Keep your own record of what you did, because nobody else is

Memory is worse at this than you expect

Ask someone in their third year what they achieved eighteen months ago and you will usually get two items, both recent, both partly wrong about the sequence and the scale. This isn’t carelessness. Working memory of a job is organised around what is happening now, and completed work stops being salient almost immediately once the next thing starts.

The result is a predictable failure at exactly the moments where evidence matters: a review, a promotion case, an interview, a conversation about pay. In each of those you’re asked to reconstruct a year or more, and you reconstruct it from whatever survives, which is dominated by the last two months and by whatever went wrong. Nobody else is keeping the record either. Your manager remembers a subset, weighted the same way, and their notes exist mainly where a process compelled them.

What is worth writing down

A useful entry is short and has four parts: what the situation was, what you actually did, what changed as a result, and who else was involved or noticed. The third part is the one people omit and the one every later use depends on. "Rewrote the onboarding notes" is a task. "Rewrote the onboarding notes because two new joiners had asked the same eight questions, after which the induction stopped consuming a day of a senior colleague’s time" is evidence.

Record things that aren’t achievements too. Problems you predicted, decisions you disagreed with, estimates you gave and what they turned out to be, feedback you received in passing. These have several uses: they calibrate your own judgement over time, they supply the specifics in a difficult conversation, and they are the raw material for noticing patterns you would otherwise experience as a vague feeling that something is wrong.

Note the names. A large part of what makes a case persuasive later is being able to say who benefited and who would confirm it, and those details evaporate faster than the work itself does.

The mechanics, kept deliberately cheap

Any system fails if it takes effort. A single document, appended to once a week or once a fortnight in ten minutes, is enough, and it beats an elaborate structure that gets abandoned in March. Attach it to something already in your calendar so it does not depend on remembering. The entries do not need to be well written; they need to be dated and specific.

Keep it somewhere personal rather than on a work system, subject to one important qualification: do not remove anything confidential from your employer. What belongs in a personal record is a description of what you did, not the customer data or the internal documents themselves. That distinction matters, both because copying material out can breach your contract and because a record full of things you should not have is useless to you anyway.

Where it changes the outcome

The obvious use is a review, where arriving with a dated list quietly changes the shape of the conversation. Instead of both parties reconstructing the year, there is a document, and the person who brought it has a large influence on what gets discussed. That is not gamesmanship; it is the same reason anyone takes minutes.

The less obvious uses are larger. A CV becomes an editing job rather than an act of recall. Interview preparation stops being an attempt to remember something that sounds impressive. A manager change costs you far less, because you can brief the new person on what you have been doing without depending on a handover that probably lasted twenty minutes. And in the unwelcome cases — a performance process, a restructure, a dispute — a contemporaneous dated record is the single most useful thing you can have, and it cannot be created retrospectively.

The honest limits

A record does not by itself make a case. Evidence still has to be selected, framed for the audience, and connected to what the organisation cares about, and a list of forty entries handed to a manager is worse than three chosen well. It also does not compensate for work that nobody values, and the exercise sometimes reveals exactly that, which is uncomfortable and useful.

There is a failure mode worth naming: keeping the record so carefully that it starts to influence which work you take, steering you towards tasks that document well over tasks that matter. That is a real risk in any environment with a strong promotion process. The record is meant to be an instrument for remembering, not a scoreboard, and the moment you catch yourself declining something because it will not write up neatly, the tool has begun using you.