some time tracking results
Brian Hicks, January 29, 2025
A while back I wrote (post 1, post 2) about how TagTime (and the then-called TinyPing) analyze time by assuming that each ping is worth 45 minutes, then getting a daily average and a 95% confidence interval. This can give you a pretty good idea of how you're spending your time, but I only did it for a simulated person with a perfect schedule.
If you haven't read about this before, here's the basic idea:
- The system randomly asks you what you're doing.
- It biases that random choice in a way that the long-term average time between pings is 45 minutes (or whatever you like)
- Eventually, you can get an idea of what a "normal" day looks like by doing a little math.
Anyway, let's do the analysis now that I've got a bit over a month of data.
What you'll see below:
- The "tag" is what I'm doing. This is a dot-separated hierarchy (e.g.
work.meeting.standup.) For analysis, I split that into three tags (work, work.meeting, and work.meeting.standup) so I can see the proportions in different categories. - The average daily time is the percentage representation of the tag in the data set times the number of minutes in the day. (E.g. if I had 100 pings, 30 of which were tagged
sleep, I'd have 30 / 100 * 1440 to get 432 minutes, or 7h12m.) - The margin of error is a 95% confidence interval of the average daily time, rounded to the nearest minute.
In this data, I've censored a few things—either because they felt too personal to share in a space like this or because they reveal the contents of work projects—but otherwise I've checked this against other accurate sources I have (e.g. my watch for sleep tracking) and it all seems to be accurate!
So, here are the top places I've been spending my time recently:
| Tag | Average Daily Time ± Margin of Error |
|---|
| Unknown | 7h17m ± 43m |
sleep | 6h58 ± 43m |
work | 2h56m ± 31m |
beeps | 48m ± 17m |
tv | 28m ± 13m |
lunch | 25m ± 12m |
work.meeting | 24m ± 12m |
breakfast | 16m ± 10m |
k8s | 14m ± 9m |
coffee | 12m ± 9m |
driving | 9m ± 8m |
dishes | 8m ± 7m |
There are a lot of improvements I could make here. For example, I've had a lot of time off recently due to holidays. If I applied that insight and re-analyzed work for only weekdays I worked, I'm sure I'd get more like 7.5 to 8 hours per day. Overall, though, I'm pretty happy with this level of insight!
If you'd like to try this for yourself, you can get the source or pre-built binaries on GitHub.