KPI drift rarely starts with a mistake. An analyst builds a metric for one report and makes a sensible choice about a lookback window or an exclusion. Someone else copies it into another report and adjusts it slightly. A data vendor changes its methodology. The person who built the original leaves. A year later, four reports show four values for what everyone calls the same metric.
The symptoms are familiar: two teams present different figures for the same KPI in the same meeting, time is spent reconciling numbers instead of discussing them, and leaders stop trusting the reports.
In pharma commercial analytics, the usual suspects are metrics with many reasonable variations, such as new-to-brand prescriptions, market share, persistence, line of therapy and gross-to-net.
The fix is not to find the one correct definition, because different questions sometimes need different definitions. It is to write each definition down, give it a distinct name if it differs, assign an owner, version changes with an effective date, and link every report to the definition it uses. Our guide to commercial KPI definitions covers how.