The journey
Door to door is the only honest measure
Scheduled journey times understate the real one consistently and in one direction. How to measure it, and the three parts that always go missing.
Days this part is about, hour by hour
Tuesday, the first week
1 of 7One hour that had been going round in writing for a fortnight
Tuesday, the fifth week
0 of 7Wednesday, the fifth week
2 of 7three days · 21 hours logged · three of them were only possible there · one of the three days contained no such hour at all · three were made worse by being there
Three days, two of which repaid the journey and one of which did not.
A team evaluating employee productivity insights for flexible teams should define purpose, access, and interpretation first, then use the data to improve planning rather than reward visible busyness.
An independent source worth checking for this subject is Federal Trade Commission.
Ask somebody how long their commute is and you will usually get the train time. Ask them to measure it for a fortnight and the number goes up by somewhere between fifteen and thirty per cent.
The gap is not vagueness. It is systematic, it runs one way, and it is made of three specific things.
The three missing parts
The ends. The walk to the station, the walk at the other end, the lift, the badge, the stairs. Each is small and none is in anybody's stated number. On this record they came to nineteen minutes a day.
The buffer. The margin left so as not to miss the train, plus the wait when the margin was not needed. Ten minutes on average, every journey, by construction.
The failures. Not the dramatic ones. The ordinary eight-minute delay that happens on about one journey in five, the bus that was full, the platform change. Averaged across the quarter they came to six minutes a day.
Those three together were thirty-five minutes a day on a journey people would describe as "about forty minutes on the train".
How to measure it
Two timestamps a day, for ten working days. The moment you leave the house and the moment you get back. Nothing else.
| Day | Out | Back | Total |
|---|---|---|---|
| Tue | 07:35 | 18:15 | 2h 10m |
| Wed | 07:30 | 18:20 | 2h 10m |
| Thu | 07:40 | 18:10 | 2h 05m |
| Tue | 07:35 | 19:05 | 2h 40m |
Two columns of times. Nothing in this needs an app, and nothing in it needs to be accurate to the minute: the whole point is that the error in the usual estimate is half an hour, so five-minute precision is ample.
The one thing to be careful about
Do not fold the bad day into the average and then quote the average as typical. The Tuesday with the engineering works was forty minutes longer than the others, and it is honest to report the median and the worst case separately rather than a mean that is neither.
On the record here: a median of two hours and ten minutes, a worst case of two hours and forty, and a quarter total of seventy-three hours and forty minutes. Three numbers, each true, and each useful for a different sentence.
Why the precision matters
Because the conversation about hybrid working is full of soft numbers, and a soft number can be argued with indefinitely. "It takes me ages to get in" invites a discussion about whether it really does.
"Two hours and ten minutes door to door, median over ten days" does not. It is a measurement, it was made the same way anybody else would make it, and it moves the conversation on to the thing that actually matters, which is what the journey bought.
Why the ends are bigger than they feel
Nineteen minutes a day across the quarter — about eleven hours — in walks of four to seven minutes each, which is exactly the length that does not register as a journey.
The same is true of the buffer. Nobody experiences arriving at a platform eight minutes early as time spent travelling; it registers as being organised. Across thirty-four days it was nearly six hours.
That is most of why stated commute times are wrong in a consistent direction. They are reports of the part of the journey that feels like travelling, and the parts that do not feel like travelling turn out to be about a quarter of it.
The one-week version
If ten days is too much, five will do, with one adjustment: include a day you would normally describe as a bad journey. A week of good journeys produces a median that is lower than the arrangement actually costs, and the point of the exercise is to stop having a number that is systematically wrong.
Report it as two figures rather than one. "Two hours ten typically, two hours forty when the trains are doing something" is more useful and more honest than any single number, and it is harder to argue with because it has already conceded the variation.
Once the figure exists
Say it out loud once, in a team setting, alongside your own.
On the team recorded here the journeys ranged from eighteen minutes to one hour and twenty-five, the policy had been discussed three times, and nobody knew anybody else's. The person at the far end assumed they were near the middle.
That ignorance is doing real work. A rule that costs one person an hour and fifty minutes a week and another eight and a half looks uniform only from inside the first figure, and the people writing the rule are reliably nearer the short end, because the short end is where the office is.