Averages Hide the Problem

Every tool reports averages. Almost every finding in personal time data is in the variation the average removes. Organisations asking the same question at team level can compare their process with this accounting guide: define the purpose, make the record visible to the people in it, and interpret time data alongside finished work rather than on its own.

What the average conceals

Two hours a day of deep work sounds fine. It might be ten hours on Tuesday and nothing for four days, which is a completely different working life and calls for a different response.

Six hours a week of meetings averages to just over an hour a day. If they are all on Wednesday, Wednesday is gone, and that is the fact that matters.

Twenty minutes a day of interruption is unremarkable. Twenty interruptions on one afternoon is a destroyed afternoon.

The average is the same in each pair, and only one member of each pair is a problem.

What to look at instead

The distribution. How many days had a stretch over ninety minutes, and how many had none.

The worst day, not the mean day. Systems fail at their worst case.

The count of occurrences, separately from their duration. Five two-minute interruptions cost far more than one ten-minute one.

And the sequence: what clusters with what.

The longest-stretch measure

For anybody whose work needs concentration, the single most useful figure is the longest uninterrupted stretch per day.

It is easy to extract, it is not an average, and it corresponds to something people can feel.

Track it over a month and the trend tells you more than any category breakdown.

Weekly totals, and why they mislead too

A week is a short period and one unusual event dominates it.

Four weeks is the shortest span where a total means much; less than that and you are describing a particular fortnight.

Which is an argument for repeating short investigations over time rather than running one long one, and for comparing like periods.

The comparison that is fair

Yourself against yourself, over comparable periods.

Not against a published figure, a colleague, or an idea of what a week should contain.

Those comparisons are unavailable — nobody's figures are collected the same way — and they produce discouragement rather than information.

Its own note covers this, because it is the commonest way people make themselves miserable with this data.

Presenting it to yourself

One line per day is usually more informative than any chart: date, longest stretch, biggest single consumer, one word for how it went.

Thirty lines of that is a month you can read in two minutes and reason about.

A dashboard of averages is a month you glance at and learn nothing from, which is why most of them are abandoned.

What to check

Does your tool show you anything other than averages and totals?

Do you know your longest stretch for each day of last week?

Are you looking at the worst day or the mean day?

And could you say which single day this month was the worst, and why?

The figure to watch

For anybody whose work needs concentration, the longest uninterrupted stretch per day is the most useful single number.

It is not an average, it corresponds to something you can feel, and it responds to the changes worth making.

Worth keeping in mind

One line per day beats any chart: date, longest stretch, biggest single consumer, one word for how it went. Thirty of those is a month you can read in two minutes and reason about.

Weekly totals mislead too

A week is short and one unusual event dominates it.

Four weeks is the shortest span where a total means much. Less than that and you are describing a particular fortnight, which is an argument for repeating short investigations over time rather than running one long one.

In short

Look at the distribution rather than the mean: how many days had a stretch over ninety minutes, and how many had none.

Systems fail at their worst case, so the worst day is more informative than the average one.

The point

One line per day beats any dashboard: date, longest stretch, biggest single consumer, one word for how it went.

A month of that is readable in two minutes. For a complementary perspective on work, measurement or planning, consult Microsoft 365.