What is the difference between vanity metrics and actionable metrics?

The dashboard looked great: 10,204 signups, all time, with a line climbing steadily up and to the right. The line nobody added sat next to it: revenue from those signups, a flat zero. Both numbers were real. Only one of them could have changed a decision, and it was not on the dashboard.

This is the vanity metric problem in one screen. The numbers that feel best predict the least, and there is a fast way to tell the two kinds apart.

Two kinds of numbers

Both are real data. Only one can change a decision.

A metric that cannot go down cannot warn you. Its only function is to be presented. That is the tell in the story above: cumulative signups climbed every single week while revenue sat at zero, so the number of decisions changed by that climbing line was exactly zero.

The 3-question test

Run any metric through these. Three noes = decoration.

1. Decision? If it dropped 20% Monday, would anyone do anything differently?
2. Denominator? 10k signups out of how many visitors? A count without a base is a headline.
3. Owner? Who is paged when it moves? No owner means no action, means vanity.

The test comes from Lean Analytics. It has aged better than most dashboards.

Notice that the test never asks whether the number is important-sounding or whether leadership likes seeing it. It only asks whether the number is wired to a decision, a base and a person. A metric can fail all three and still be true; it just cannot be useful.

The translation table

Every vanity metric has an actionable cousin. Swap, do not delete.

The pattern behind every row: add a rate, add a denominator, add a time window. A raw count becomes a conversion, an all-time total becomes a monthly movement, and suddenly the number can go down, which means it can finally tell you something.

Swap rather than delete for a practical reason too: the vanity metric usually contains the raw ingredient. Total signups is the numerator of signup-to-active; pageviews are the top of the funnel that visits-reaching-the-goal measures. The data was never the problem. The framing was.

The dashboard graveyard

Four smells that a metric is decorative:

• Always up: cumulative anything. See also: the keynote trick of charting a running total.
• No target: a number with no threshold cannot be red or green.
• No segment: one blended average hides the cohort that is on fire.
• No owner: a metric everyone watches and nobody owns changes nothing.

A good BA deletes more charts than they add. Every decorative chart on a dashboard costs attention that the actionable ones need, and the graveyard grows one harmless-looking addition at a time. The smells above are the exit criteria: a metric that matches two or more of them is a candidate for deletion, not for a redesign.

The takeaway

If a number only ever goes up, has no denominator, no target and no owner, it is decoration. Run the 3-question test, swap each vanity metric for its actionable cousin, and let the dashboard shrink. The test and the full translation table fit on the print-ready PDF above; bring it to the next dashboard review.