Why guesses are always wrong
Ask anyone where their week went and you will get an answer built from memory, which systematically over-weights whatever was memorable and under-weights whatever was frequent.
Nobody remembers the eleven small interruptions. Everyone remembers the two-hour meeting. That bias is why teams cut the wrong thing: they cancel the meeting people complain about and leave untouched the daily half-hour of copying information between systems that nobody has ever mentioned because it feels like just part of the job.
Whatever your own numbers turn out to be, the shape is usually the same: the expensive thing is invisible because it is distributed in small pieces across every day.
How to run it in one week
Say what it is for, and what it is not for
Tell the team you are looking at how the team's time is spent, that individual sheets will not be reviewed, and that nothing here feeds into performance.
Then hold to that absolutely. The moment someone's audit data is quoted back at them, you have lost the ability to run an honest one ever again.
Say it in writingGive them eight categories, no more
Fixed list, agreed in advance. Long lists produce inconsistent categorisation, which makes the totals meaningless.
Include an "other" category and watch it. If other exceeds about 15%, your categories are wrong and worth revising for the next run.
Eight, plus otherLog in half-hour blocks, twice a day
Not real-time tracking. Two minutes after lunch and two minutes at the end of the day, filling in rough blocks.
Precision is not the goal and chasing it destroys participation. You want to know whether admin is 5% or 25%, not whether it was 18% or 19%.
Two minutes, twice a dayLog interruptions as a simple count
A tally each time work is broken by something unplanned. No categories, no notes, just a count.
This number surprises people more than any other, and it is the one that explains why a week with only six hours of meetings still felt fragmented.
A tally, not a descriptionAggregate and share the whole picture
Totals by category across the team, with no names. Share it with everyone who took part, including the parts that reflect badly on management decisions.
Sharing the full result is what earns participation next time. Collecting data and going quiet reads as surveillance, whatever you promised.
Share everything, name nobody
The categories to use
Adapt these to your business, but keep the count at eight and keep them mutually exclusive.
| Category | What belongs in it |
|---|---|
| Core work | The thing this role exists to do |
| Meetings | Any scheduled call, including 1:1s |
| Admin and data entry | Updating systems, copying between tools, filing |
| Communication | Email, chat, writing updates |
| Rework | Redoing something because of an error or a change |
| Waiting or blocked | Time lost waiting on someone else |
| Learning | Reading, training, figuring something out |
| Other | Watch this. Over 15% means the categories are wrong |
Rework and waiting are the two most people leave out, and they are usually where the uncomfortable findings live. A team spending 12% of its time blocked has a coordination problem that no amount of individual productivity advice will touch.
Reading the results
Look for four things, in this order.
The gap between belief and reality. Ask everyone to guess the split before you show them the totals. The categories where the guess is furthest off are where the leverage is, because those are the ones nobody has thought about.
Admin as a share of the whole. Anything over 15% for a role that is not an admin role points at a tooling or process problem rather than a discipline one.
Rework. Usually caused by unclear requirements or missing context at handover, not by carelessness. High rework is a specification problem.
Blocked time. If this is significant, it is almost always a dependency or approvals bottleneck, and it is the cheapest thing on the list to fix.
Do not over-interpret one week. This is a diagnostic that points you at two or three places worth investigating, not a measurement system.
Acting on it without causing damage
Pick two things. Not eight. A time audit that produces a list of twelve improvements produces none of them, for the same reason a retrospective with twelve actions does.
Fix the process, never the person. If admin is high, the answer is fewer places to enter the same information, not asking people to be faster at data entry.
Say what you changed, and why. Explicitly connecting a change back to the audit is what makes the next one worth running. Silence after a data-collection exercise is how these die.
Where the results point at meetings, our guides to running a meeting people don't resent and deciding if a meeting should be an email are the practical follow-ons. Where they point at repetitive admin, what to automate first covers which of it is safe to remove and which is not. And where the finding is that the same information is being entered in three places, that is the specific gap Siela is built to close, by keeping CRM, meetings, tasks, and team work on one shared record rather than four.
Frequently asked questions
- How long should a time audit run?
- One normal week. Two weeks produces better data and much worse participation. If the week you pick is unusual, a holiday period or a launch, pick a different one rather than accepting distorted results.
- Will people be honest in a time audit?
- Only if they are certain it is not a performance measure. Say explicitly that individual results will not be reviewed and that only aggregate categories will be discussed, then hold to it. One breach of that promise makes every future audit worthless.
- Should the audit be automated with tracking software?
- For a one-week diagnostic, no. Automated tracking measures which application was open, which is not the same as what someone was doing, and installing monitoring software changes the relationship with your team far more than the data is worth.
- What if the results show someone is underperforming?
- Do not use audit data for that. You promised otherwise, and one week of self-reported categories is weak evidence anyway. If there is a performance concern, handle it separately and on its own evidence.
- How often should you repeat it?
- Once or twice a year, or after a significant change in how the team works. Repeating it quarterly turns a diagnostic into a ritual, and participation quality drops each time.