AI and Performance Reviews: What It Needs From Your HR Data
← All insights Insights

AI can summarise a performance review. It can’t invent one.

11 November 2026 · 4 min read · Hubb Group

Key takeaways

  • AI is good at compressing and drafting; it cannot invent evidence your review process never collected.
  • What a model needs is structure — several raters on the same questions, competencies mapped to roles, self-scores beside reviewer scores, calibration and a history.
  • Anonymity is worth more than any summary: small reviewer groups should be pooled so a single voice is never isolated.
  • Run one proper cycle first. Then a summary is worth reading — and a manager can act on it without one.

Every HR platform is bolting AI onto appraisals, and the demos are seductive. Paste in a year of feedback, get back a tidy paragraph on where someone is strong and where they need support. It looks like the end of the worst week in the HR calendar.

Then you connect it to your own company and it has almost nothing to read.

That is not a failure of the model. Language models are genuinely good at this kind of work: compressing a lot of text, noticing what three people said in different words, drafting a first pass a human then fixes. What they cannot do is invent the evidence. If your review process produces one manager’s notes, once a year, written the night before the meeting, then a summary of it is a tidier version of one person’s opinion — and the tidiness makes it more convincing, not more true.

The input problem

Ask what a model would actually be reading. In most companies the honest answer is a single rater, no comparison, and a scale nobody agreed on. Was that four out of five generous? Did "a great team player" mean the same thing from the two managers who wrote it? Nothing in the data says, so nothing in the summary can either.

Worse, the gaps are invisible. A model given thin evidence does not refuse — it writes you a confident paragraph. The output looks identical whether it was built from forty considered answers or four rushed ones.

What a model would need to be worth having

  • More than one rater on the same person, answering the same questions — manager, peers and direct reports, not a single point of view.
  • A competency library mapped to roles, so "leadership" means a defined level for that grade rather than a word everyone reads differently.
  • The self-assessment alongside the reviewer scores, so blind spots and undervalued strengths are in the data rather than in a conversation nobody recorded.
  • Calibration, so a four from one panel means roughly what a four from another panel means.
  • A history — the same person, the same competencies, cycle after cycle. Most of what is useful about a review is the shape of the line, not this year’s dot.

None of those are AI features. They are process features, and they are the thing to fix first — whether or not you ever point a model at the result. They are also exactly what a 360° cycle in Hubb People produces: multi-rater campaigns against one question set, a self-versus-reviewers radar per competency, calibrated final ratings, and the same structure repeated every cycle so the comparison holds.

One thing you should not hand to a model

Anonymous feedback. The reason people write candidly in a 360 is that they believe it cannot be traced back, and that belief is fragile — a small reviewer group plus a distinctive turn of phrase is often enough. Hubb People pools reviewer groups smaller than three automatically so a single voice is never isolated, and that protection is worth more to the quality of your data than any summary.

The same care applies to where the text goes. Review comments are personal data about named employees. Before any of it reaches a third-party model, somebody should be able to say plainly which service processes it and on what terms. "The vendor added AI" is not an answer to that question.

Start with the part AI cannot do for you

Run one cycle properly. Ask several people the same questions, compare the answers with the subject’s own, calibrate the scores, and turn the result into agreed goals with owners and dates. Do that twice and you have something a model can genuinely help with — and, more usefully, something a manager can act on without one.

If you would rather see it than read about it, Hubb People loads a realistic sample team and a part-finished 360 on day one, so you can click through every screen before putting a real person in it. We also build the AI side of this work for clients — that is what Hubb AI is — and the order has never changed: get the record right, then automate it, then let a model read it.

Frequently asked questions

Can AI write performance reviews for me?

It can draft one from evidence you already have, and that is genuinely useful. It cannot supply the evidence. Given one manager’s hurried notes it will still produce a confident, readable paragraph — which is the risk, because the output looks the same whether it was built from forty considered answers or four.

What data does AI need to be useful in HR?

Multi-rater answers to the same questions, a competency library mapped to roles so ratings mean something specific, the subject’s self-assessment alongside reviewer scores, calibrated final ratings, and the same structure repeated each cycle so year-on-year comparison holds.

Is it safe to put employee feedback into an AI tool?

Review comments are personal data about named people, so before any of it leaves your system somebody should be able to say which service processes it and on what terms. Anonymous feedback deserves particular care — a small reviewer group plus a distinctive phrase can be enough to identify the author.

Does Hubb People use AI for reviews?

Hubb People’s job is the structured part: 360° campaigns, a self-versus-reviewers competency radar, calibration, development plans and anonymity protection. That is the input any AI would need. We build AI tooling separately — Hubb AI — and the order matters: get the record right first.

Want this kind of thinking on your project?

Tell us what you're working on — we'll come back with a clear plan and a fixed quote.

  • Fixed price, agreed up front
  • A named lead on your project
  • No obligation, and no sales sequence

No obligation. We reply within one business day. Privacy