Your performance review started 11 months ago

It's Sunday night. The self-eval form has been open for forty minutes. You've written your name. You remember a launch in May, a difficult quarter that resolved somehow, a Slack message from your manager that you wish you'd screenshotted. The cursor blinks in the box labeled 'Notable contributions.' You consider writing 'demonstrated leadership' and feel something die. This is most people's relationship with performance review prep — the form arrives, the dread arrives with it, and the writing happens against a memory that has quietly compressed nine months into a blur and the last six weeks into the entire year.

The reason May is loud and January is silent

Memory isn't a filing cabinet. It's a compression algorithm, and the compression isn't uniform. Recent events stay vivid. Older events shrink into adjectives. By November, 'the migration I led in February' has become 'a project I worked on early in the year' — which is exactly the kind of sentence that produces a self-review with no nouns in it.

This is recency bias, and it isn't a discipline problem. You can't out-discipline how human memory works. Asking yourself, in October, 'what did I do in February?' is a question your brain is structurally unable to answer well. The answer was there once. It got compressed. It's gone.

The structural failure of performance review prep is that the cycle asks you to evaluate twelve months from inside a two-week memory window. The form expects a narrative. Your brain offers fog. The fix isn't to remember harder — it is to capture earlier, back when the events are still events and not adjectives.

What 'starting early' actually costs

When people hear 'start prepping for your review earlier,' they imagine a discipline they don't have time for: a daily journal, a weekly retro, a Notion database with relations and tags. This is the wrong picture. It is also why most people don't do it.

Starting early costs two minutes a week. One sentence, somewhere, on Friday afternoon. Two minutes times fifty-two weeks is one hundred and four minutes a year — less time than you will spend reformatting the self-eval doc the night it is due.

The cost of not doing it is the cost you already know. You give up eleven months of evidence in exchange for an hour of recovered Sundays, and then you spend that hour back, plus interest, in the week of the review. The trade is bad. You already know the trade is bad. The reason you don't make the better one is that 'journaling for your career' sounds like a project, and the actual thing is a sentence.

The four things worth capturing

When you do sit down for two minutes, capture these. Skip everything else.

• Wins — what shipped, what closed, what got measurably better because you were in the room. Not 'worked on the redesign' but 'shipped the redesign on March 14.' The verb matters. The date matters.
• Metrics — the number, not the adjective. 'Cut latency 40%' beats 'improved performance.' If you don't have a number, write down the rough one. You can verify later. You can't verify a number you never wrote down.
• Feedback, verbatim — screenshot the Slack message from your manager. Save the email. Copy the exact words your customer used. The phrasing is what makes the quote land in a self-review six months later.
• Stretch moments — the thing you didn't know how to do until you did it. Easy to miss in the moment because they feel like surviving, not winning. Six months later, in the section labeled growth, these are the only entries actually about growth.

One note on what not to capture: don't write the narrative yet. You're not drafting your review in March. You are collecting raw material. Save the storytelling for the form.

If your review is in two weeks: salvage mode

If you're reading this with two weeks on the clock, the long-game advice isn't useful yet. Here is the recovery plan.

• Slack search — search your own DMs for 'thanks,' 'great work,' 'appreciate,' 'well done.' That is your feedback file, assembled in ten minutes. Screenshot what you find.
• Sent mail — filter sent mail to your manager and skim the subject lines. You will remember the projects you forgot. Add the dates to your evidence sheet.
• Calendar walkback — open your calendar to January. Scroll forward one week at a time. Each recurring meeting that ended marks a project that shipped. Each one-off was probably a decision.
• Tickets and PRs — filter Linear, Jira, or your git log to your name, sort by date. The work is documented. You just have to look.

Build a one-page evidence sheet from these sources before you touch the self-eval form. Then write the form from the sheet, not from memory. The sheet is your external memory — use it.

If your review is in six months: the rhythm

If you have time, build the rhythm now. The version that works for most people is this: Friday afternoon, two minutes, three prompts.

• What did I ship this week?
• What number moved because of me?
• What did someone say about my work?

One sentence per prompt. Skip any prompt that didn't happen. Some weeks all three are blank — that is information too.

The point of the rhythm isn't the writing. The point is the capture. Future-you is going to be tired and panicked and writing under deadline. Present-you is calm and remembers what happened this week. Let calm-you do the work for tired-you. It is the cheapest gift you can give your future self.

You don't need a system to start. You need one sentence. Open whatever you'll see again — Notes, a doc, the back of an envelope, an app — and write down one thing that happened at work this week. Even if it's small. Even if it's 'I helped Alex unblock the API thing.' Tomorrow's entry doesn't matter yet. Next week's doesn't either. This week's entry is the experiment. The goal is to find out that two minutes is, in fact, two minutes, and that the result is a sentence you'll be glad to have in eleven months. That's the whole practice — everything else is decoration.