The resource you are actually spending

The first false picture to remove is that the problem is time.

Time is fixed and visible. Everyone in your organisation has the same amount of it and most people can tell you roughly where theirs went. Attention is neither fixed nor visible, and it is the thing that actually gets consumed by an interruption.

The direct observational evidence is unusually specific. Twenty four information workers, seven of them managers, were shadowed in situ for an average of 25 hours and 42 minutes each. They spent an average of 11 minutes and 4 seconds working on one thing before switching to another. More than half of those working segments, 57.1 per cent, were interrupted rather than completed. When an interrupted task was resumed the same day, the average delay before returning to it was 25 minutes and 26 seconds, with an average of 2.26 other tasks in between (Mark, González & Harris, 2005).

That is a sample of 24 people in one workplace type, and it is descriptive rather than causal. Treat it as a well measured picture of a particular kind of office, not as a law.

The figure you have probably heard, that it takes 23 minutes and 15 seconds to refocus, is not in that study or in any peer-reviewed paper by the researcher it is attributed to. It appears to have entered circulation through a 2006 interview. The published numbers are 25 minutes 26 seconds for same-day resumption and 20.31 to 22.77 minutes for total task completion in a later experiment. This matters less because the numbers differ than because you have almost certainly repeated a figure to a colleague that has no paper behind it.

There is a second finding that reverses the intuition. In a controlled experiment with 48 participants, interrupted people did not work more slowly. They finished faster: 20.31 and 20.60 minutes against 22.77 minutes uninterrupted. What rose instead was stress, from 6.92 to 9.46, frustration, from 4.73 to 6.63, and effort, from 9.50 to 11.04 (Mark, Gudith & Klocke, 2008).

Interruption does not usually show up as work not done. It shows up as the same work done under compression, by someone who is more depleted at 17:00 than the output would suggest. That is why the cost is invisible in a status report and obvious in a person.

The compression finding has a specific implication for how you will be judged.

If your team absorbs interruption by working faster, the interruption is free to the organisation and expensive to the individual. Nothing in the delivery record shows a problem. The signal arrives later and in a different channel: someone becomes short with a colleague, or stops volunteering, or leaves.

This is the first case of a pattern that runs through the whole course. The costs of poor leadership are almost never charged to the account where they were incurred. They are charged late, to a different line, and by then the causal chain is long enough that nobody argues about it because nobody can trace it.

The practical consequence for Level 0 is that you cannot use output as your measurement instrument. If you could, you would already know. You need a record of the inputs.