You've just finished writing a review. You felt fair. You mentioned her "great energy," his "strong delivery," her "collaborative spirit," his "ownership." Then you hit submit. What you probably didn't notice: one of those two people just got evaluated on results, and the other got evaluated on personality.
That's how gender bias enters a performance review. Not with a slur, not with a policy. Through adjectives you didn't think twice about, criteria applied unevenly, and a vague memory of "how someone came across" in a meeting six weeks ago. At a time when more teams are remote and reviews are written more than they're spoken, those words travel further and cost more. Recognizing and addressing gender bias in performance reviews is not a compliance chore. It is the difference between promoting the right person and promoting the loudest one.
Key takeaways
- Gender bias in reviews is usually linguistic and structural, not intentional — it lives in criteria, adjectives and memory.
- Women tend to be judged on personality and likeability; men on results and potential.
- Ambition and directness read as leadership in men and as a problem in women.
- The fix is procedural: shared criteria, calibrated ratings, rewritten comments and structured check-ins.
- Vague praise feels kind and quietly stalls careers.
How gender bias actually gets into a performance review
Most of the time, it walks in through the door you left open: the criteria. If your rating form asks for "leadership presence," "executive communication" or "culture fit," you've built a room with no walls. Every rater fills that space with whatever feels right, and "feels right" is where bias lives.
The second door is memory. Reviews written weeks after an event rely on what stuck. A single interrupted meeting, a single visible mistake, a single moment of sharpness in a Slack thread — those stay. The 40 quiet deliverables that shipped on time don't. When someone's reputation is already less generous going in, that recall problem compounds.
The third door is the sentence itself. "She's a strong collaborator but can be a bit abrasive." Nine words, and you can't quite defend them in a room where she asks what "abrasive" means.
How does gender bias affect performance reviews?
It changes what gets measured and what gets rewarded. In practice, women tend to be reviewed more on personality and likeability, and men more on results and potential. That means identical work can produce different ratings, different salary recommendations and different promotion paths — not because the rater is hostile, but because the criteria never forced the rater to be specific.
The knock-on effects are more damaging than a single low rating. Less demanding feedback means fewer visible wins. Fewer visible wins means slower advancement. Add a mix of vague praise and personality notes, and you've built a review that feels encouraging and produces nothing.
How to avoid bias in performance reviews (a working protocol)
Here's the version that actually changed my practice, in four steps. I'll admit I resisted step one for a while, because rewriting criteria sounded like paperwork. It wasn't. It cut the number of unarguable, subjective comments in my own reviews by roughly half.
- Rewrite your criteria as evidence. Replace "leadership presence" with "led X project from brief to launch and unblocked at least two cross-team dependencies." Replace "strong communicator" with "wrote project specs that engineering shipped without rework." If you can't name what evidence counts, delete the criterion.
- Separate results from style. Two questions, two boxes: what did this person deliver, and how did working with them go. Never let the second box override the first. Never let the first box absorb the second.
- Calibrate ratings across raters. Before you finalize anything, read three reviews side by side — one from each of your direct reports. If the personality-heavy feedback clusters around one group, you have a pattern, not a one-off.
- Ask the same question twice. "What specifically did they do?" and "What would have to change for this rating to be higher?" If the second answer is fuzzy, the rating is fuzzy too.
None of this requires a new HR system. It requires you to write criteria that another person could check.
What gender bias actually sounds like: concrete examples
You won't spot it in a policy document. You'll spot it in a table of comments, side by side. Here's a small version of a pattern I've seen repeatedly in review batches:
| What you wrote | What it evaluates | What it should have said |
|---|---|---|
| "She has great energy and is lovely to work with." | Likeability | "She shipped the migration a week early and flagged the rollback risk before it hit production." |
| "He can be a little blunt sometimes." | Perceived tone | "He gave direct feedback in the debrief; check whether it landed and whether the result improved." |
| "She's a strong collaborator but needs to be more assertive." | Tone policing | "She raised the budget concern twice in planning; the second time, it was ignored. What does 'assertive' mean here?" |
| "He's ready for the next step — huge potential." | Halo, promise | "He delivered the top two accounts and mentored one junior. Both are evidence for promotion." |
Notice what the left column has in common. All four are abstract. The middle column names what's actually being judged. The right column names what could be verified. That's the whole method.
Can you give me an example of performance bias?
Yes, and it's usually the least dramatic one. A reviewer writes "reliable, steady, great team player" for a competent engineer. Sounds positive. But "reliable and steady" doesn't win promotions. It signals someone you depend on, not someone you bet on. That phrase, repeated across cycles, is a career limiter dressed as a compliment.
The ambition double standard
Direct the same behavior at two different people and you get two different stories. Assertiveness in one person reads as "leadership potential." In another, it reads as "difficult." You don't need a study to see it. You just need to sit in calibration and listen to which reviews carry the words "too much" or "just a bit."
Which brings up the real question: what do you do with a rater who says "she rubs people the wrong way"? You ask for the specific moment, the specific person, and the specific behavior. Then you check whether the same behavior got flagged for the other direct reports reporting to the same manager. The answer is usually "no."
The vague feedback problem nobody budgets for
Look, I've been guilty of this too. In one review cycle early in my career, I wrote "keep doing what you're doing" for a colleague I genuinely rated highly. When she asked what specifically to keep doing, I could list three things off the top of my head, but none of them were in the review. She got praise. She didn't get direction.
That's a different kind of bias, and it's easier to miss because it feels generous. Vague, personality-based comments go to the person you like. Specific, actionable feedback — the kind that builds a case for promotion — goes to the person you think is going somewhere. If that second group skews one way across your team, you've got a pattern.
The fix is embarrassingly simple. Before you hit submit, ask: could the recipient act on this sentence? If not, rewrite it. If you can't rewrite it, delete it.
What to change first (and what to let go of)
If you only do one thing this cycle, rewrite your rating criteria so they describe observable evidence rather than qualities. If you can do two things, add a calibration session where every rater reads three reviews from colleagues whose performance was rated at the same level. You'll find the pattern in under an hour. I know, because I found mine.
The reviews we write are the most visible, most quotable artifacts we produce about the people we manage. They outlast the meeting, the quarter, the manager. That's a lot of weight on a few lines of text — and the people who read them will remember the adjectives long after they've forgotten the project. Choose the ones you'd be willing to defend in a room where the person is sitting next to you. That's not a high bar. It's just the right one.
