What is recognition bias?

The pattern in which recognition consistently goes to some people more than others for reasons unrelated to their work: visibility, proximity to the manager, role, or identity. It is usually unintentional and shows up clearly once counted.

Who tends to be under-recognised at work?

People whose work has no visible artefact, remote staff, those in back-office roles, and members of groups whose contributions are read as expected rather than notable. The common thread is that their work is less seen, or seen differently.

How can a company reduce recognition bias?

Count. Tally recognition by person, team, location and role each quarter and look at who is missing. Then widen who can recognise, so that colleagues who see the invisible work can name it, and tie recognition to specific behaviours rather than general impressions.

Recognition glossary

Recognition bias

Recognition bias is the systematic tendency for recognition to go disproportionately to certain people, such as those who are more visible, closer to the manager, in customer-facing roles or from majority groups, regardless of the distribution of good work.

Nobody sets out to recognise the same people every time. It happens anyway, and the reasons are structural before they are personal.

How it arises without anyone meaning it

Recognition follows attention, and attention is not evenly spread. A manager sees the people who present, those who sit nearby and those whose work produces a thing that can be looked at. The engineer who keeps the build green, an analyst whose reports are never wrong, a coordinator who prevents the problems that would otherwise have been visible: none of them generate a moment in which they can be noticed.

Then there are the filters people carry. Research on workplace feedback finds repeatedly that the same behaviour is described differently depending on who does it: assertive from one person, abrasive from another; helpful from one, leaderly from another. Recognition inherits this. The DEI post on this site covers the evidence, and the pattern in the second example above, where women are thanked for support and men for delivery, is among the best documented.

Location is the newest filter. Hybrid teams recognise the people in the room more than the people on the call, and the effect is large enough to show up in a quarter’s data.

What it does

The obvious harm is unfairness. The less obvious harm is that recognition is information about what the company values, and a biased distribution teaches the wrong lesson. If the awards all go to presenters, the message is that presenting is the work. If remote staff are never named, the message is that proximity is what counts. People adjust their behaviour to the signal, and the quiet, necessary work gets quieter.

Seeing it

Bias is hard to notice in the moment and easy to notice in a table. Once a quarter, list every recognition given and tally by recipient, then by team, role, location and any demographic dimension the company already tracks. Compare against headcount. The metrics post describes the mechanics.

The pattern is usually visible immediately. A department where 60% of the recognition went to 15% of the people is not a department with 15% stars. It is a department where 85% of the work is not being seen.

Reducing it

Widen the field of view. Peer-to-peer recognition exists partly for this reason: colleagues see the invisible work that managers miss, and a program that lets anyone recognise anyone corrects for a single manager’s line of sight.

Anchor recognition to behaviours. A message tied to a named value or a specific act is harder to give on impression alone, which is the argument the values-based recognition page makes.

Prompt for the missing. A program that tells a manager which of their reports has not been recognised in six weeks is doing the counting for them, and the reach metric is the number that makes the gap visible.

What it looks like

A year of spot awards in a forty-person department, tallied: every one went to somebody who presents at the Friday demo. The people maintaining the systems the demos run on received none.
A peer recognition feed where women are recognised more often for helping and supporting, and men for delivering and leading, on a team where the actual work is shared roughly evenly.
A hybrid team where the three people who come into the office with the manager receive twice the recognition of the five who do not, for comparable output.

Questions people ask

What is recognition bias?: The pattern in which recognition consistently goes to some people more than others for reasons unrelated to their work: visibility, proximity to the manager, role, or identity. It is usually unintentional and shows up clearly once counted.
Who tends to be under-recognised at work?: People whose work has no visible artefact, remote staff, those in back-office roles, and members of groups whose contributions are read as expected rather than notable. The common thread is that their work is less seen, or seen differently.
How can a company reduce recognition bias?: Count. Tally recognition by person, team, location and role each quarter and look at who is missing. Then widen who can recognise, so that colleagues who see the invisible work can name it, and tie recognition to specific behaviours rather than general impressions.