A consortium can spend six months developing a credible technical solution, assembling work packages and negotiating partner roles, then lose points because a requirement was treated as background reading rather than a response obligation. You only get one chance to submit. Knowing how to map call requirements before drafting is therefore not an administrative exercise: it is the first control against an evaluation summary report that says the proposal is good, but insufficiently demonstrated.
The call text is not a single set of instructions. It contains conditions for admissibility and eligibility, the expected scope and outcomes, conditions of participation, the award criteria, and sometimes call-specific instructions that alter how a standard criterion should be read. A useful map separates these layers. If it does not, teams tend to give equal attention to everything, including material that does not affect the score, while leaving scored claims unsupported.
Start with the documents that govern the score
For Horizon Europe and Digital Europe, begin with the call topic page and the applicable work programme, then identify the proposal template and evaluation form for the action type. For Erasmus+, the programme guide, call text and relevant application form perform similar functions. Do not assume that a familiar template or a previous call’s evaluator brief still applies. Versions, sub-criteria, thresholds and expected outcomes can change.
The practical distinction is this: admissibility and eligibility determine whether the proposal can proceed at all; award criteria determine the score once it is evaluated. Page limits, submission channels, mandatory annexes, legal entity conditions, country rules and consortium composition can be fatal even where the technical case is strong. They belong in the map, but not in the same column as Excellence or Impact evidence.
Create one controlled source file. Record the document title, version, publication date and the precise location of each requirement. A paraphrase without a source reference is not sufficient. When the team later disputes whether a sentence is mandatory, expected or merely contextual, the original wording must be available.
How to map call requirements into a proposal matrix
The most reliable format is a requirements matrix with one row per testable obligation. A paragraph of call text often contains several obligations and should be split accordingly. For example, a topic may require a demonstrator, contribution to specified policy objectives, engagement with named stakeholder groups, interoperability with existing infrastructure and credible exploitation. That is five rows, not one.
Each row needs a requirement ID, verbatim source text, source location, classification, evaluation criterion, planned proposal location, owner, evidence needed and status. The classification should distinguish at least: admissibility, eligibility, scope, expected outcome, destination impact, implementation instruction and award-criterion requirement.
The evidence field is where the matrix becomes useful rather than bureaucratic. Do not write “addressed in Section 1”. State what an evaluator will be able to verify: a quantified baseline, letters of commitment, a named pilot site, an exploitation route, a governance decision right, a partner capability, a data-management procedure or a measurable KPI. If no evidence can be named, the requirement has not yet been properly planned.
A compact row might read: “Demonstrate validation in operational conditions”, classified as expected outcome and linked to Excellence and Impact. Its proposal locations may be the methodology, pilot work package and impact pathway. Its evidence might be access agreements, a defined test protocol, readiness assumptions, pilot-user roles and outcome metrics. The owner is not “the consortium”. It is the person who must obtain the evidence before the internal review date.
Preserve the evaluator’s reading order
Map requirements against the criterion in which an evaluator is most likely to test them, but accept that many requirements carry across criteria. A pilot can support Excellence by validating the methodology, Impact by evidencing uptake, and Implementation by showing feasible access to facilities. Repeating the same claim word for word across all three sections is weak. The matrix should specify the different question each section must answer.
For Excellence, the questions are typically whether the objectives are clear, ambitious and credible, and whether the methodology is sound. For Impact, evaluators test whether the pathway from project results to outcomes is plausible, whether barriers are recognised, and whether dissemination, exploitation and communication are proportionate. For Quality and Efficiency of Implementation, they look for a coherent work plan, credible resources, risk management, decision-making and the capacity to deliver.
The exact formulation, weighting and sub-criteria depend on the action type and published evaluation form. Do not import scoring logic from a Research and Innovation Action into an Innovation Action, a Coordination and Support Action, Digital Europe, or Erasmus+ simply because the headings look familiar.
Translate each requirement into an evaluator test
Call language is often broad by design. “Contribute to European competitiveness” is not a drafting instruction. It becomes one only when translated into an evaluator test: what result will exist at project end, who will use it, what changes as a result, what prevents uptake, and what evidence makes the chain credible?
Use three tests for every scored requirement. First, is there an explicit response? An evaluator should not need to infer that the consortium intends to comply. Second, is there evidence? Assertions such as “strong market demand” or “leading expertise” attract scrutiny when no market evidence, customer route, prior result or relevant partner record accompanies them. Third, is the response consistent across the proposal? A target presented in Impact must be resourced in the work plan and owned by a named partner.
This is also where a red-team reading earns its place. It looks for unsupported causal jumps: a technical output treated as an adopted solution; a dissemination activity treated as impact; a broad partner list treated as delivery capacity; a risk register that names risks without defining triggers, mitigation owners or contingency consequences. These are not stylistic faults. They are reasons for an evaluator to withhold confidence.
Build compliance and persuasion separately
A proposal needs both. Compliance proves that every requirement has a home. Persuasion gives the evaluator a reason to believe the plan will work. Combining them too early creates a familiar failure mode: a polished narrative in which mandatory topic elements are hard to locate.
Draft the compliance layer first. For each matrix row, mark the exact section, table, work package, deliverable, milestone or annex where the response will appear. Then assess the persuasion layer: whether the response is specific, evidenced, internally consistent and proportionate to the claim.
This distinction matters especially for expected outcomes. Listing them in an opening paragraph may establish recognition, but it does not show contribution. A defensible response connects each outcome to activities, outputs, uptake actors, indicators and assumptions. If the route relies on a policymaker, procurer, regulator, investor or end user outside the consortium, say what will secure their participation and what happens if it does not.
Check thresholds, score bands and contradictions before freeze
A high overall score cannot rescue a criterion below its threshold where the evaluation rules impose one. Equally, a proposal that appears comfortably above threshold can still be uncompetitive when the available budget is limited and competing proposals are strong. Map requirements to protect against both failure modes: the obvious omission and the significant weakness that keeps a criterion in a lower score band.
Before submission, run a traceability check from both directions. Read each matrix row and find the exact proposal evidence. Then read each proposal section and ask which call requirement it discharges. Material that has no evaluative job may be consuming scarce page space; requirements with no identifiable evidence are unresolved.
Where scores are expressed on a 0-5 scale, do not treat a half-point difference as cosmetic. It can reflect the distinction between an adequate response and a response with a significant weakness. The internal question is not “have we mentioned this?” It is “would a sceptical evaluator find the claim substantiated on the page?”
An independent pre-submission assessment can test that question without inheriting the consortium’s assumptions. BidShark’s automated assessment uses the published marking scheme for the relevant form version and preserves disagreement where independent readings diverge; its Expert Q&A package is the point at which a human EU evaluator addresses written questions. Neither changes the call requirements. Both can expose whether the draft actually answers them.
Treat the completed matrix as a live decision record until submission, not a planning artefact filed away after the kick-off meeting. Every unresolved row tells the team exactly what it must obtain, decide, evidence or remove while revision is still possible.