What to ask for
The fortnight that would settle it
Ten days, one letter an hour, two timestamps a day — and the four numbers you will have at the end of it.
Days this part is about, hour by hour
Wednesday, the second week
3 of 7Planning day
Wednesday, the fourth week
3 of 7The new starter's first day
Thursday, the fifth week
1 of 7three days · 21 hours logged · seven of them were only possible there · one was made worse by being there
The three days worth asking for more of.
Before adding another fixed office day, teams can take a closer look at the product to check whether capacity, coordination, or project design is the real constraint.
For a separate perspective on technology, privacy, or flexible work, consult Remote.
Twelve of the thirty-four days, hour by hour
Tuesday, the first week
1 of 7Wednesday, the first week
2 of 7Thursday, the first week
0 of 7Tuesday, the second week
0 of 7Wednesday, the second week
3 of 7Planning day
Monday, the third week
0 of 7Wednesday, the third week
1 of 7Thursday, the third week
0 of 7Tuesday, the fourth week
2 of 7Wednesday, the fourth week
3 of 7The new starter's first day
Tuesday, the fifth week
0 of 7Wednesday, the fifth week
2 of 7twelve days · 84 hours logged · fourteen of them were only possible there · five of the twelve days contained no such hour at all · eleven were made worse by being there
Twelve of the thirty-four days, chosen to cover the range.
The argument about hybrid working has been running for several years and most of it is conducted without data on either side. Both parties are reasoning from memory, and memory on this subject is distorted in a known direction: the hours that justified the journey are vivid and the ones that did not are gone by the following month.
A fortnight of counting ends that, for you, for about three minutes a day.
The whole method
One letter at the end of each hour, with four words so you can recognise it later:
n if it could only have happened here, b if it was possible
anywhere and better here, s if it was identical anywhere, w if the
building made it worse.
Two timestamps a day: when you left the house, when you got back.
Ten working days, including the days at home. Then stop.
Nothing else. No ratings, no reflection, no app. Anything larger does not get finished, and an unfinished record answers nothing.
What you will have
Four numbers that almost nobody has about their own working life:
- How many hours actually needed the building.
- Which weekday they fell on.
- Your real travel time, door to door, which is reliably longer than the number you would have given.
- How large the fourth column is.
And one sentence that fits in a conversation: seventeen of my thirty-one useful hours were Wednesdays, and I am in on Tuesdays, Wednesdays and Thursdays.
What it will not tell you
It will not tell you whether three days is the right number for your organisation. One person's fortnight cannot settle that and nothing on this site claims it can.
It will not tell you anything about anybody else. A manager's
distribution differs from an engineer's; somebody four months into a
job will have far more n hours than somebody four years in, for good
reasons that are about them and not about the office.
And it will not tell you what the office is worth in the things that do not come in hours — the sense of a place, knowing who people are, the reason a Tuesday in an empty building still feels different from a Tuesday at the kitchen table. Those are real and this method does not reach them. A count of hours is a narrow instrument and it should be read as one.
Why it is still worth doing
Because the narrow thing it does measure is the thing the whole arrangement is justified by, and currently nobody on either side of the argument has a figure for it.
On the quarter recorded throughout this site the figure turned out to be thirty-one hours out of two hundred and thirty-eight, concentrated on one weekday, bought with seventy-three hours and forty minutes of travel. None of that was knowable in advance and all of it was knowable in a fortnight.
Whatever your numbers are, they will be yours, and they will be the first thing in the discussion that neither side has to take on trust.
The three objections, answered honestly
"A fortnight is too short." It is short. It is long enough to see the distribution, which is the output, and it is short enough to be finished — and a finished short record beats an abandoned long one by an enormous margin. If a fortnight shows something surprising, run another one in a different month before acting on it.
"The classification is subjective." Partly. The boundary between
b and s involves judgement and different people will draw it
differently. The boundary that matters — between n and everything
else — is much harder to get wrong, because the question is whether the
hour was possible elsewhere, which is usually a fact rather than an
opinion. Report n and travel; treat the rest as texture.
"One person proves nothing." Correct, and nothing here claims otherwise. What one person's record does is replace that person's guess with that person's measurement, and supply a number to a discussion currently conducted entirely without one. If it prompts two colleagues to keep their own, the objection stops applying.
What to do with the result if it is boring
It might be. You may find your days are well chosen, the overlap is good, and the ratio is around one. That is a real result and it is worth having.
It means the arrangement is working for you, which is useful to know with confidence rather than by assumption — and it means that when somebody else on your team says theirs is not working, you have a basis for believing them rather than assuming they are describing the same quarter you had.
The last thing
The quarter described throughout this site was not a disaster. Nobody behaved badly, no policy was unreasonable on its face, compliance was ninety-three per cent, and several days were plainly worth the journey.
What it was, was unexamined. Thirty-four journeys were made because a document said three days, fifteen of them returned nothing, and the information that would have fixed most of it existed the whole time in nine separate heads and was never once assembled.
That is the ordinary case. It is not a failure of anybody's effort or anybody's intentions, and it will not be corrected by either, because the thing that is missing is not effort. It is a count of something, kept by somebody, for two weeks.