Field Guide
Metrics
You’ll notice a restaurant’s star rating can climb while the actual experience of eating there stays exactly the same, or gets worse.
A restaurant’s online rating starts as a real, useful signal — 4.6 stars, built up honestly over hundreds of ordinary meals, telling a prospective diner something true about the food and service. Then the owner starts watching the number closely. A comped dessert appears at the first hint of a lukewarm review. Servers get coached to specifically ask happy tables for five stars. The rating climbs to 4.8. Some of that increase reflects a better restaurant. Some of it doesn’t.
A metric is a chosen, simplified stand-in for something real. The actual dining experience is enormous and layered — different for every table, every night, every server — far too much to hold in view at once. The rating compresses all of that into a single number someone can track and act on. That compression is exactly what makes a metric useful, and exactly what makes it partial: a representation of the thing, not the thing itself.
The trouble isn’t the metric. It’s what happens when the number and the reality it stands for come apart and nobody is watching for the gap. A metric can be moved upward through means that never touch the underlying experience — or that trade against it, like a rushed comp instead of an actual fix. The number goes up. The reality behind it stays flat, or quietly gets worse, and for a while nothing else will show it.
Strength · Shadow · Watch for
Turns something too large to hold in view — a whole dining experience, a whole institution’s work — into something small enough to track, compare, and act on.
The simplification that makes a metric useful is the same thing that lets it get mistaken for the whole of what it represents.
Visible effort going into moving the number itself — the ask for a five-star review, the comp offered before the complaint is addressed — rather than into the thing the number was supposed to reflect.
Places to look: A meeting where the metric gets discussed at length and the experience it represents doesn’t come up at all. A metric that’s improved for several periods running, with no one asking lately what it might no longer be catching.
What would have to be true about the actual experience — not the number — for this metric to really deserve the credit it’s getting?
Cases
What a Calculation Can Reach, on how a Metric can become the operative definition of the problem an institution is addressing.