Activity Scores
Monitoring tools produce a single figure — a productivity percentage, an activity score, an engagement index. It is worth knowing precisely what goes into it. For a related reference, see further details.
What the number is made of
Keyboard and mouse events per interval. Whether the input device moved.
Application classification. Somebody decided which applications count as productive, usually by category and usually by default.
Idle detection. A threshold, typically a few minutes, after which the interval is scored as inactive.
Sometimes window titles and URLs, classified the same way.
None of these is a measure of work. They are measures of input device activity while a categorised application is in focus.
What this misses by construction
Thinking. Reading a specification, working out an approach, sitting with a hard problem. Scored as idle.
Talking to somebody. The most valuable ten minutes of a difficult week frequently produce no keystrokes.
Anything on paper. Sketching, calculating, reading a printed document.
And anything on a device the tool does not see. The same blind spots recur in every monitoring product.
The work that scores highest is repetitive keyboard work, which is also the work most likely to be automated.
How it is gamed, immediately
Not maliciously — reflexively, by people protecting themselves.
That is the score changing behaviour, not measuring it.
Mouse movement is trivially maintained. Any organisation deploying activity scoring produces, within weeks, an informal culture of knowing what registers.
And the honest are penalised. Somebody who steps away to think shows as idle; somebody who reads the news with a document open does not.
What a score is legitimately for
Being fair, because the category is not worthless.
Aggregate capacity planning, where individual accuracy does not matter and the trend across a department is the question.
Detecting a machine problem — a workstation that shows as idle because software is failing.
And in a small number of regulated contexts, where activity logging is a compliance requirement rather than a management choice.
In none of these is the individual score used to judge a person, which is the use it is sold for.
What to use instead
Whether the work arrived. Output at whatever granularity the work has.
Whether commitments were met, which is the reliability question and needs no instrumentation.
And a conversation, which is the tool that detects a struggling person about a month before an activity score does.
The short version
- An activity score is built from input device events, application categories set by default, and an idle threshold
- It misses thinking, conversation, paper work and anything on an unmonitored device by construction
- The work that scores highest is repetitive keyboard work, which is also the most automatable
- 24% report taking fewer breaks to avoid appearing idle — the score changing behaviour rather than measuring it
- Legitimate uses are aggregate capacity planning, detecting machine faults, and specific regulated logging — none at individual level
- Use whether the work arrived, whether commitments were met, and a conversation, which detects a struggling person a month earlier
For additional context on this topic, see GIMP.