Keeping a supervision record that survives a handover

Every research group eventually runs the handover test. A supervisor takes a sabbatical, moves institution, falls ill, or retires, and someone else picks up their scholars. How badly that goes is decided long before it happens, by what was written down during four years of ordinary supervision.

Most labs fail this test. Not because nobody kept records, but because the records that exist were written as personal aides-mémoire — fragments meaningful only to the person who wrote them, scattered across a notebook, an inbox, and the margins of printed drafts. The scholar, who has the most to lose, typically holds the most complete picture and the least authority to assert it.

What the incoming supervisor actually needs

The temptation is to record everything. That produces an archive nobody reads. A handover needs four specific things, and a record that captures only these will beat an exhaustive one that buries them.

The current state of the work. Not the topic — the state. Which questions are settled, which are open, what has been tried and abandoned and why. The "and why" is what disappears fastest and costs most. Without it the new supervisor suggests the approach that was ruled out eighteen months ago, and the scholar has to relitigate a decision they thought was closed.

The decisions, with their reasoning. A list of what was decided is nearly useless without why. Decisions in research are provisional and get revisited when circumstances change; you cannot tell whether circumstances have changed if you do not know what the decision rested on.

The trajectory. Is this scholar ahead, behind, or on track — and against what? A single snapshot cannot answer this. A run of dated records can.

The obligations. Funding conditions, ethics approvals and their expiry, committee dates, teaching commitments, visa constraints. This is the category most often held entirely in the departing supervisor's head, and the one with hard external deadlines attached.

Write for a stranger

The practical test for any supervision note: would this make sense to a colleague who has never met this scholar?

"Discussed the sampling issue, agreed on the approach" fails. In the moment it is perfectly clear. Two years on it is clear to nobody, including the person who wrote it.

"Sampling: the cluster approach undercounts rural sites. Agreed to switch to stratified sampling with the 2021 census as the frame, accepting we lose comparability with the pilot data. Revisit if the census frame turns out to be unusable" — that survives. It is three sentences. It took ninety seconds.

The difference is not length. It is that the second version records the reasoning and the trade-off accepted, not just the outcome.

Records the scholar can see

A supervision record the scholar cannot read has a specific failure mode: it is never corrected.

Supervisors misremember. You will occasionally record a decision the other way round, attribute a finding to the wrong experiment, or note that something was agreed when the scholar left believing it was still open. If they can see the record, these get fixed within days, usually with a mild correction that costs nothing. If they cannot, the error sits there and becomes the official version — and it is the version the incoming supervisor will inherit.

There is a fairness dimension too. Supervision records feed into progression decisions, references, and occasionally disputes. A scholar assessed partly on a record they have never seen and cannot contest is in a poor position, and institutions increasingly take a dim view of it.

Institutional memory is not an archive

The reflex when handover comes up is to demand an annual report. Most annual reports are written the week they are due, summarise a year nobody remembers accurately, and are read once by someone ticking a box.

What actually transfers is the incremental record — the meeting notes, decisions and actions accumulated as the work happened. The annual report is a compression of that record. If the underlying record exists, the report takes twenty minutes and is accurate. If it does not, the report is a reconstruction, and reconstructions are systematically optimistic. Nobody writes "this year went badly and I am not sure why" in a form their head of department will read.

If you are choosing where to spend administrative effort, spend it on making the incremental record easy and habitual, not on making the annual summary more thorough.

What to do if you are inheriting a scholar right now

If the record does not exist, do not try to reconstruct it from the departing supervisor. Reconstruct it with the scholar, and do it in writing, in the first fortnight.

Sit down and build the four categories above together. The scholar knows most of it — they were at every meeting. What they typically lack is the reasoning behind decisions that were made for reasons never explained to them, and permission to treat their own account as authoritative.

Then give them the document and ask them to correct it. Their corrections are the most valuable part, and it establishes from the first week that the record is something you both maintain rather than something being kept about them.

That document becomes the baseline, and from there the ordinary discipline applies: decisions with reasoning, actions with dates, written where both of you can see them. It is not much to ask of any single meeting. It is the difference between a lab that can survive a supervisor leaving and one that cannot.