How to Build a Research Consortium

A consortium is a set of legally independent organisations that apply together and sign one grant agreement. Every collaborative programme sets minimum conditions, and they are conditions on organisations and countries, not on people. Read them before the first email, because they decide whether a conversation is worth having at all.

Standard Horizon Europe research and innovation actions require at least three legal entities, independent of each other, established in three different eligible countries, at least one of them an EU Member State. The rule sits in the General Annexes of the work programme, published with the other binding documents on the Funding and Tenders Portal, and individual topics can tighten it.

Independence is a legal test, not a social one. Two institutes under the same parent organisation usually count as one entity, which is the trap that catches consortia built inside a single national system.

The minimum is a floor, not a target. A three-partner consortium is normal for a coordination action and unusual for a large collaborative project, and the call text often states an expected size or an indicative budget per project that implies one.

Some topics add conditions that are easy to miss on a first reading: a partner from a widening country, a clinical site, an industrial participant, a specific type of organisation, or a maximum number of partners.

Other schemes work differently. A single-principal-investigator scheme such as an ERC grant has no consortium at all, and adding partners to it misreads the scheme rather than strengthening the proposal.

Partner is a loose word and the grant agreement is precise. Getting the role right at the start prevents the most common late crisis, which is discovering that the organisation you built a work package around cannot hold the status you assumed.

The coordinator is one organisation. It submits the proposal, signs first, receives the funding and distributes it, chairs the governance, and is the funder's single point of contact for reporting. It is a job with real administrative weight, and it is usually carried by a research office rather than by the researcher whose idea it was.

Beneficiaries are the other funded partners. Each signs the grant agreement, holds tasks and a budget, reports its own costs, and is liable for its own part of the work.

Associated or unfunded partners contribute without receiving money from the grant. They are useful for organisations that cannot or will not take funding, such as a company that would rather stay outside the reporting regime, but they cannot be given accountability for a deliverable the project depends on.

Affiliated entities, linked third parties and subcontractors each have their own cost eligibility rules. Subcontracting core project tasks is normally restricted, so a subcontractor is not a way to add missing expertise.

Being coordinator is not a prize. Take it if the idea is yours and your institution has the administrative capacity. If your research office is stretched, negotiate a work package leadership or a scientific coordination role instead, which carries the influence without the payment administration.

A consortium that reads well on paper has one property: a reviewer can point at any partner and say what it does that nobody else in the group does. That property comes from designing the work breakdown before drawing up the list of names.

Write the work packages first, with an objective, an owner and a deliverable each. Then ask which competence each package needs. The gaps in that list are your partner specification, and they are what you should be searching against.

Give every partner one visible job. A partner whose only contribution is dissemination or a letter of support is what reviewers call a passenger, and it costs points under implementation criteria in every framework that has them.

Balance disciplines against the work, not against a quota. If the project has a modelling package, a field package and a policy package, the consortium needs a modelling group, a field group and a policy group, and stapling a fourth discipline on because it sounds interdisciplinary weakens the story.

Balance countries against eligibility and access, not against geography for its own sake. A partner is there because it holds a cohort, an instrument, a licence, site access or a language. Spreading partners over eight countries for its own sake is expensive to manage and easy to criticise.

Consider the sector mix the call implies. Topics that ask for uptake beyond academia expect an organisation that will actually use the result, and a letter of interest from a company is not the same thing as that company holding a work package.

Keep the group as small as the work allows. Every added partner adds coordination cost, a share of the budget that has to be justified, and one more organisation whose internal deadlines you now depend on.

The first two partners come from your address book and the third from a partner's address book. After that the network runs out, and the search has to become deliberate. Four routes cover almost every case.

Databases of already funded projects tell you who has done this kind of work and survived the reporting. CORDIS lists EU funded projects and their participants, NIH RePORTER covers NIH awards, and Gateway to Research covers UKRI. Searching them by topic returns organisations with proven capacity rather than only people with publications.

National Contact Points exist for exactly this and are free. Most EU programmes maintain a network of them per country, and part of their job is partner search, including circulating a partner profile you write. Your own research office usually knows the local one by name.

Networking instruments are the slow route that works. COST Actions and similar networks put you in a room with the groups working on your problem across a continent, one or two years before you need them.

The literature is the route that finds groups nobody in the consortium knows. Recent papers on the specific missing competence, rather than general standing in the field, are what make the fit legible when a reviewer reads the partner descriptions.

Our research partner finder belongs to that last route and nothing more. It reads a project description and suggests researchers whose published work sits close to it, with a reason for each name. It knows nothing about funders, calls, country eligibility, organisation type or capacity, and it contacts nobody. It shortens the reading, and every eligibility question stays yours and your research office's.

The gap between a researcher saying yes and their organisation being able to participate is where late proposals collapse. Everything below is quicker to check in week one than to repair in the final week.

Confirm the organisation, not the person. Ask which legal entity would sign, and whether it is already registered with the funder. In EU programmes that means a participant identification code, which takes time to obtain if it does not exist yet.

Confirm eligibility against the call, per partner

country, organisation type, and any role restriction. A researcher rarely knows their own institution's status, and the research office answers this in a day.

Confirm capacity honestly. Ask what else the group is submitting to the same deadline, and whether the named person would still be there at the project's start, which for large programmes is a year or more after submission.

Ask for their internal deadline, not the funder's. Institutional sign-off, financial approval and signature of a partner declaration commonly close one to three weeks before the call, and each partner has a different one.

Get the administrative contact early, in writing, and copy them on anything with a deadline. Proposals are lost to an unsigned form far more often than to a weak idea.

Agree that a consortium agreement will be needed and when it will be drafted. Most collaborative EU projects require one between the partners before the grant starts, and models such as DESCA exist precisely so nobody starts from a blank page. The conversation that has to happen alongside it is in our collaboration discussion checklist.

Evaluation reports repeat themselves. The recurring criticisms of consortium composition are structural, and all of them are visible before submission if someone reads the proposal as a stranger would.

The passenger. A partner with no distinct deliverable, present because they asked or because someone owed them a favour. It is the single most common composition criticism, and removing them strengthens the proposal.

The late partner. Adding an organisation three weeks before the deadline means their tasks are written by someone else, their budget is a guess and their sign-off is a gamble.

The imbalanced budget. A coordinator taking a disproportionate share, or a partner with a large budget and one small task, invites a specific and easily justified criticism.

The duplicated competence. Two partners who do the same thing look like padding unless the proposal explains why two are needed, for instance two clinical sites for recruitment volume.

The invisible early career researcher. If a postdoc will lead a task, name them and their role. Unattributed effort reads as a group with spare hands rather than as a plan.

The unmanaged risk. A consortium that depends on one partner for the data with no fallback will be told so, and naming the dependency and its mitigation beats pretending it is not there.