A performance review is written for two readers at once

One document, two audiences

A written review looks like a message from your manager to you. It is also, and often primarily, an entry in a record that will be read by people you have never met: whoever calibrates ratings across teams, whoever approves pay and promotion, and possibly whoever inherits you when your manager moves on. Those two audiences want different things from the same paragraphs.

This is why reviews so often read oddly — formal in places, evasive in others, praising something you thought was minor. Your manager is writing to you and building a file at the same time, and where those purposes conflict the file usually wins, because the file has consequences that outlast the conversation.

Ratings are comparative, whatever the labels say

Most rating scales are described as absolute — meeting expectations, exceeding them — but in practice they are distributed. Managers submit proposed ratings, those ratings get compared across a group, and adjustments follow so that the overall spread looks sensible to whoever is approving pay. Some organisations enforce this openly, others do it by convention and would deny the label.

The consequence is that your rating carries information about your standing relative to peers, not only about your work. A strong year in a team of strong years can produce a middling rating. This is genuinely unfair at the individual level and it is also a predictable output of a system that has to allocate a finite pot, and knowing that is better than concluding that your manager thinks less of you than they said.

Reading the wording

Review language is conventionalised, and the conventions are learnable. Concrete, specific praise with named outcomes is real. Generic warmth about attitude, reliability or being a pleasure to work with, unattached to any achievement, is frequently filler used where there was nothing specific to say.

On the other side, phrases like "would benefit from", "an area for development" and "at times" are the standard softeners for a real criticism. The test is whether a point appears more than once, or reappears from the previous cycle. A development area named twice is not a suggestion; it is the thing that will be cited if a harder conversation ever happens, and it deserves a direct question about what would count as fixed.

It is also worth reading what is missing. A review that describes how you work but says nothing about the scope or difficulty of what you worked on is not building a case for anything, whatever its tone. If you are hoping to be promoted, the sentences that matter are the ones a stranger could use as evidence that you are already operating at the next level, and their absence is more informative than any adjective in the document.

What to do while you are in it

Most reviews invite a self-assessment, and most people write it badly, either by listing tasks completed or by being so modest that they supply nothing usable. What your manager needs is material they can lift: outcomes, dates, scope that changed, work that other people relied on. Write it as though it will be copied, because parts of it usually are.

If you disagree with something, the useful move is to establish the specific evidence rather than to argue about the characterisation. "Can you give me an example of where that happened?" is a fair question, and if no example exists the point tends to soften on its own. Most systems also allow a written response, which is worth using when something is factually wrong — calmly, briefly, and without relitigating the tone.

Timing helps here too. Reviews that draw on peer input usually collect it weeks before you see anything, so the moment to influence what appears is when nominations are being made rather than when the draft lands. If you are asked who should be consulted, name people who actually saw the work rather than people who like you, because vague endorsement from a distant colleague adds nothing a reader can use.

Where the whole thing is worth less than it looks

Some review processes are close to ceremonial: forms completed in a rush the week before a deadline, ratings decided in advance, feedback that arrives once a year and could have been given in March. In that environment the document still matters for pay and for the record, but you should not mistake it for a source of development, and you will need to get real feedback elsewhere.

The genuine early-career failure is treating the annual review as the moment you find out how you are doing. Anything in that document should have been said already, and if it was not, that is a signal about your manager rather than about you. Ask often enough during the year that the review contains nothing new, and it stops being an event and becomes a summary — which is what it was supposed to be.