You don't forget the work. You forget the evidence.

The blank page the night before

The self-review is due tomorrow. The box is empty. And the strange part isn't that you have nothing to say — it's that you know you had a good six months, and you can't retrieve any of it.

You scroll back through Slack. You open old PRs. You find a doc you wrote in March and think, oh right, that. An hour in, you've reconstructed maybe a third of what you actually did, and the version you'll submit is the version you happened to remember at 10pm on a Sunday — not the version that was true.

We've been on that blank page more than once. It's the moment that eventually made us build a tool — but the tool isn't the point of this post. The reframe is, because for a long time we had the problem diagnosed wrong.

It feels like a competence problem. It isn't.

The night-before scramble feels like evidence that you didn't do enough. You stare at the empty box and the quiet voice says: if the work were really that good, you'd remember it.

That voice is wrong, and it's wrong in a specific, fixable way. You're not failing to do impactful work. You're failing to keep it. Those are different failures, and only one of them is about how good you are at your job.

Here's what we mean, concretely — the same handful of things go missing every cycle.

Your impact gets scattered. A win doesn't live in one place; it's spread across a ticket, a thread, a design doc, a meeting nobody wrote down, and your own head. None of those add up on their own, and by review time you're the only system that ever held the whole picture — and you've garbage-collected most of it.

You forget the small wins. The quiet fix that saved the on-call rotation a bad week. The gnarly bug you chased for two days. The thing that felt minor when you shipped it, and would have read as real, cumulative judgment if you'd written it down.

And you lose the invisible work entirely — mentoring a teammate through their first big feature, unblocking another team so their launch didn't slip, the process change everyone quietly adopted. This is often the most valuable work you do, and it shows up in exactly zero tickets. It's the first thing to decay and the last thing you'll reconstruct under pressure, which matters because it's frequently the work that decides your level.

Memory is a terrible evidence base

The deeper issue is that we treat memory as if it were a record, and it isn't. It's lossy, it's recency-biased, and it degrades fastest on precisely the work that's hardest to point at later.

Reviews and promotions aren't decided on the work you did. They're decided on the work you can articulate when it counts. That's a little unfair — but it means the scramble isn't a writing problem you solve with a better vocabulary the night before. It's a retrieval problem. By the time you're staring at the box, the evidence is already gone, and no amount of wordsmithing brings it back.

Once you see it that way, the fix stops being "get better at self-promotion" (uncomfortable, vague) and becomes "keep a record" (boring, doable). That's a much better problem to have.

Keep the receipts while you still have them

The move is to separate the two jobs you're currently doing at the same time: remembering and composing. Right now you do both at review time, which is why it's miserable — you're running an archaeological dig and writing an essay in the same sitting.

So do the remembering continuously, cheaply, in the moment. When something goes well — you ship it, you unblock someone, you fix the thing — take thirty seconds and drop the messy version somewhere durable. Not a polished bullet. Just enough that future-you can reconstruct it: what happened, what you did, and any number you have while you still have it. Then, at review time, the writing is just editing. The evidence is already there.

You don't need a tool for this. A plain text file works. A note per week works. The whole trick is lowering the cost of capturing one thing to near zero, so you'll actually do it when the moment's fresh — instead of promising yourself you'll remember, which you won't.

The reason we eventually built something for it is that our own text file kept dying — which is its own post, and honestly the more interesting one. But you can get most of the value tonight, for free, by starting the file before you need it.

Your next review is closer than it feels. The work you'll want to point to is happening right now. The only question is whether you're writing it down while you still remember it — or planning to reconstruct it, again, the night before.