The reason an unmeasured task disappears

A support team is given a first-response-time target. Response time improves for eleven straight months. Over the same period, the internal notes that used to explain unusual cases to the next person thin out and then stop, and nobody notices until an escalation arrives with no history attached.

What everyone sees

Measurement is treated as neutral observation, a lens pointed at part of the work. It is not neutral, because the observed and unobserved tasks draw on one finite supply of effort. Adding a measure to one task changes the relative return on every task that competes with it, including tasks nobody intended to deprioritise.

What is actually happening

Holmström and Milgrom’s multitask analysis shows that when an agent divides effort between tasks of unequal measurability, strengthening the incentive on the measurable task reduces effort on the others — and that the optimal contract may therefore involve weaker incentives than the measurable task alone would justify, precisely to protect what cannot be counted. Kerr’s survey of reward systems documents the same pattern across organisations, with stated goals routinely undermined by the metrics installed to pursue them. The modern version is faster because dashboards make the measured task continuously visible: a number that updates hourly does not merely rank higher than an unrecorded task, it is the only one of the two that exists during the working day.

Why it stays hidden

The loss hides because it produces no signal until much later. Unmeasured work has no series to fall, no threshold to breach and no owner to report it, so its disappearance is invisible in the period it happens. When the consequence surfaces, the metric that caused it is showing an unbroken record of improvement and is not a suspect.