What to record, and at what granularity
Too coarse and the data answers nothing. Too fine and it does not get recorded at all.
Every studio that starts recording time faces the same design decision, and most get it wrong in the same direction: they build a category structure detailed enough to answer every conceivable question, and within a month nobody is filling it in.
The minimum useful record
- Client — always. Without it you cannot tell which relationships are profitable.
- Project — always. This is what the estimate was made against.
- Phase or activity — a short fixed list. Design, build, revision, project management, admin.
- Billable or not — a single flag, and the one most often omitted.
- Duration, and when it happened.
Five fields. Anything beyond this needs to earn its place by answering a question you have actually asked. Some teams also evaluate employee PC activity tracking; that is a separate decision from choosing the right level of project-time detail.
Without it you know how long the project took and not how much of that you can charge for. The gap between those two numbers is the entire margin conversation.
Keep the activity list short
Five activity codes get used consistently. Twenty get used inconsistently, which is worse than none — the data looks detailed and is not comparable between people or projects.
The test for adding a category is whether a decision would change based on the answer. 'How much time goes into revision' changes how you quote. 'How much time goes into revision on Tuesdays' does not. For UK working-time context, GOV.UK guidance on calculating working hours explains how working hours are calculated.
Record in blocks, not to the minute
Fifteen-minute granularity is sufficient for almost every purpose and is far more likely to be recorded honestly than six-minute increments.
The precision of the underlying data is limited by human memory and attention anyway. Recording to the minute produces false precision and adds friction at the exact moment you are trying to reduce it.
Capture it during the work
This is the whole game. Time recorded as work happens is approximately accurate. Time reconstructed on Friday is a plausible narrative, and it is systematically wrong in the same direction — it rounds toward the estimate, because that is what the person remembers expecting.
The mechanism matters less than the timing, and the practical requirement is that starting a record takes one action. A notebook works. A spreadsheet open in a tab works. Timers built for the purpose — Monitask, Toggl Track, Clockify and others occupy this space — work because they start and stop against a named project in a single click and total it afterwards by client. Anything that takes three steps will be done retrospectively.
Record the unbillable too
The strong temptation is to record only client work, since that is what gets invoiced. It hides the thing you most need to know: what proportion of the week is actually sellable.
Proposals, admin, invoicing, learning, business development — recorded honestly, these produce the utilisation figure that every rate calculation depends on. A studio that does not know its utilisation is guessing at its own pricing.