Metrics turn goals into numbers, and the choice of number shapes what a team pays attention to.
A bad metric can make a team work hard on the wrong thing.
Defining one carefully is often the most valuable part of an analysis.
A good definition states the numerator, the denominator, the time window and any exclusions.
For example, conversion rate might be orders divided by unique visitors per day, excluding staff and bots.
Rates and averages per user are usually more informative than raw totals, which grow with size alone.
A metric should be sensitive enough to move when things change and stable enough not to swing on noise.
It should be hard to game, or it will improve without the goal improving.
Leading metrics move early and predict later results, such as trial sign-ups predicting revenue.
Lagging metrics, such as yearly revenue, confirm results but arrive too late to steer by.
Counter metrics, tracked alongside, catch harm that the main metric would miss.
Changing a metric's definition breaks comparisons with the past, so changes should be rare and recorded.
Writing definitions in one shared place stops teams calculating the same metric in different ways.
In practice, a metric that no decision depends on is usually not worth tracking.
The common mistake is choosing whatever is easy to count rather than what reflects the goal.