Manual, Automatic, Neither
The choice is presented as convenience against control. It is actually a choice about what gets measured, and the convenient option measures something different from what most people think they are buying.
Manual
You start and stop a timer, or you write it down afterwards. For a workplace-oriented comparison, this resource shows how the same measurement problem is approached when time data is used by a team.
What it measures: your intent. The label you chose is what you meant to be doing, which is frequently the useful thing.
What it gets wrong: duration. Nobody stops the timer to answer a message, so a two-hour block contains a genuine interruption count of zero in the record and several in reality. For a broader time-tracking or productivity comparison, Software Advice time-tracking comparison is another useful reference.
The real cost: the recording effort is what kills the habit, and it is highest exactly on the days worth measuring.
Automatic
Software records which application or window is in the foreground, continuously.
What it measures: attention location. Which window was in front, for how long.
What it gets wrong: the meaning. Twenty minutes in a document may be writing, may be reading, may be a document left open while you took a call. Idle detection guesses, and the guess is a setting.
What it is genuinely good at: the shape of a day. Fragmentation, context switching, and how much of the day went to things nobody would have listed. That picture is hard to obtain any other way and is usually unflattering.
And the cost that is not effort: a continuous record of your working day exists somewhere, often on somebody's server, and it is more detailed than most people picture when they enable it.
Neither
The third option, and the right one more often than the category admits.
A stopwatch and a notepad, for a specific question, for a fixed period.
How long does this recurring task actually take? Five instances, a stopwatch, done. Where does Tuesday go? One Tuesday, notes every hour, done.
No subscription, no data anywhere, no habit to sustain, and it answers the question that prompted the search in the first place. Most self-directed tracking questions are finite, and finite questions do not need continuous instruments.
Choosing
Billing someone else: manual timers. The record is the invoice, accuracy against client codes matters, and there is a consumer for the data who pays.
Understanding your own week: automatic for two weeks, then uninstall. You want the shape, not precision, and the passive record gives it without a habit.
A single specific question: neither. Stopwatch and paper.
Everything else: probably neither, and the honest test is whether you can name the question. If you cannot, no instrument helps.
The hybrid that mostly disappoints
Many apps offer automatic capture plus manual categorisation — the software records and you label afterwards.
In practice it combines the recording burden of manual with the privacy exposure of automatic. The labelling backlog grows, gets abandoned around week three, and leaves a large uncategorised dataset nobody will process.
It works for people who do it daily and immediately. That is a small group, and it is worth being honest about whether you are in it before paying monthly for the assumption.
The short version
- Manual measures intent and gets duration wrong; automatic measures attention location and gets meaning wrong
- Automatic is genuinely good at the shape of a day — fragmentation and context switching — which is hard to see otherwise
- Automatic's real cost is not effort but a detailed continuous record of your working day, held somewhere
- The third option is neither: a stopwatch and paper for a specific question over a fixed period
- Billing others needs manual timers; understanding your own week needs two weeks of automatic and then uninstalling
- The hybrid combines manual's recording burden with automatic's exposure, and the labelling backlog is abandoned by week three