PerfCopilot
Get started

Related: how to reduce bias in performance reviews.

Proximity Bias in Performance Reviews (and How to Fix It)

By Samira Bahmanyar · HR Manager

Definition Proximity bias in performance reviews is the tendency to rate the people you see most often, usually those physically near you, more favorably than people whose work reaches you through a screen, even when their output is equivalent. Visibility gets read as productivity, and presence gets read as commitment.

Two people on the same team deliver the same quarter of work. One sits ten meters from their manager. The other works from home three days a week, or from another city entirely. At review time, the first one "just feels" stronger. Nothing in the work explains the gap. Where it happened does.

Key Takeaways

  • Simply being seen at work shapes how people judge you: observers infer traits like "dependable" from ordinary face time, without intending to and without noticing (Elsbach, Cable & Sherman, 2010).
  • Remote work does not have to cost anyone a rating. A large randomized hybrid trial found no harm to performance reviews or promotions (Bloom, Han & Liang, 2024). The risk sits in the review process, not in the location.
  • The fix is structural: judge every person on the same artifacts, and check your ratings by location before calibration.

Try it on your own data: PerfCopilot runs a bias checker on any review draft, and its performance review generator drafts from the work itself (tickets, pull requests, deals, projects), so a remote contributor's output carries the same weight as an in-office one's. Free for teams of up to 5.

What is proximity bias?

Proximity bias is a shortcut in how managers judge contribution. The brain treats what it sees often as more important and more reliable than what it has to look up. In an office, that means the person who is always at their desk, in the hallway conversation, or first into the meeting room accumulates an impression of effort that a remote colleague, whose work arrives as a merged change or a closed deal, never gets the chance to build.

It is not a new idea. Long before hybrid work, organizational researchers studied "face time": simply being seen at work. In a pair of studies, Kimberly Elsbach, Daniel Cable and Jeffrey Sherman found that observers drew trait conclusions from passive face time alone. Being seen during normal hours led to inferences that a person was "dependable", and being seen outside normal hours led to inferences that they were "committed". Crucially, those inferences were spontaneous: people formed them without meaning to and without awareness (Elsbach, Cable & Sherman, Human Relations, 2010).

That is the core of the problem. A manager does not decide to favor the in-office employee. They simply absorb more "dependable" and "committed" signals from that person, and those signals feel like evidence when the review form opens.

Why distributed teams make it worse

Hybrid and remote work did not invent proximity bias, but it widened the gap between the people who generate visibility signals and the people who do not. On a fully co-located team, everyone gets roughly the same face time. On a hybrid team, face time becomes uneven by design, and uneven inputs produce uneven impressions.

There is direct, if older, evidence of the cost. In a well-known randomized experiment at the Chinese travel company Ctrip, call-center staff who worked from home performed better than their office-based colleagues, yet were less likely to be promoted once performance was taken into account (Bloom, Liang, Roberts & Ying, Quarterly Journal of Economics, 2015). The work was there. The credit was not.

More recent data is more encouraging. A 2021 to 2022 randomized trial with 1,612 employees at Trip.com (the same company, renamed) found that working from home two days a week had no negative effect on performance reviews or promotions, while reducing quit rates (Bloom, Han & Liang, Nature, 2024). Read honestly, the two studies together say something useful: being remote is not automatically penalized, but it can be, and what decides the outcome is whether the review process looks at the work or at the person's presence.

Vendor research points the same way. BetterUp reports that in an analysis of 360-degree evaluations run with ICONIQ, managers rated remote employees less favorably than in-person employees on most dimensions (BetterUp). Treat that as directional: it is a vendor's own analysis rather than a peer-reviewed study.

The honest trade-off: sometimes the in-office person really is doing more. Proximity bias does not mean every rating gap between office and remote staff is unfair. It means the gap has to be explained by evidence you could show either person, not by who you happened to bump into.

What proximity bias looks like on real teams

It rarely announces itself. It shows up as small, reasonable-sounding judgments:

The fingerprint is the vocabulary. Words like "always around", "very present", "highly engaged" and "go-to person" describe how often you saw someone, not what they produced.

How proximity bias interacts with other review biases

Proximity bias rarely travels alone, and it amplifies the biases next to it.

That overlap matters because the fixes overlap too. For the full taxonomy, see 12 types of performance review bias.

How to reduce proximity bias in your next review cycle

General advice like "treat remote staff fairly" fails because, as the face-time research shows, the impression forms without awareness. These steps change the process instead of asking managers to out-think their own perception.

