What actually counts as “activity” for time trackers

The raw signals a tracker can read

Strip away the branding and almost every desktop tracker samples the same handful of inputs: keystrokes, mouse clicks, cursor movement, scrolling and which window currently holds focus. Some also note the application or URL behind that window. That is the entire vocabulary. A tracker records that these events happened and roughly where they landed — it does not read the words you typed or understand what was on screen.

Because the raw inputs are shared, the activity percentage is fundamentally a measure of input coverage, not of thought or output. A second with a single keystroke counts the same as a second with thirty. This is the first thing to internalise: the metric was built from what a computer can cheaply observe, not from what a manager actually wants to know.

• Keystrokes — any key, regardless of what you type.
• Mouse movement, clicks and scrolling.
• Which window or application holds focus (on many trackers).
• Not the content of your input, and not what is on the screen.

How those signals become a percentage

Trackers slice your tracked time into fixed intervals — 10 minutes is the common choice — and, within each one, sample repeatedly for any input at all. The activity percentage for that interval is simply the share of sampled moments that contained input. The daily figure on the dashboard is the average of those intervals, each weighted the same.

Two consequences fall out of that design. First, evenly spread input scores higher than the same amount of input delivered in one burst, because coverage rewards presence across the whole interval. Second, one very quiet block — a long call, a walk to think through a problem — drags the daily average down more than most people expect, and a frantic hour later cannot fully undo it.

What never counts as activity

If an action produces no keyboard or mouse input, it is invisible to the tracker, no matter how much real work it represents. Reading a specification, watching a recorded call, reasoning through an architecture, listening on a video call — all of it generates near-zero input, so the tracker cannot tell deep concentration from an empty desk.

The same blind spot covers anything that happens off the tracked machine. Work on a second laptop, notes on paper, a whiteboard session or a phone call away from the desk simply do not exist as far as the percentage is concerned.

• Reading a spec, a long thread or documentation.
• A call where you mostly listen.
• Thinking, planning or sketching on paper.
• Work done on a second device or phone.

Where the big trackers differ

The raw signals are shared, but the details around them vary by brand, and those details are worth knowing before you compare two numbers. Interval length, the exact sampling rate and the threshold for counting a moment as "active" all differ from one tracker to the next, so a 70% in one tool is not automatically a 70% in another.

Some trackers add layers on top of the input signal. Several weigh which application or window held focus — DeskTime, for instance, generally classifies apps as productive or unproductive — while Upwork attaches activity segments to the screenshots in its Work Diary. Time Doctor is known to flag unusually regular input patterns, and tools like WebWork publish their own per-minute activity formula. When you are unsure of a specific mechanic, it is safest to assume the general model above and check the tracker's own documentation.

Why honest work often scores low

Because input density is a weak proxy for output, a genuinely productive day can land far below a repetitive one. A day of research, code review and meetings can be entirely real work and still sit near thirty percent, while an hour of mindless data entry sits near the top of the scale.

This is structural, not a sign of slacking. Designers reviewing mockups, engineers reading logs, writers staring at a paragraph and anyone on back-to-back calls all produce naturally low numbers. If a role is judged on the percentage alone, the people doing the most thinking are penalised the most — which is exactly backwards from what the number is meant to encourage.

What the number actually rewards

Read literally, the metric rewards input that is spread evenly through the interval rather than concentrated in one spike. A steady, slightly uneven rhythm — climbing during hands-on work, dipping for reading and calls — is what a real working day looks like, and it reads as a person.

A report pinned at or near 100% for hours, day after day, is the real anomaly. No human hand produces input in every sampled moment; real work has pauses built into it. Consistency at the very top of the scale draws more scrutiny than a healthy mid-range figure, which surprises people who assume higher is always safer.

How to read your own activity report

Open your own timeline before anyone else does and look at its shape across the day rather than the single headline number. Where are the dips, and do they line up with meetings, reading and breaks? A percentage becomes a story about a real day once you can see the intervals behind it.

Compare your number against your own quiet weeks, not against a colleague in a different kind of role. If a genuinely low day needs explaining, the interval view is your evidence — a dip that matches a two-hour design review explains itself the moment someone can see it.

Where Reaction fits

Reaction sits in one narrow place: the stretches when you step away from the keyboard but the session should stay alive — the coffee, the stretch, the school run. It keeps a natural, human-like level of presence within a range you set, using varied cursor motion, a living keyboard rhythm, occasional scrolling and window switches, so the line stays even instead of collapsing into a dead gap.

It starts only when the computer is genuinely idle and steps aside the instant you touch the mouse. It cannot make low-input work look like high-input work while you are actually at the desk doing it — that is a limit of the metric, not a feature. What it does is stop a normal break from reading as an absence. Use it within your agreements; see the acceptable-use note below.