What to Track, and What Not to Bother With

Category design is where most time tracking quietly fails. Too many categories and people pick approximately; too few and the data answers nothing; the wrong ones and it describes the list rather than the work.

Start from the question

The categories follow the purpose, and different purposes need different lists. See what time tracking is for.

For a vendor-side perspective, Monitask also has a page on employee attendance tracking software.

Billing needs client and matter.

Estimating needs work type, so you can compare like with like.

Finding where effort goes needs activity types — and this is the one people get wrong, because project categories cannot answer it.

Capacity needs very little detail.

Pick the primary purpose, design for it, and accept that the others are served roughly.

How many

Five to nine. Above that, accuracy falls faster than the extra detail is worth.

The test: can someone choose the right one in under two seconds, without thinking? If not, they will pick approximately and consistently wrongly.

Fewer categories used accurately beat more used approximately, every time.

And you can always split later if the data shows one category is doing too much work. Starting broad and narrowing is easier than the reverse.

The categories that must exist

The ones that reveal the useful findings, and the ones most often missing.

Waiting or blocked. Time lost to someone else, an approval, a dependency. The single most valuable category in most datasets, and the first to be omitted.

Rework. Doing something again. Separating this from original work is what makes quality problems visible.

Admin. Invoicing, expenses, timesheets themselves, tooling.

Meetings, if they are a meaningful share.

Interruptions or unplanned work, in environments where that is significant.

Learning. Training, reading, working out how something works.

Without these, all of it goes into project hours, which inflates project cost and hides the actionable part entirely.

The catch-all

Have one, and watch it.

Its size is a quality indicator for the whole dataset. If "other" holds 30% of hours, your category list does not match the work, and every other number is suspect.

Under 5% is healthy. Over 15% means redesign.

And look at what is in it periodically — the contents usually name the missing category.

What not to bother tracking

Minute-level precision. The recording method does not support it. Rounding to five or fifteen minutes loses nothing real. See why time data is almost always wrong.

Sub-tasks within a task, unless a specific question needs them.

Anything you will not look at. The test: name the report this category will appear in. If you cannot, drop it.

Breaks and lunch, unless compliance requires it.

Which application or website was used. That is monitoring, not timekeeping. See the line between timekeeping and surveillance.

And a second dimension "for analysis later". Two-dimensional category systems — project and activity and client and phase — halve compliance and produce the data nobody wanted.

Designing the list

With the people who will use it. They know what actually happens; a list designed by management or derived from the finance system describes something else. For broader independent background, see NIST Privacy Framework.

Use their words. If the team says "chasing", call it chasing.

Test with a real week. Ask two people to categorise last week retrospectively and note every time they hesitated. The hesitations are the design problems.

Then reduce. Whatever list you arrived at, it is probably too long.

Reviewing it

After the first month, which is when the mismatches are obvious.

Then every six to twelve months.

Look at: the catch-all size, categories with almost nothing in them, and categories with implausibly much.

Changing categories creates a discontinuity in the data. Mark the date, and do not compare across it without saying so. See reading time reports.

For individuals

Fewer still. Five is plenty.

Client or project, plus admin, marketing, learning, and one catch-all.

Record non-billable time, or your effective rate is fiction — this is the whole reason the categories matter solo. See time tracking for freelancers.

And drop any category you have not looked at in three months.

The short version

Categories follow the purpose, and different purposes need genuinely different lists.

Five to nine. If choosing takes more than two seconds, people pick approximately.

Waiting, rework and admin must exist — they are the most valuable and the most commonly missing.

Watch the catch-all. Over 15% means the list does not match the work, and every other number is suspect.

And do not add a second dimension for later analysis. It halves compliance and produces data nobody uses.