Treating the review as a process with rules and a timeline, not a verdict delivered to you.
A performance review is a process with a shape most people never learn: a self-assessment, a manager's write-up, usually a calibration discussion among managers where relative ratings get settled, and only then the conversation you actually attend. Knowing that shape changes what you do. It means your self-assessment is partly written for people who have never seen your work, that the decisions are largely made before the meeting, and that the meeting itself is better used for finding out what would change next time than for contesting what happened this time. This is a different skill from receiving feedback well. It's mostly about evidence and timing.
Writes the self-assessment the evening before, from memory, and produces an account of a year that is in fact an account of the last six weeks with two unusually memorable incidents from earlier. The strong work in March is gone entirely. Has kept no record all year on the reasonable-sounding grounds that they would obviously remember the important things, and has now discovered empirically that this is not how memory works.
Writes a self-assessment for their manager, who already knows all of it, full of shorthand and internal references that make sense to exactly one reader. The document then travels to a room of managers who have never seen the work and are comparing people across teams, where it reads as thin. Is rated accurately against what the room could tell, which was considerably less than there was, and receives feedback about visibility that lands as unfair and is in fact a precise description of what happened.
Mentions wanting a promotion for the first time in the review itself, having assumed the work would make the intention obvious. The cycle's decisions were effectively settled some weeks earlier, in a conversation where their manager, not knowing there was an ask, advocated for something else. Receives a supportive response and a plan for next cycle, which is the earliest thing that was ever actually available, and will do the same again next year, having drawn no conclusion from it.
Receives the rating as a judgment handed down, thanks their manager, and leaves without asking the two questions the hour was for: what specifically would have made it the next level up, and what evidence would need to exist by the next cycle. Has treated an inherently negotiable and partly noisy process as a fact about themselves, and will spend the following year working hard against a target nobody has described.
Your review is in three weeks and you have no record of the year. What would you do — reconstruct it from memory and calendar, or do that and then start a running note for next cycle, since the same thing will happen again?
You want a promotion this cycle. What would you do — let the work speak and see how the review goes, or tell your manager months in advance so they can advocate with notice?
You get a rating one level below what you expected. What would you do — make the case that it's wrong, or ask what specifically would have made the difference and what evidence would need to exist next time?