Most companies have plenty of metrics and no hierarchy. Twenty dashboards, each department steering by its favorite number, and executive meetings that relitigate which chart matters. A north star metric exists to end that argument: one measure of delivered value that the whole company agrees predicts long-term revenue, with every team's work laddering into it visibly.
The test of a good north star
It has to be a proxy for value received, not activity generated. At WisOwl AI, signups were the flattering number: 10,000+ of them, acquired organically. The honest north star was successful matches: a recruiter and a candidate connected in a way that progresses a real hire. Signups can grow while the product fails; matches can't. At CaaStle-style subscription businesses the equivalent is active engaged subscribers instead of gross adds, because subscription economics depend on retention more than acquisition. I watched a ~$50M+ ARR portfolio managed with exactly that discipline, and our experimentation program's biggest wins came from optimizing for retention instead of acquisition.
What the engagement produces
- The metric itself, chosen through a structured evaluation: does it capture value delivery, is it measurable weekly, can teams actually influence it, and does it correlate with revenue on your own historical data.
- A driver tree, decomposing the north star into the three-to-five input metrics each team owns.
- Guardrails, because every north star can be gamed. Optimizing matches shouldn't degrade match quality; optimizing engagement shouldn't manufacture addiction. We define guardrail metrics the same day as the north star.
- A review cadence that keeps the metric in charge, and a scheduled reassessment, because north stars have lifespans and outgrow themselves as strategy shifts.
Two to three weeks, leadership workshop included. The alignment matters more than the number. Once every team can say how their quarter moves the same metric, most prioritization arguments settle themselves.