Why Your Performance Review Sounds Like ChatGPT (And How to Fix It)
Performance reviews are starting to sound eerily similar across teams and organisations. Here's why that's happening, why it matters, and what you can actually do about it.
Why Your Performance Review Sounds Like ChatGPT (And How to Fix It)
If you've read a performance review lately and thought, "this could have been written about literally anyone," you're not imagining it. Across organisations of all sizes, written feedback has started to converge on the same safe, polished, oddly interchangeable language. "Consistently meets expectations." "A strong team player." "Demonstrates good communication skills." Sound familiar?
The uncomfortable truth is that a lot of performance reviews now read like they were drafted by a language model. And increasingly, some of them actually were.
How Did We Get Here?
Templates Made Life Easier (And Reviews Worse)
Over the past decade, HR platforms and content sites have handed managers something they desperately wanted: a shortcut. Phrase banks, competency libraries, downloadable templates. All designed to help time poor people get through review season without a breakdown.
The problem is that when everyone draws from the same pool of stock phrases, the output becomes indistinguishable. "Needs to improve communication." "Takes initiative." "Brings energy to the team." These phrases describe nobody in particular, which means they're functionally useless as feedback.
Managers Are Uncomfortable, So They Play It Safe
Reviews are genuinely hard. They require honest assessments of real people, often in writing, with paper trails. Research on appraisal effectiveness consistently shows that vague, non specific feedback tends to come from managers who are either under time pressure or simply uncomfortable delivering anything that could start a difficult conversation.
The result is reviews that are technically positive but practically meaningless. "Keep up the good work" tells an employee nothing about what they actually did well, why it mattered, or what they should do more of.
Generic AI Drafted It, Nobody Questioned It
Generative AI can make performance reviews significantly better. But it can also make the generic review problem worse, and right now, a lot of organisations are experiencing the latter.
The issue isn't the technology. It's the workflow. Managers who use AI to surface themes from a year's worth of notes, flag vague language, and sharpen a draft they've already written in their own voice? That's AI working well. Managers who type a two line prompt, accept whatever comes back, and move on? That's where reviews start to feel like they were written about a fictional composite employee rather than an actual person.
HR and legal experts call this automation bias: the AI output looks polished and professional, so it gets waved through without the scrutiny it needs. The result is a review that reads smoothly and says almost nothing. Worse, the employee on the receiving end usually knows.
Why Vague Feedback Is a Genuine Problem
This isn't just an aesthetic issue. Research is pretty clear that feedback quality directly affects performance, motivation, and perceived fairness. Vague feedback, especially when it's the only kind someone receives, undermines self confidence, creates uncertainty about what's actually expected, and can lead to disengagement.
There's also a well documented equity dimension. Studies examining reviews at large organisations have found that women are significantly more likely to receive vague praise ("great team player", "well liked by colleagues") and less likely to receive feedback linked to concrete outcomes and business results. Men's reviews more often contain specific references to projects delivered, metrics hit, deals closed. This isn't a small stylistic difference. Vague feedback correlates with lower performance ratings for women but not for men, suggesting it operates as a quiet but meaningful form of bias.
Generic, templated language isn't just unhelpful. It can actively disadvantage people.
How to Tell If Your Reviews Have the Problem
A quick sense check: could the review paragraph you've just written apply to three or four other people on your team with only a name swap? Are there any specific projects, incidents, or metrics mentioned? Does the language reflect how this particular person actually works, or does it sound like it came from a brochure?
If the answer to most of those is "well, sort of" or "not really," the review has a specificity problem. Whether AI wrote it or not is almost beside the point.
What Good Feedback Actually Looks Like
Behaviours and Outcomes, Not Traits
The fix isn't complicated, but it does require a bit more effort. High quality feedback describes observable actions and connects them to real results. Not "strong communicator" but "led the Q3 client update calls and reduced escalations by about 30% over the quarter." Not "takes ownership" but "identified the data discrepancy before the board presentation and resolved it without being asked."
This kind of specificity is harder to write. It also happens to be genuinely useful to the person reading it.
Use SBI as a Simple Framework
The Situation Behavior Impact (SBI) framework is one of the most practical tools available for structuring feedback. It's three questions: When and where did this happen? What did the person actually do? What was the effect on the team, the customer, or the business?
Combined with documented goals or KPIs from the review period, SBI keeps feedback grounded in evidence rather than impressions. It also makes it much harder to accidentally fall back on personality descriptors.
Make Development a Conversation, Not a Paragraph
One of the better shifts in modern performance management thinking is separating the formal evaluation from the development conversation. The written review captures what happened. The conversation is where you talk about what comes next, what support the person needs, and what growth looks like for them. Treating the written document as the end of the process, rather than the beginning of a dialogue, is part of why reviews feel so impersonal.
Using AI Without Losing the Point
Here's where I'll be direct: AI assisted performance reviews, used well, are a genuinely good idea. The key phrase is "used well."
Using AI to summarise a year's worth of notes, flag potentially biased language, or generate a first draft that you then refine and edit? That's a reasonable use of the technology.
The model that works is human led, AI aided. You bring the judgment, the context, and the accountability. The AI helps you express it more clearly and consistently. That distinction matters enormously, both for the quality of the output and for the fairness of the process.
Managers should also be trained to push back on generic AI outputs. If a generated draft describes your direct report in terms that feel too unspecific or slightly off, that's a signal to rewrite, not just tweak. The AI doesn't know what happened in that difficult client meeting or how the person handled the restructure. You do.
A Few Practical Steps
For managers who want to immediately improve their reviews:
Keep a running note of key contributions, outcomes, and examples throughout the year. Memory is unreliable; notes aren't. Draft in your own words first, even if just bullet points. Then use AI to help with structure or phrasing if needed. Every time you write a trait ("great communicator," "lacks ownership"), add a concrete example and explain the impact. Use forward looking language for development areas. "Next quarter, focus on X" is more useful than "needs to improve at X." Treat the written review as an opening, not a verdict. Ask the person how the feedback lands.
For employees who receive reviews that feel generic: ask for examples. Ask how the feedback connects to your goals or your path forward. Bring your own evidence of what you've delivered. You can nudge a vague review toward a more useful conversation.
A Better Way to Write Reviews
If you want to make the shift from boilerplate to genuinely useful feedback, Perform Review is built exactly for that. The platform helps individuals and teams produce high quality, professional self assessments and peer reviews with AI assistance that guides rather than replaces the process. The result is feedback that's specific, structured, and actually sounds like it was written by a human who paid attention, in a fraction of the time. Because it was.