design.

Audit the chain from method to result. For each major activity, ask what it produces, who validates it, and what decision follows. If Work Package 2 collects requirements, where is the method for prioritising them? If pilots generate user data, how will that data alter the solution? If the proposal claims replication, what conditions must be present for another region or organisation to adopt the model?

The red-team question is simple: what passage in the proposal supports this claim? Statements about maturity, scalability, access to users, prior results, standards alignment or competitive advantage require evidence in the text. A confident adjective is not evidence, and evaluators are not required to infer it from partner reputation.

3. Trace outputs through to impact

Impact sections often contain the most polished language and the weakest causal chain. An output is what the project delivers: a prototype, curriculum, toolkit, standard, dataset, training course or policy recommendation. An outcome is a change in behaviour, capability, adoption or performance. Impact is the wider, longer-term effect to which those outcomes contribute.

A logic audit must show how each step occurs. Who uses the output? Why would they use it? What resources, approvals, procurement decisions, regulatory changes or follow-on investment are required? Which of those conditions are within the project’s control, and which are external dependencies?

Where the pathway depends on a third party, name the dependency and the mitigation. It is better to acknowledge that uptake relies on a national authority, platform owner or procurer than to imply that dissemination will cause adoption automatically. Dissemination creates visibility. Exploitation or uptake requires a user, a route, an incentive and usually an owner after the grant ends.

KPIs should sit at the right point in that pathway. Counting workshops, publications and website visits may demonstrate delivery, but it rarely proves the expected impact. Pair activity indicators with measures of reach, adoption, behavioural change or operational value where the call requires them. Be realistic about attribution. A three-year action cannot credibly claim to have caused a Europe-wide outcome that depends on policy decisions beyond its lifetime.

4. Reconcile implementation with the promised results

Now audit whether Implementation can carry what Excellence and Impact promise. Every objective and material impact pathway should be represented in work packages, tasks, deliverables, milestones, resources and governance. This does not mean repeating the narrative. It means making the operational route visible.

Look for orphan claims. These are commitments stated in the impact narrative that have no task, budget, owner or decision point in the implementation section. The reverse also matters: a costly task with no evident contribution to an objective can make the plan appear unfocused.

Milestones should record decisions or verifiable states, not merely calendar events. “Pilot completed” is weak unless completion has acceptance criteria. A stronger milestone identifies what must be true, who verifies it and what happens if it is not achieved. Risks deserve the same treatment. A risk register that lists generic risks but does not alter task sequencing, governance or contingency resources is not a management mechanism.

5. Audit consortium capacity as evidence, not biography

Evaluators assess whether the consortium has the capacity and role distribution to deliver the work. Partner descriptions commonly overstate credentials while leaving the actual delivery logic unclear. A strong publication record does not establish access to end users. A major company name does not establish exploitation responsibility. A public body’s letter of support does not establish authority to implement a result.

For each critical task, identify the accountable beneficiary, the relevant capability, the resources assigned and any external dependency. Check that roles match the claimed contribution. If a partner is essential to pilot recruitment, data access, standardisation or market uptake, that relationship should be operationalised in the work plan rather than left in a profile paragraph.

Use disagreement as a diagnostic

The most useful audit is not a consensus reading by people who helped write the text. It is a controlled comparison of independent readings against the same form. If one reader sees a clear impact pathway and another sees an unsupported assertion, the difference is evidence. It points to language, structure or missing proof that a real panel could also read differently.

Record disagreements by criterion and passage, then adjudicate them against the evaluation form. Do not average them away. A single low reading may expose a significant weakness that a mean score conceals, particularly where an individual threshold applies.

This is the practical value of an external pre-submission assessment. BidShark applies independent criterion-based readings, shows the supporting passages and preserves material disagreement before a final adjudicated score. It is an automated assessment, not a human panel verdict; the separate Expert Q&A package is where a serving or former EU evaluator answers written questions after reviewing the report.

Revise the chain, not just the wording

Once a weakness is found, resist the instinct to add more prose everywhere. Repair the specific broken connection. Add the missing baseline, allocate the absent task owner, convert a generic milestone into an acceptance test, or qualify an impact claim that the project cannot substantiate. Then check that the change is reflected consistently across objectives, work packages, KPIs, risks, budget and partner roles.

The final question before submission is not whether the proposal sounds persuasive to the people who built it. It is whether a sceptical evaluator, with limited time and no access to the consortium’s unstated knowledge, can trace every material promise to evidence, delivery and a credible result. Find the gaps while they can still be corrected.

About 15 minutes · before the panel

Evaluate
before you
submit

One proposal. One deadline. One opportunity to improve it before submission. Upload your proposal and the call, and get the score, the reasons and what to fix.

Start
Chat with us on WhatsApp