1. Gather the same evidence for every person

Before writing a word, pull the full period's work for each person from the same sources: tickets closed, pull requests reviewed, deals moved, documents written, projects delivered, peer feedback received. If an in-office employee's review rests on "I saw them handle it" and a remote employee's rests on artifacts, you are holding them to different evidence standards.

2. Write evidence before ratings

For each competency, list two or three dated examples first, then pick the rating. If you cannot find any artifact for "collaboration" for the person you see every day, your rating is an impression. The same rule protects remote staff, whose collaboration often lives in written threads and reviews rather than conversations you overheard.

3. Split your ratings by location before calibration

This is the single most useful check. Sort your team by where they mostly work and compare the rating distributions. If in-office staff cluster at the top and remote staff in the middle, do not assume bias, but do require that each gap be explained by evidence. Bring that split into the calibration meeting, where a peer manager can challenge it.

4. Add peer input from people who work with remote staff daily

A manager may see a remote employee once a week. Their closest collaborators see their work every day. Structured peer feedback from the people who actually depend on someone's output is a direct counterweight to a manager's location-skewed view.

5. Equalize the moments that create impressions

Proximity bias is fed by informal moments: the corridor update, the post-meeting chat. Give everyone a regular one-to-one with the same agenda, and move decisions and status updates into written channels everyone can see. That reduces how much of the review depends on who happened to be nearby.

6. Screen the draft for presence language

Before submitting, read the review for words that describe visibility rather than output: "always around", "very present", "engaged", "go-to person", "heads-down", "quiet". Each one should either be backed by a specific artifact or rewritten. Our pre-submit bias checklist covers the full screen.

A note for remote employees: keeping a brag document of your own work helps, because it gives your manager the evidence they did not see. But the durable fix belongs to the process, not to you. A review that requires you to perform visibility is measuring the wrong thing.

Where software fits

Most of the work above is evidence gathering, and that is exactly where a busy manager cuts corners for the people they see least. This is the gap PerfCopilot was built for: it drafts each review from the tools your team already uses (tickets, pull requests, deals, projects, peer feedback), and every claim in the draft traces back to a work artifact the employee can see and dispute. An artifact looks the same whether it came from the desk next to you or from another time zone.

To be clear about scope: PerfCopilot is a review-writing layer, not a full performance platform. It does not run engagement surveys or compensation, and it does not split your ratings by location for you. It gives every person the same evidence base. You still own the judgment and the calibration.

Frequently asked questions

What is proximity bias in performance reviews?

Proximity bias is the tendency to rate people you see often, usually those in the office near you, more favorably than people whose work you mostly see through a screen, even when their output is the same. Being visible gets mistaken for being productive.

Is proximity bias the same as recency bias?

No. Recency bias over-weights when work happened, so the last few weeks dominate the review. Proximity bias over-weights where and how visibly work happened, so the people you physically see get more credit. They often stack: a remote employee's early-cycle wins are both distant in time and invisible.

Does hybrid work automatically hurt someone's performance rating?

No. A large 2024 randomized trial at Trip.com found that a two-day-a-week hybrid schedule did not hurt performance reviews or promotions (Bloom, Han & Liang, 2024). Proximity bias is a risk in how a manager gathers evidence, not an inevitable penalty for working remotely. A process that judges the work rather than the presence removes most of it.

How do you reduce proximity bias in a review cycle?

Gather the same evidence for every person regardless of location, write evidence before ratings, split your team by location before calibration and compare the distributions, add peer input from people who work with remote staff daily, and check your draft for presence words like "always around" or "very present".

Should remote employees just make themselves more visible?

Keeping a running record of their work helps them, but putting the whole fix on the employee teaches people to perform visibility instead of doing the work. The durable fix belongs to the manager: a review process where every rating rests on artifacts that look the same whether the person was in the room or not.

The bottom line

Proximity bias is not a character flaw. It is what happens when one group of people generates a steady stream of "dependable" and "committed" impressions and another group does not, and the review form asks a manager to summarize a year from memory. The research is clear that remote work does not have to cost anyone a rating. Whether it does comes down to one question: did you judge the work, or the presence?

The structural answer is the same evidence for everyone. That is the bet behind PerfCopilot: every claim cited to real work, so the person you never see in the hallway gets judged on exactly the same terms as the person you see every day.

Sources