Proximity Bias Is Costing Remote Engineers Their Promotions
I shipped the migration that saved the team six hours a week of manual deploys. Wrote the postmortem template we still use. Onboarded three junior engineers who now outrank me on the internal leaderboard for PR velocity. And I watched two people who sat closer to the VP get promoted to staff before I got a title change.
I was remote. They weren't. That's the whole story, and for a long time I refused to believe it was that simple.
The promotion gap has a number now
For years, "proximity bias" was a phrase HR decks used to sound thoughtful about hybrid policy without committing to anything. In 2024, Stanford economist Nick Bloom's research on hybrid work gave it teeth: fully remote employees were promoted at roughly half the rate of employees working in an office full time. Employees on a hybrid schedule — specifically two days a week in the office — kept pace with their fully in-office peers on promotions, while also cutting attrition by about a third compared to five-day-in-office mandates.
Read that again, because the shape of it matters more than the headline. The gap isn't remote versus office. It's fully invisible versus visible enough. Two days a month of face time didn't move the needle. Two days a week did. That's not a productivity threshold — nobody's output triples between zero office days and two. It's a noticing threshold: the minimum dose of physical presence required for a manager's brain to file you under "someone I actively track" instead of "someone whose name appears in Slack."
This tracks with what Microsoft's Work Trend Index has been reporting out of leadership surveys for the past several cycles: a large majority of leaders say they struggle to trust that people they can't see are actually working, even when the metrics in front of them say otherwise. That's not a data problem. Their dashboards already show the work. It's a wiring problem — a mismatch between what humans evolved to trust (a body in a chair) and what remote engineering actually produces (commits, docs, decisions in a thread at 11pm).
Why your GitHub graph doesn't fix this
Every remote engineer I know has the same reflex when they sense they're being overlooked: ship more. Write more RFCs. Close more tickets. Build the dashboard that finally, finally makes the invisible work visible.
It doesn't work, and here's the mechanism why. Promotion decisions aren't made by reading your commit history — they're made in a calibration room, days or weeks after the work happened, by a manager reconstructing your impact from memory. Memory is associative, not archival. It retrieves what it can attach to a scene: the time you caught the bug in the standup, the hallway conversation where you talked someone off a bad architecture decision, the look on your face when the release went sideways and you owned it in front of the room. None of that indexes searchably. None of it shows up in a PR count.
A Future Forum Pulse report on hybrid equity put a number on the leadership side of this: a large share of executives named preventing an in-office/remote fairness gap as a top priority for their organizations — which is itself evidence the gap was already showing up in their own promotion data, visibly enough to become a stated goal rather than a hypothetical risk. You don't build a mitigation plan for a bias you haven't already measured.
So the fix isn't "do more work nobody sees." It's making sure the work gets seen by the right person, in the room where calibration happens, before memory decays. That's a distribution problem, not a productivity problem — and most remote engineers are solving the wrong one.
What actually closes the gap
I stopped trying to out-produce the bias and started designing for the two-days-a-week effect deliberately, without needing an actual office day to do it.
I made my manager's memory do less work. Instead of a monthly self-review nobody reads until performance-cycle week, I started sending a two-paragraph "what moved and who noticed" note every two weeks — not a brag sheet, a factual record with names attached: who I unblocked, whose incident I closed, whose design I pushed back on and why. It gave my manager pre-written material for the calibration room instead of asking them to reconstruct three months of my work from a Slack scroll the night before.
I picked the meetings that function as the office. Not every sync matters, but the ones where cross-team priorities get argued out loud — those are the modern equivalent of hallway visibility. I stopped skipping them to protect focus time and started treating them as the two office days Bloom's data says close the gap. Camera on, first to speak, name attached to the idea before someone else restates it as their own.
I asked for the unglamorous, visible work on purpose. The incident channel at 2am. The cross-team migration nobody wants to own. Not because it's good engineering — plenty of quiet, excellent engineering never gets attributed — but because visible ownership of a mess is the fastest way to get filed under "reliable" by people who don't see you daily.
None of this is fair. Compensating for a bias by performing more visibility isn't equity, it's a workaround, and it privileges the engineers who can spare the emotional labor to do it. But "this shouldn't be necessary" and "this is what closes the gap right now" are both true at once, and only one of them helps you at your next calibration cycle.
The org-level fix nobody wants to ship
The honest fix isn't individual — it's structural, and it's the one companies keep deferring. If Bloom's data is right that two office days close the promotion gap almost entirely, the actual lever isn't "return to office" mandates, which torch retention for exactly the people doing the best remote work. It's forcing visibility into the promotion process itself — written impact logs required for every calibration, cross-manager peer input instead of single-manager memory, skip-levels who actually read the artifacts instead of defaulting to who they remember seeing.
Some companies are already doing versions of this — not because it's virtuous, but because the remote work isolation research is forcing hybrid policy to get specific instead of aspirational. The ones that don't will keep losing their best remote engineers to the moment those engineers finally do the math on how long they've been quietly passed over.
You can out-work a lot of things in this industry. You cannot out-work a manager's memory of who was in the room. Until the process accounts for that, the fastest way to get promoted remains sitting closer to the person deciding — no matter how good the code is.