How to Find Research Collaborators

Every productive search begins with a written sentence describing what the project cannot do with the people already on it. Without that sentence you end up with a list of impressive researchers and no reason for any of them to reply.

Name the missing capability precisely

a method you cannot run, an instrument you do not have, a population or cohort you cannot access, a modelling skill, a language, a country you need for eligibility.

Decide whether you need a person or a service. A core facility, a paid analysis or a data access request solves some gaps faster than a collaboration does, and without the authorship negotiation.

Write down what you bring. A collaborator is agreeing to spend their own time, and the answer to "what is in this for them" has to exist before you send anything.

Decide the size of the first ask

a question by email, a call, a small joint analysis, a shared student, a full grant. Almost every failed approach asks for the largest one first.

The single most reliable source of collaborators is the literature you are already reading, used systematically rather than passively. The goal is to find the people solving your specific problem in a neighbouring field, since those are the ones whose skills complement yours rather than duplicate them.

Take the paper whose method you need and list everyone who has cited it in the last three years. That forward-citation list is a directory of people currently using the technique.

Read methods sections rather than abstracts. Abstracts tell you what a group studies; methods tell you what they can actually do, which is what you are shopping for.

Follow the datasets. Search for other papers using the same public dataset, cohort or instrument, since those authors already have the access and the tooling you need.

Check preprint servers as well as journals, because they show what a group is doing this year rather than what it finished two years ago.

Look one field over, deliberately. The person who can solve your statistical problem is often not in your field at all, and the fastest route to them is the methods paper you keep citing.

Our collaborator finder automates the first step of this from a document: upload a paper or a project write-up and it shortlists authors from the public literature whose work sits close to it, with the overlap stated. It suggests names to read, it does not check whether anyone is available or interested, and it never contacts anyone.

Conferences work for this, but not by chance and not at the plenary. What produces collaborations is small rooms, specific questions and a follow-up within a week.

Prefer small, focused meetings and satellite workshops over large annual conferences. Twenty people in a room for two days produces more contacts than two thousand in a convention centre.

Read the programme in advance and pick five people to speak to. Ambient networking produces business cards; a prepared question produces a conversation.

Work the poster sessions, including the posters of doctoral students. The student is often the person who actually ran the method you want, and their supervisor is easier to approach afterwards with a specific reference.

Talk to session chairs and organisers, who know who else is working on the problem and are usually willing to say so.

Follow up within a week, referring to the specific thing you discussed. A message sent a month later reads like a cold email with a false claim of acquaintance. The outreach templates include a conference variant.

Summer schools and methods workshops deserve a separate mention: they put you in a room with people at your own career stage who will still be around in fifteen years.

Some of the most productive collaborations start from a shared resource rather than a shared interest, because the resource creates a reason to talk and a structure to talk inside.

Core facilities and national research infrastructures

the staff who run an instrument know everyone who uses it and are frequently willing to make an introduction.

Cohorts, biobanks and registries with an access committee

applying for access puts you in contact with the custodians and often with other approved users working on adjacent questions.

Public datasets and their user communities, including the mailing lists, forums and issue trackers of the software built around them.

Multi-site networks and distributed replication projects, which exist specifically to recruit contributing sites and are one of the few places where an unsolicited offer to join is expected.

Funded networking instruments rather than research grants

many agencies fund workshops, exchanges and networks whose entire purpose is to introduce groups to each other before a project exists.

Open-source scientific software

contributing a fix or a feature to a tool your field uses is a slow route, but the people it introduces you to are the ones who build the methods.

The most underused source of collaborators is two floors away. Institutional structures exist to make these introductions and are usually idle because nobody asks.

The research office or grants office, which knows who else at the institution has applied to the same call and who has capabilities in a given area.

Internal seed funding for interdisciplinary pilots, which is often undersubscribed and exists precisely to pay for a first joint piece of work.

Doctoral schools and graduate programmes, whose co-supervision arrangements are a low-risk way to start a collaboration with a shared student.

Institutional seminars outside your department, where an hour of your time buys you a map of who does what nearby.

Technology transfer and industry liaison offices when a company partner is what the project needs, since they also handle the contracts that follow.

Bibliographic tools are good at one thing: showing you who has published on a topic and with whom. Every other question you have about a potential collaborator is outside their reach, and treating a shortlist as an answer is how people end up writing to someone who retired in 2023.

What they do well

topic and method overlap, co-authorship structure, recency of output, affiliation history, and finding people who describe the same problem in different vocabulary.

What they cannot know

whether someone has time, whether their group is funded, whether they are interested in your question, whether their institution is eligible for your call, and whether they are pleasant to work with.

Always verify the basics before writing

current affiliation from the departmental page, an institutional email address from a recent corresponding-author line, and evidence of activity in the last two years.

Check the candidate against your own co-author list. If the point of the search is to widen the circle, the names you already publish with are the wrong answer even when they rank highest.

Remember what a new co-author costs elsewhere

once you publish together, that person becomes a conflict of interest for reviewing your work, under the windows described in the conflict of interest guide.