Closing a collaboration properly: the conversation nobody schedules

Grant close-out gets a financial report and an administrative sign-off. What it rarely gets is a conversation about who may do what with the data, the unpublished results, and the manuscript that never quite got finished.

That conversation costs an hour. Skipping it is why partners end up unable to use material they helped generate.

Settle six things

Data: who holds the originals, who may use them for what, for how long, and whether use requires the other party's consent.

Manuscripts in review: who takes them through, and whether authorship changes if that person does most of the remaining work.

Stalled drafts: whether anyone will finish them, by when, and what happens if nobody does.

Unpublished results: whether they will be written up, and whether a party stepping back objects to the others publishing.

Equipment, samples, code remaining in either location.

Residual obligations to the funder and to participants — reports, deletion of identifiable data, results feedback to the community. Who does each, and when.

The last item is the one most often dropped entirely, and it is the one involving people who cannot follow up: the participants.

A fair rule for stalled manuscripts

This is the most common source of post-collaboration conflict, and a simple rule prevents most of it.

Set a deadline — six months is reasonable. Within it, the lead author works to complete the paper with all co-authors. After it, any party may take the work forward, provided they notify the others and preserve proper credit for what was already done.

Write it down with the start date. It protects both sides: nobody can sit on a manuscript indefinitely, and nobody can quietly appropriate someone else's work.

Data rights: the asymmetry to watch

If the data was generated in Vietnam and the working copy sits on your institution's servers, the default outcome is that your partner has to ask you for access to material their team collected.

That is worth fixing explicitly at close-out rather than leaving to goodwill and email response times. Confirm in writing that they may use the dataset for their own studies and for their students' theses, and say whether notification or consent is needed.

If this was agreed at the start, close-out is just confirmation. If it was not, this is the last point at which the conversation happens while everyone is still responsive.

Residual obligations need a named owner and a budget line

Deleting identifiable data, producing a plain-language summary for the community, answering participant queries next year — none of these happen once the funding stops and everyone has moved on.

Two practical measures: name the person responsible for each, and where the work falls on your partner, make sure it was resourced in the original budget rather than assumed.

"Our partner will handle participant queries" is only true if somebody is paid to be there.

When it ends badly

If the collaboration is ending because of a dispute, the close-out matters more, not less.

Keep correspondence professional and factual, reference what was agreed, and use the institutional mechanisms rather than bilateral pressure. Where you hold the greater institutional power, be aware that an unresolved close-out usually means the other party gave up rather than that the matter was settled.

And separate the dispute from the obligations: participants and the funder are owed the same things regardless of how the parties feel about each other.

Keeping the relationship usable

A collaboration that closes cleanly is a collaboration you can restart. Three things that cost almost nothing:

A closing note naming each person's specific contribution — useful to them internally in ways you may not see.

Asking whether they need anything from you in the coming year: a letter, a data extract, a reference.

Sending them anything that publishes later from the shared work.

The research community in any field is smaller than it feels, and the people on the other side of this close-out appear later in review panels and funding committees.

The document

Not a legal instrument. An email summarising the six items, sent to everyone, with a request to confirm.

Send it within a month of the work winding down. Later than that and a request to formalise things reads as a demand rather than routine housekeeping.

Its value is not enforceability. It is that writing it forces everyone to be specific, and that you can open it two years later when someone asks.