0.2 The interruption log

The log is three working days. Not a week, because a week of logging degrades by Wednesday, and not one day, because one day is always atypical for reasons you will find persuasive.

For each interruption, record five fields. Time it arrived. Where it came from, named. What it displaced. How long you spent on it. Whether you returned to what you were doing, and how much later.

Then, at the end of each day, classify each entry into one of five categories. These come from the source material for this course and they are chosen because each one implies a different action.

Necessary. It had to reach you and it had to reach you then.
Useful but poorly timed. The content was legitimate, the arrival was not.
Unnecessary. Someone else could have handled it, or it did not need handling.
Avoidable. A predictable consequence of an arrangement that could be changed.
Self created. You opened it. Nobody sent it.

The fifth category is the one people leave empty, and the logged evidence suggests it should be the fullest. In a study that instrumented 32 people's machines for five working days, participants averaged 566 application switches per day, with a range from 183 to 1,035. Email was visited an average of 74 times a day, with a maximum of 373, and the average visit lasted 32 seconds (Mark et al., 2015).

Seventy four visits of half a minute each is about 40 minutes of email per day, distributed as an interruption that arrives 74 times. No colleague sent most of those. They were opened.

The field that makes the log worth keeping

"What it displaced" is the only column that turns a list of events into an argument. An interruption with no displacement cost is not a problem. An interruption that displaced the only ninety minutes available that week for the thing you say matters most is a decision someone made, and now you know who.

There is a predictable failure in the third category. "Unnecessary" is the judgement most people apply to interruptions from below them and least often to interruptions from above them, and the log will show this bias clearly if you record the source by name.

When you find that pattern in your own log, resist the first explanation, which is that senior people simply have more legitimate reasons to interrupt. Sometimes true. Also exactly what the pattern would look like if you were classifying by status rather than by content, and you cannot tell the two apart from inside.

The test is mechanical. Take three interruptions classified necessary that came from someone senior, and three classified unnecessary that came from someone junior, and swap the senders. Read the content again. If the classifications move, you were reading the sender.