Measurement perturbs what it measures. In physical systems that is a technical nuisance. In systems made of people it is the central fact, and it is the one thing a monitoring tool is structurally unable to report.
Effort moves toward whatever is legible
People respond to being observed by producing observable things. This is not cynicism and it does not require anyone to be trying to deceive: if part of your work is visible to the people who decide about you and part is not, attention shifts toward the visible part without any decision being made.
The specific movements are consistent. More time in tools that record activity, less in ones that do not. Faster responses and shallower ones, because responsiveness is measurable and thinking is not. More time spent documenting what one did, which is work about work.
The things that shrink are the ones nobody wants to lose
Asking for help becomes expensive, because a person asking looks stalled and the colleague helping looks idle. Two people solving a problem together registers as two people producing nothing, which is the gap described in the article on what cannot be measured, now with a consequence attached.
Exploration shrinks for the same reason. An hour spent finding out whether an approach is viable produces no artefact if the answer is no, and under observation a no-artefact hour is a liability. So people choose the legible task over the valuable one, and they choose it quietly and reasonably.
Monitoring does not usually make people work less. It makes them work more visibly, which is not the same thing and is occasionally the opposite.
Presence displaces substance
Where the measure is availability, the behaviour it produces is availability: staying online, answering immediately, keeping the status green. Each of those is a small interruption to somebody, and a workforce optimising for responsiveness interrupts itself continuously.
The organisation then observes fragmented attention and low output on substantial work, and frequently responds by measuring more.
Where transparent tracking genuinely helps
The honest counterweight, and it is real. For hourly and field work, a documented record protects the worker as much as the employer. It settles disputes about hours worked, supports claims for unpaid time, and removes the situation where a person's account of their own week is weighed against a manager's impression.
Workers in those settings frequently prefer having the record, and opposition to monitoring as a category tends to come from people whose work is not paid by the hour. The distinction that matters is between a record that documents an agreed exchange and a system that scores somebody's character, and the same product can be configured as either.
The effect is strongest on the people you least want to lose
Reactivity is not uniform. Someone whose work is routine, well defined and already visible loses very little when it is recorded, because the record matches what they were doing anyway.
Someone whose contribution is judgement, mentoring and knowing which problem is worth solving is the person whose work the system reads worst, as the article on activity sets out. They are also the person with the most alternatives, and the one for whom the implied estimate is most insulting. A measure that misrepresents seniority most severely is applied to the people most able to leave, which is a staffing consequence of a monitoring decision and appears in neither budget.
Trust is spent, not consumed
Introducing monitoring says something whether or not anything is said. It communicates an estimate of the workforce, and people adjust their own contribution to match the estimate they were given. The discretionary part of the work, the part nobody could require, is the part that responds.
That effect is slow, invisible to the tool, and mostly irreversible within the tenure of the manager who caused it.
What to do about it, since the tool cannot tell you
Ask the people. The reactivity effect is not measurable by the system producing it, so the only available instrument is asking, and asking in a form where an honest answer is safe. A short anonymous question after three months, repeated, produces information no dashboard contains.
Keep measurement at the level of the group wherever the question is about a process, which is nearly always. Give people their own data. State the purpose narrowly and publicly.
And write down the condition under which it stops. A programme with no stated exit does not have one, and the article on what cannot be measured explains why: the tool will never report that it is not worth having, because its costs are not among the things it collects.
What we cannot verify
Effects of observation on work are studied across organisational research and the findings are contested, vary by context, and are not reducible to a number we would publish; the direction described here is better supported than any magnitude. The illustration in this article is labelled as illustrative for that reason. Vendor claims that a product improves morale or engagement come from the vendor. What is happening in your organisation is answerable only by asking the people in it.
The short version
- Measurement perturbs the measured, and with people that is the central fact.
- Attention shifts toward the visible part of the work without anyone deciding.
- Asking for help and exploring an idea both register as producing nothing.
- Monitoring usually produces more visible work rather than more work.
- For hourly and field work a documented record protects the worker too.
- Write down the condition under which it stops, because the tool will never say.