Proximity, and What It Does to an Assessment
The person you see most is the person you can describe best, and description is what assessment runs on.
three things · one reached them · one with no channel at all · the room side is not counted
The three questions a promotion panel will ask.
Six beliefs one manager held going into a promotion round, with the source of each
She handles a difficult client better than anybody else on the team
Sat through four of those calls in the same room, and none of it is written anywhere
He is the person the others go to when something breaks
Watched it happen repeatedly; it is in no objective and no record of his work
Her written proposals are the strongest the team produces
Six documents anybody can read, which is why this one would survive being questioned
He delivers close to his estimates, more than anyone
Checked against eighteen months of dates, which took twenty minutes and settles the matter
She has seemed less engaged since the spring
Formed from shorter messages and a camera that is usually off, and resting on nothing at all
He is not interested in leading anything
Said once, months ago, to a third person, and never put to him
six claims · two rest on hearsay or on nothing
Six beliefs, four of them about people outside the room, and the two with no real source are both about somebody's state of mind. The two that would survive a panel are both about written output, because written output is the only thing that travels without a witness. One manager's own notes before one promotion round; what any organisation permits a manager to rely on differs by employer and by what it has agreed to record.
Assessment runs on description. The person you can describe in specific terms gets a strong case made for them, and the person you can only describe in general terms does not, regardless of what either of them actually did.
The evidence problem in “Proximity, and What It Does to an Assessment” is easier to discuss when the underlying record can be inspected and corrected. A nonprofit reviewing continue to the provider page for mouse jiggler detection should use time and project context as one input, explain the collection to staff and volunteers, and keep consequential judgements with an accountable manager.
Proximity determines who you can describe, which is how seating turns into career outcomes.
For a separate nonprofit perspective on “Proximity, and What It Does to an Assessment”, see BoardSource governance resources. Use the outside material to test governance, inclusion and implementation questions against the organisation’s real circumstances.
What a panel actually asks for
An example. Not an impression, not a rating, not a summary of the year: a specific thing somebody did, in enough detail that the panel can judge it themselves.
Anybody who cannot produce one loses the argument.
Where examples come from
Almost entirely from having been present. The examples a manager reaches for are moments they witnessed, and they witness what happens near them.
This is not a failure of recall or of fairness. It is that the raw material for a specific description is an observation, observations are made by being there, and nobody can be there for somebody in another building.
The shape of the resulting bias
Two people of equal ability produce very different-looking cases. One case is made of incidents; the other is made of adjectives.
Panels prefer incidents, correctly.
Why calling it bias is not quite right
Bias suggests a thumb on the scale. This is closer to an instrument problem: the measuring device works at short range and reports nothing at long range, and the absence of a reading is treated as a low reading.
Everybody involved can be entirely fair and the outcome still falls the same way. Which means exhorting managers to be fair does not touch it, and most organisations' entire response to this problem is exhortation.
The same two people, described from the same desk:
Handled the escalation in March: took the call himself, got the client to agree a revised sequence, and had it in writing by the Friday.
Reliable. Thoughtful. A safe pair of hands.
The recency component
Proximity has a time dimension as well. The person you saw yesterday is easier to describe than the person you saw in March.
Somebody outside the room is permanently in March.
What tends to get over-credited
Visible effort, availability, speaking in meetings, and the handling of small urgent things. All of these are easier to observe from across a room than from anywhere else.
None of them is worthless and none is the job.
What tends to get under-credited
Sustained, difficult, uninterrupted work that produces one good thing after six weeks. It is nearly invisible while it is happening and the result arrives without a story attached.
This is disproportionately the work people do when they are not in a room, which compounds the problem rather than offsetting it.
The test for your own case
Write two sentences about each person on your team from memory, now, without looking anything up. Then look at the length and specificity of each.
The ones you struggled with are the ones the system is failing, and the exercise takes ten minutes.
What the test usually reveals
A clean split by proximity, with the people you sit with described in incidents and everybody else described in traits. It is an uncomfortable list to look at and it is the most useful ten minutes in this section.
Do it before the assessment cycle rather than during it, because during it you will be constructing a case rather than taking a reading.
Calibration meetings make it worse
Where managers meet to compare people across teams, the comparison is conducted in anecdotes, out loud, under time pressure. The manager with vivid specifics wins the room.
That process is not neutral between a case built from observation and a case built from reading somebody's work carefully, even though the second is better evidence. Knowing this in advance is worth a great deal: it means preparing two or three concrete incidents to say out loud, not just a well-organised file nobody will open.
Correcting for it honestly
Not by inflating anybody. By going and getting the missing material: reading their work properly, asking the people who do work with them, and doing it in the months before the decision rather than the week of it.
The correction is research, not generosity.
What research can realistically recover
Output, and other people's first-hand accounts. That is a substantial amount and it is not the same thing as having watched somebody work.
It is enough to produce a case made of incidents rather than adjectives, which is the thing that was missing.
What stays uncorrected
The accumulated sense of somebody that a manager builds over a year of sharing a room, which influences every judgement and cannot be assembled deliberately. Nothing in this note reaches it.
The honest position is that a manager in a room will always know the people in the room better, that this will always show up in assessments, and that the available remedy is to work much harder on the evidence for the people they see least — knowing that the result is a reduction in the gap rather than its removal. Pretending the gap has been closed is how people end up being told their case was considered equally when it was not.