Crafting Your Map

Key takeaways

• A KPI is a map: a deliberate simplification of the business that leaves things out on purpose.
• The trouble starts when you forget the omission and treat the map as the territory.
• Good maps are sparing. A few measures, chosen because they move with the thing you actually care about, beat a dashboard of forty.
• Targets quietly become anchors, and then they become identities, so set them knowing they will be hard to revise later.
• The working test of a metric is whether it is meaningful, measurable, and actionable. If it cannot change a decision, it is decoration.

A KPI is a map, and a map leaves things out on purpose

Every metric you choose is a reduction. You take a sprawling, partly unknowable business and you draw a small picture of it that fits on a screen. That is the point of a map, and it is why maps are useful: a map the size of the territory would help nobody. So a dashboard is a sketch of the bits someone decided were worth watching, not the business itself, and it is drawn at a moment that has already passed.

This is easy to say and hard to feel, especially when the sketch is a good one. General Patton is said to have understood that the map is not the territory. The warning matters most for the person who has stared at the map so long that they have stopped noticing what it left out.

The trouble starts when you mistake the map for the ground

Peter Bernstein warned that our lives teem with numbers, and that we forget they are only tools, with no soul, until they quietly become fetishes. You can watch it happen. A team gets a clean ROAS figure, the figure becomes the conversation, and within a quarter the business is being run to make that one number rise. Spend gets cut on anything the number cannot see. The map has replaced the place.

There is a related trap I see constantly, which is mistaking motion for momentum. A metric moves, everyone feels productive, and nobody asks whether the movement connects to anything that pays the wages. Numbers need movement to be worth watching. Movement is not the same as progress.

Fewer measures, chosen for what they stand in for

The instinct under pressure is to add more metrics, on the theory that more coverage means more safety. In practice it usually adds noise. Statisticians have a word, parsimony, for a model that explains as much as possible with as few moving parts as possible, and it is a good discipline for a dashboard. Checklists in aviation and surgery tend to run to between five and nine items, roughly the limit of working memory, and a working set of KPIs benefits from the same restraint.

I once asked a founder to tell me, without looking, the five numbers that run her business. She got to eleven and was still going. Her dashboard had forty-three tiles. When we cut it to six, the arguments in her Monday meeting changed, because for the first time everyone was looking at the same small map instead of each defending the tile that flattered their own team.

Targets become anchors, then identities

A target is a number you make up about the future, and the Your Brain post on anchoring earlier in the series explains what happens next. The stretch goal you set in January becomes the reference point every later forecast is judged against. Worse, people attach themselves to it. Missing it stops being information about a hard quarter and starts being a verdict on a person. Set targets by all means, but set them knowing they will calcify, and build in an honest moment to ask whether the target still describes the world or just your January optimism.

If a number cannot change a decision, it is decoration

The plainest filter I know for a metric runs to three words: meaningful, measurable, actionable. Meaningful, so it connects to something you care about. Measurable, so you are not re-arguing definitions every week. Actionable, so a change in it would change what you do. The renaming treadmill, from ROI to TACOS to ROAS to MER, often fails the third test: the label changes, a small cult gathers around the new acronym, and the decision the number is meant to serve never gets any clearer.

A map is worth keeping. It is also worth distrusting, in proportion to how much you have come to rely on it. The next tier of this series, on KPIs that corrupt, is about what happens when a map stops describing the territory and becomes the thing people are paid to move. That is where a good metric can turn into a bad incentive, and where the question stops being "is this accurate?" and becomes "what is this making people do?"