What the travel buys, and what it does not

What it buys

Unscheduled minutes with specific people. Nineteen of the thirty-one hours that needed the building had nothing in the calendar. This is the real product and it is bought by proximity to the right person, not by proximity to the building.

A surface for arguments with a shape. A wall, a table, eight people looking at the same thing. Three hours of the planning day did in one morning what a fortnight of writing had not.

Access to physical things. Hardware, printed material, a machine that is in one place. Small on this record — about four hours — and absolute when it applies.

A boundary. For a number of people, leaving the house is what makes the working day end. This is a real purchase and it is usually left out of the formal case because it sounds personal rather than operational.

What it is assumed to buy and did not

Collaboration in general. Not as such. The forty-four hours that were better for being in the building were better by a margin, and nearly all of them involved people who were present that day. On the fifteen days when the relevant people were elsewhere, no amount of being in the building produced any of it.

Visibility that counts. The argument that being seen matters is not wrong, but on this record the days with the most people in were the days when nobody had time for anybody. Being seen in a crowded room on a Wednesday is not obviously more visible than being useful in writing on a Thursday.

Focus. Sometimes the opposite. Eleven hours of the quarter were deep work that plainly went worse in an open room than it would have gone elsewhere.

Telling which you are getting, before you set off

Four questions, answerable the evening before, each one worth about the same as the others:

1. Will at least two people I actually work with be there?
2. Is there one thing in the day that needs a room and a wall?
3. Is a room booked?
4. Is there anything physical that has to be handled?

Two yeses out of four and the day will almost certainly repay the trip. Zero yeses and it will almost certainly not — on this record, every one of the fifteen days that produced nothing scored zero or one.

That checklist is not an argument for skipping days. It is an argument for choosing which days, which is the decision almost everybody has and almost nobody makes deliberately.

The checklist, scored across the quarter

Eleven days met none of the four conditions and produced nothing, eleven times out of eleven. That is the most reliable relationship in the whole record.

Why this is not a licence to skip days

Two reasons, and the second is the serious one.

The unscheduled hour depends on people being in without a specific reason. Nineteen of the thirty-one hours had nothing in the calendar, which means they happened because somebody was there anyway. A team that only came in when the conditions were met would eliminate the category it was trying to protect.

And the conditions are only knowable approximately. "Will two people I work with be there" is a guess unless somebody has said, which is why the sharing of plans matters more than the checklist does.

The checklist is a tool for ranking days you are going in anyway, not for deciding whether to go. Used the first way it is worth a substantial fraction of the quarter. Used the second way it quietly destroys the thing it was measuring.

The one condition that dominates

If only one can be checked, check the first. People.

On this record, every day with two or more relevant colleagues present produced at least one n hour, and no day without them produced any, regardless of how the other three conditions scored. The room, the wall and the journey all modulate the value of a day. The people determine whether there is any.