The policy
Compliance is not the outcome
Thirty-four days of near-perfect compliance produced thirty-one hours of what the policy was for.
Days this part is about, hour by hour
Monday, the third week
0 of 7Came in to make up a day missed through illness
Wednesday, the third week
1 of 7Tuesday, the second week
0 of 7three days · 21 hours logged · one of them was only possible there · two of the three days contained no such hour at all · four were made worse by being there
Three days the policy treats as identical.
A fair pilot starts with a shared baseline; teams can review how the platform works before comparing what changes after a scheduling or process decision.
For further reading outside this collection, see material published by Krebs on Security.
Compliance on this record was ninety-three per cent: thirty-four days attended of thirty-nine required, with the five gaps being two days of illness, a day at a client, and two exceptions asked for and granted.
By any reporting standard that is a success. It would appear in a quarterly review as a policy working well.
What the quarter actually produced was thirty-one hours that could not have happened anywhere else, fifteen days that produced none at all, and forty-five hours that went worse for being in the building.
Both descriptions are of the same quarter. Only one of them is measured.
Why the measurable thing takes over
Not through cynicism. Through availability.
Attendance can be counted without anybody designing anything, because the building already counts it for other reasons. Overlap, purpose and outcome cannot be counted without someone deciding to count them, building something, and persuading people it is not surveillance.
So one number exists and the others do not, and a number that exists will be used. Within two quarters the policy is managed against the proxy, and the thing the proxy stood for has dropped out of the conversation entirely — not because anybody decided it should, but because nobody has a figure for it to put next to the one they do have.
The specific failure mode
The failure is not that compliance is high and value is low. It is that the two are not connected, so the instrument has no steering on the outcome.
- If value is low and compliance is high, the available move is to raise the requirement. This adds journeys and does not add overlap.
- If value is low and compliance is low, the available move is enforcement. Same result.
- If value is high, nobody asks why, so the thing that produced it — on this record, four planned days with a room and a purpose — is never identified or repeated.
In all three branches the lever being pulled is days, and days are not what varies.
What could be measured instead
Nothing elaborate. Two questions, asked of each person once a quarter, and not aggregated upwards beyond a team:
- On your office days last month, how many of the people you work with most closely were there?
- How many hours in the building last month could not have happened anywhere else?
Neither needs a system. Both would have shown, on this record, that a quarter reported as ninety-three per cent successful was producing about two and a half useful hours a week.
The thing to say in a review
When a policy review opens with an attendance figure, the useful intervention is not to dispute it. It is to put a second number beside it:
Attendance was ninety-three per cent. On my own days, the median number of my team who were in with me was three of nine, and fifteen of my thirty-four days contained nothing that needed the building.
That is not an argument against the office. It is the missing half of the evidence, and without it the review has only one number and will act on it.
The same quarter, described twice
| The reported version | The measured version |
|---|---|
| 93% attendance | 31 of 238 hours needed the building |
| 34 of 39 required days | 15 days produced nothing at all |
| Two exceptions, both granted | 45 hours worse for being there |
| No complaints raised | 73h 40m of travel |
| Policy working | 2h 26m of travel per useful hour |
Both describe the same thirteen weeks. The left column is what exists in the organisation. The right column exists only because somebody wrote down a letter at the end of each hour.
Why adding a measure usually fails
Most attempts to measure the right-hand column turn into surveillance, and the reason is structural rather than malicious.
A measure that is reported upwards becomes a target. A target about collaboration becomes a thing people are assessed on, and people assessed on collaboration will report collaboration. Within two quarters the new number is as hollow as the old one and considerably more intrusive to collect.
The protection is scope. The two questions worth asking stop at the team: they inform the people who can act on the answer, they are not aggregated, and nothing is kept. A number that travels no further than the person who can change something does not need to be defended against.
What a team can do without any of this
Ask the two questions of itself, once a quarter, in a meeting, out loud, and write nothing down.
How many of us were in together, typically? What happened last month that actually needed the room?
Five minutes. The answers are approximate and they are enough, because the decisions they inform — which day, what goes on it, where to sit — do not need precision. They need somebody to have noticed.