Your performance rating is not decided in your review. It is decided weeks earlier, in a calibration meeting you will never attend, by a group of managers who mostly know you as a name on a slide and whatever your manager manages to say in the ninety seconds your row is on screen.
I sat on the manager side of that table for years while running IT operations teams at a large telecom, and later helped a few companies design their review cycles as a fractional COO. The pattern held everywhere: the people who did well in calibration were not always the strongest performers. They were the people whose case was easiest to repeat.
That matters more in 2026 than it did five years ago. When Meta announced it would cut roughly 5% of staff as “low performers” in early 2025, Mark Zuckerberg wrote that he wanted to “move out low-performers faster”. Microsoft, which had publicly killed stack ranking in 2013, followed with its own performance-based cuts and a policy that gave low-rated employees five days to choose between severance and a performance plan. A rating is no longer just a bonus multiplier. At some companies it is the list.
Product managers are unusually exposed in that room. Your output is a decision, not a deliverable. Engineers have shipped code, designers have screens, sales has a number. You have outcomes that a dozen people touched. Here is what actually happens to your case once it leaves your manager’s hands.
Seven things that happen to your case in calibration
1. The rater matters more than you do
This is the uncomfortable research finding that every calibration process is quietly built around. In a study of 4,492 managers rated by bosses, peers and subordinates, Steven Scullen, Michael Mount and Maynard Goff found that the rater’s own tendencies explained 62% and 53% of rating variance in their two datasets. Actual performance explained 21% and 25%. Deloitte cited that finding when it rebuilt its own review system, concluding that ratings revealed more about the rater than the ratee.
Calibration exists to correct for this: the lenient manager gets pulled down, the harsh one gets pulled up. The side effect is that your rating now depends on how credible your manager is with their peers. A manager who has inflated before gets discounted, and so do you.
2. Your manager gets about two minutes
A calibration session for a product org of forty people might run three hours. Subtract the arguments over the top and bottom of the curve and the middle of the distribution gets almost nothing. In the sessions I ran, the median employee got discussed for less than two minutes, and most of that time went to whoever objected.
Your manager is not reading your self-review aloud in that window. They are saying two or three sentences. If those sentences are not already written down, they get improvised.
3. Vague praise loses to a specific number
“She’s great with stakeholders” is not an argument. “She killed the pricing page rewrite after three customer calls and saved a quarter of engineering time” is one. Shelley Correll and Caroline Simard reviewed performance evaluations at three tech companies and a professional services firm and found that vague feedback correlated with lower ratings, and that women were more likely to receive it. Only 40% of women’s reviews tied feedback to business outcomes, against 60% for men.
You cannot control how your manager writes, but you can control what material they have. If the only artifact is a list of features shipped, expect a vague sentence about shipping features.
4. Someone else in the room has an opinion about you
Calibration is the one meeting where the engineering director whose roadmap you pushed back on gets a vote. Peer managers are invited specifically to challenge ratings. One sharp comment (“I found her hard to work with during the platform migration”) can move you a full band, and your manager may not have a counter ready.
The defense is not avoiding conflict. It is making sure the senior people you have conflict with also see your reasoning, ideally in writing, before the cycle closes.
5. Scope gets compared across teams that are not comparable
A PM on a mature billing product and a PM on a new AI feature get placed on the same grid. The room will reach for whatever comparison is easiest, which is usually visible scope. This is why owning a problem beats doing a lot of work at review time: “owns retention for SMB accounts” survives the compression to one sentence. “Ran the sprint ceremonies and coordinated three teams” does not, and a lot of that coordination is the glue work that rarely gets credited.
6. The curve is still there, even when it officially isn’t
When Microsoft dropped stack ranking, its HR chief said there would be no more predetermined distribution of ratings, but the rewards budget would continue. That is the honest version of most “no curve” systems. If the bonus pool is fixed, someone has to be below the middle for someone else to be above it. Calibration is where that trade happens, whether or not anyone calls it a curve.
7. Last year’s rating is the anchor
The first question in many rooms is “what was this person last cycle?” A move up needs a story; staying put needs none. If you were rated “meets” last year, your manager has to argue for change, which costs them credibility points they may be saving for someone else. If you got a strong rating after a big launch, expect the room to ask what you have done since.
What to put in your manager’s hands before the meeting
Most advice about reviews focuses on the self-assessment. The self-assessment matters less than people think, because nobody in calibration reads it. What matters is the two or three sentences your manager will say out loud. Write them for them.
A one-line scope statement. The problem you own, in words a stranger understands. Not your team name.
Two outcomes with numbers and dates. “Cut onboarding drop-off from 41% to 29% between March and July.” If you do not have a number, use a decision and its consequence: what you stopped, what it freed up.
One preemptive answer to the likely objection. If you know the platform director thinks you slowed their migration, give your manager the short version of why, and what happened after.
A peer who will back you. Before the cycle, ask a senior person outside your reporting line whether they would speak to your work if asked. Many managers reach out to peers before calibration; make sure at least one of the answers they get is specific.
In my experience the employees who did this were rare, perhaps one in ten, and their managers loved them for it. Arguing a case you have to build from memory, under time pressure, in front of peers who are each protecting their own people, is hard. Handing your manager a clean, defensible paragraph is one of the most useful things you can do for both of you.
The question to ask your manager this month
Calibration happens before you are told your rating, which means the window to influence it is now, not when the review lands. Ask your manager directly: “When my name comes up in calibration, what will you say, and what do you expect someone to push back on?”
Some managers will be surprised by the question. The good ones will answer it. If the answer is vague, you have just learned what the room will hear, and you still have time to change it. If it is honest and uncomfortable, that is the feedback that gets scarcer as you get more senior, and it is worth more than any rating.