The person who gets promoted is often the one who can explain their impact, not the one who did the most work. Your manager has a dozen other things on their mind. If you cannot say what you shipped last spring and why it mattered, they probably cannot either.
The fix is boring. Write wins down when they happen. Then, when the conversation comes, you are editing a record instead of trying to remember.
Start with the outcome
Most people describe activity: “Worked on the CI pipeline.” Your reader wants the result: “Cut deploy time from 25 minutes to 15.” Put the outcome first and the work second.
If you do not have a number, use a fact. “Nobody on the team had to be paged for the billing job after that” is still evidence.
Pick the wins that fit the next level
You do not need everything. Look at what the next level asks for and pull the wins that match. Four kinds tend to carry the most weight:
- Things you shipped that mattered to someone besides you, with your specific role in them.
- Problems you found that others had walked past.
- Other people’s work you made better, through mentoring, reviews or onboarding.
- Decisions you drove without being asked.
A win that took a weekend and saved the team a week usually beats a six-month project nobody remembers.
Group by theme, not by date
A chronological list reads like a diary. Three or four themes read like a case. “Reliability”, “Mentoring” and “Customer-facing work” each give you a heading and a handful of wins underneath, and you can see at a glance whether the pattern points to the next level.
Watch for the direction as well. If your wins show your scope growing, from fixing tickets to owning a system to setting its direction, say so directly. That trend is the argument.
Do it before you need it
Tag wins as you log them, with something like impact or leadership, and filter by your review period when it is time to write. In Hype Doc you can also ask an AI tool to pull those wins and draft a first version. See Draft a review with AI for the setup.
The draft will be rough. That is fine. It is much easier to cut a long list of real examples than to invent one from scratch the night before.