Measuring Performance Without Presenteeism

The anxiety behind most surveillance tooling is simple: a manager cannot see the work happening and does not know how else to tell whether it is.

The uncomfortable answer is that watching people work never told you much either. It told you they were present.

Why activity metrics fail

Keystrokes, mouse movement, hours logged in, messages sent, green status indicators. All are measurable, all correlate weakly with value, and all are trivially gamed.

Worse, they are gamed most effectively by the people you least want to keep, while the people doing the difficult work — thinking, reading, talking to a customer, deleting code — score badly.

The predictable outcome is a team optimising for the appearance of activity. That is not a hypothetical risk; it is the normal result.

The alternative is not complicated

Agree what the person is responsible for. Not tasks — outcomes. What should be true in three months that is not true now.

Agree how you will both know. A number where a number exists, a demonstrable artefact where it does not, a specific judgement by named people where neither applies.

Check at a defined cadence. Monthly is usually right. Weekly turns into surveillance by another route; quarterly is too slow to intervene.

Look at the trajectory, not the week. Everyone has bad weeks; remote work only makes yours less visible.

This is ordinary management. Its absence is what makes surveillance feel necessary.

When there is no clean output measure

Plenty of valuable work does not produce a countable artefact: support, coordination, unblocking others, maintaining something that consequently does not break.

For these, the honest instrument is structured judgement rather than a metric.

Ask the people who depend on the person. Not "is she good" but specific questions: what happened when you needed something from her, what would be harder without her, where did you have to work around her.

Three such conversations tell you more than any dashboard, and they are harder to game because they are not announced in advance.

Distinguish output from availability

Both are legitimate expectations. Confusing them causes most of the trouble.

Output is what the person produces. Availability is being reachable during agreed hours so colleagues are not blocked.

Someone can have excellent output and unacceptable availability, and that is a real problem worth naming — but name it as an availability problem, not by inventing doubts about their productivity.

Set availability expectations explicitly and separately: which hours, what response time, how urgent matters reach them.

Signals worth watching

Not as metrics, as prompts to ask a question.

Work slowing without explanation. Could be blocked, could be overloaded, could be disengaged. Ask.

Withdrawal from conversation. Frequently the earliest sign of a problem.

Rising response times. Sometimes a workload issue, sometimes avoidance.

Someone becoming a bottleneck for others. Usually structural, and usually invisible until it is severe.

Each of these is a reason for a conversation, not a conclusion.

What monitoring software actually costs

Beyond the effects on trust, which are real and well documented, there is a practical objection: it produces data that does not answer the question.

You will learn that someone was active for six hours. You will not learn whether those hours produced anything worth having. And you will have taught the team that you do not know how to tell — which is the thing they will remember.

If you cannot assess someone's contribution without watching them, the gap is in how the work is defined, not in your visibility.