This might sound silly for those who know how to do it and do not see that it is a problem but for some people like me it is.

Since I started working in academia, I have had the problem of building collaborations, or to be more precise, of maintaining the collaboration and making it fruitful. I know how to take the initiative and start the collaboration but that's it. For example, if I meet someone at a conference who is working on something exciting for me, I usually communicate with them and even schedule a call but that is the end of this collaboration (can we even call it that??). In the call, I explain what we are doing and I hear what they are doing but do not know how to move to the next step or even what is the next step.

There are people/researchers who have a magic ability to build collaborations and most importantly accomplish something out of them. Any advice?

  • 11
    SF author Larry Niven said (at a panel at a convention I attended) that a collaboration is a case where two authors each do 90% of the work.
    – Boba Fit
    Dec 1, 2022 at 13:37
  • 1
    In my career, I had collaborations (research and papers wrote together) with people I know or met informally, but I had these collaboration 5 or even 10 years after the first contact point and after many failures. Among my peers (PhDs when I was a Phd, Postdoc when I was postdoc), the only one that collaborated where the one bringing funds from one professor/advisor to the other, or similar agreements, nothing to do with the idillyac picture of researchers working cooperatively towards a goal.
    – EarlGrey
    Dec 1, 2022 at 13:52
  • 1
    In my career, I find out that only 10/20% (max) of these calls lead to a collaboration. It takes time, effort and luck. Dec 1, 2022 at 18:33
  • I agree with @AnderBiguri . From my experience, the people who are really good at collaborations are usually very talkative. They are also strategic with who they talk to, but they definitely do not miss an opportunity to talk about their research to others. These people also tend to have a large pool of unsolved problems, but not necessarily.
    – Yanko
    Dec 1, 2022 at 20:27
  • If after the first call there is not at least one reasonable next step that comes to mind, then that is a signal that there may simply not be a useful collaboration possible here at this point in time. No shame in this, as @AnderBiguri notes, not all first contacts lead to a collaboration. Perhaps better to leave it at this and stay in touch - you always meet twice, and perhaps in a year or two there will be another meeting where a next step does immediately occur to everyone. Dec 3, 2022 at 10:45

3 Answers 3


As you suspect, you haven't even started collaborating.

The next step is to figure out what you want to work on together (and whether you do indeed want to collaborate). I see three different types of choices for a project:

  1. One of you has a nail, and the other one a hammer.

This could be the combination of one person having identified an interesting question, and the other one being an expert of a technique that might help solve it; or one person having a data set (or the means to create it) and the other person the know-how for how to analyze it, etc.

  1. By our powers combined!

As you talk, you realize that by combining your speciality methods / rare equipment / other know-how the two of you together might be able to accomplish something new. Maybe the contribution is the sheer novelty, the creativity in coming up with it. Maybe you tackle a challenge that neither of you alone would have even dreamed of.

  1. Hanging out is just fun!

Sometimes, what you do together can be something either of you could have done alone - but you work together, because that's more fun (and probably somewhat higher quality in the end, too). This obviously relies on a strong drive to collaborate, and may not be the motive you should aim for.

  • 3
    That's a great list.
    – xLeitix
    Dec 1, 2022 at 15:48

I don't have any evidence for any of these observations, but I think most of the keys to a successful collaboration need to be in place before it starts.

1. Make sure you (and they) have the time and energy to commit.

The fastest way for a collaboration to fail is for one (or all) of the collaborators to be overextended. If you barely have time to do your own research projects, how will you be able to be a fruitful collaborator? I regularly politely decline collaborations where the time ask is too great. Likewise, it might sound exciting to collaborate with a Nobel laureate, but if they don't have time to actually bring anything other than their name, is that what you want?

2. The best collaborations are when the different parties contribute meaningfully novel expertise or resources.

Almost all of my successful collaborations have been a match between two or more parties that have expertise or resources that the other parties just don't have. The access to something collaborators can't easily replace motivates them to pull their weight in the project.
- They are experts in a particular disease, you know how to use machine learning.
- You have interesting genome sequencing data, they know how to do de novo assembly.
- You have 50 participants with a rare condition, they have 10 more participants with that rare condition.

3. Make sure all parties are excited about the project.

I tend to avoid collaborations where the parties are not excited about the topic. I find that collaborators that just want to "pay the bills" so to speak struggle to deliver high quality products. The same goes for oneself. Is it worth it to work on a project that doesn't excite you just to get another publication?

4. Clearly agree on authorship and funding early.

This is key to avoiding drama later. Obviously, as the project goes on, things can change, but it's vital to have a baseline that everyone expects. One of the comments suggests that collaborations are frequently only productive when there is transfer of funds. That has not been my personal experience, although I admit that I have been privileged to conduct research in high resource settings.

5. Set realistic, measurable, and time-bound deliverables for each member.

The best collaborations, in my opinion, are when clearly defined deliverables are agreed upon at every meeting and the team members complete them or are agree to be accountable for why they can't be delivered. At the end of each meeting, we "go around the (often virtual) room" and agree to our "homework".
- By next meeting, Ian will write a draft for the introduction to the manuscript
- By next meeting, Nasir will commit the code for training the deep learning model to the Github repository
- By next meeting, Linda will write aim 3 for the R01 submission

Also, it's important to schedule the next meeting before concluding the current meeting. For bonus points, the meeting could automatically recur.

6. Don't be afraid to end a collaboration if it's not working.

At some level, collaboration is like dating. They don't all work out. If you or your collaborators aren't able to meet or aren't able to make progress on the deliverables, I think it is reasonable and necessary to end them. I've started collaborations where others weren't able to meet when they agreed multiple times, and that just had to be the end. Sometimes collaborations can just drift off into the night, but sometimes you need to just say "I'm very sorry, but I don't think I can continue this collaboration".

  • 2
    I worry a bit about your #5. "Dividing up" the project is probably sub-optimal, just as it is in the creation of interesting software. Try to replace "you do this and I'll do that" with "we do this together". Hard, I know.
    – Buffy
    Dec 1, 2022 at 15:53
  • 3
    @Buffy I think it's interesting to see how people from different fields can disagree on things like this. To me, #5 is almost certainly the most important part of a collaboration. I can imagine that in theoretical fields, though, that such an approach might be counterproductive.
    – Ian
    Dec 1, 2022 at 16:12
  • I kind of agree that #4 and #5 are probably good advice, although I've hardly actually done this, and had lots of great collaborations.
    – Arno
    Dec 2, 2022 at 9:18
  • (+1) For both the scheduling (which seems relatable in my experimental field) and mentioning funding explicitly. When both parties are being paid to do something else entirely and have no prior collaboration history, joint projects can become notably hard to maintain.
    – Lodinn
    Dec 3, 2022 at 0:30

It is harder to join someone else's project than to invite someone to join yours. Another idea is that you need to have more than "interest" in someone's work. You need to bring something of value to the conversation.

The conversations that lead to collaborations can be either local, at your university or in a small region with several universities, or at conferences where people with similar interests gather. But you have to offer ideas.

One way to make collaborations continue once started is to gather some funds to invite people to your place (or for yourself to travel elsewhere) to give talks on a topic of mutual interest and use the co-location to explore ideas.

A phone conversation or short zoom session isn't going to do it. You need to find ways to explore what each can contribute. That takes a bit of time. And once started, find ways to keep it going. This may also require some funds. At some point grants might be possible to fund co-located activities leading to collaboration, but it would probably be hard to do that as initial steps.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .