In 1975, British economist Charles Goodhart formulated an observation—originally about monetary policy, but one that turned out to be a universal rule for any management system: when a measure becomes a target, it ceases to be a good measure.
The logic is as follows. As long as a metric merely records what is happening, it remains reliable because no one tries to influence it directly. But the moment a metric is turned into a target and tied to performance evaluations, bonuses, or status, people's behavior changes: they begin optimizing the metric itself rather than the underlying phenomenon it was originally intended to reflect. The gap between the "indicator" and the "reality it measured" widens over time—sometimes imperceptibly, sometimes dramatically.
In HR, this happens regularly. If a recruiter's KPI is the number of closed vacancies, quick but low-quality hiring decisions start to emerge. If the metric for training effectiveness is the number of completed courses, employees click through courses without retaining anything. If the engagement metric is survey participation, people respond purely formally just to avoid ruining the department's numbers. The metric keeps growing. The real problem it was supposed to track does not get solved.
This works in an especially insidious way with metrics that are easy to measure but only indirectly related to what the company actually wants to achieve. The number of closed vacancies is not the same as the quality of hire. The number of completed training sessions is not the same as the real growth of competencies. Goodhart is precisely about this gap.
What helps:
using multiple metrics simultaneously rather than a single "primary" one—making it harder to optimize the system around a single point;
periodically changing or revising metrics if behavior around them becomes unnatural;
directly asking yourself not "is the metric growing?", but "is the real problem for which we introduced it actually being solved?"