Every PM knows to pick a North Star metric. It's one of the first things you learn — align the team around a single number, optimize relentlessly for that number, and let it guide your roadmap. Simple. Clean. Usually wrong.
The problem isn't the concept of a guiding metric. The problem is that most teams pick their North Star based on what's easy to measure and what sounds good in a board deck, not on what actually maps to the value customers get from the product.
The Measurement Trap
Here's how it usually goes. You launch. You need something to track. You pick a metric that's easy to instrument — sessions, DAUs, activation rate, conversion. You build a dashboard. You start optimizing.
Three months later, the metric is up. But the customers who matter most are churning. The feature that moved the metric didn't move the outcome. You've optimized the map while the territory shifted.
A metric is not the thing you care about. It's a proxy for the thing you care about. Proxies drift.
What a Real North Star Looks Like
A genuine North Star metric has three properties:
-
It captures customer value, not product activity. "Messages sent" is activity. "Conversations that led to a booked meeting" is value. The latter is harder to measure and worth every bit of the extra effort.
-
It's a leading indicator of retention. If your North Star goes up, renewal rates should follow. If there's no empirical connection between the metric and long-term retention, you're steering by a star that's already moved.
-
The team can influence it. A metric nobody on your team can move is a vanity metric with better branding.
The Alternative: Metric Trees
Instead of a single North Star, build a metric tree. At the top: the one thing that, if true, means the business is working. Below it: two or three drivers that predict whether the top-line metric will move. Below those: the leading indicators that tell you, this week, whether you're on track.
The tree is harder to build and harder to explain to a board. But it's honest about the fact that value is multidimensional, that drivers interact, and that optimizing any single number at the expense of others creates distortions.
North Stars are not wrong. They're just often too simple for the problem they're meant to solve. Build the tree. Pick your battles carefully. And check every six months whether the metric you're optimizing for still maps to the value you actually care about.