Choosing a North Star Metric You Will Not Regret

The north star metric exists to solve a coordination problem. Ten teams pulling on ten different numbers produce motion without direction. One number, understood by everyone, makes trade-offs discussable. That is the entire value, and it explains the criteria: the metric has to be something teams can genuinely affect and something the business genuinely cares about.

Four criteria

• It measures delivered value, not activity. Registrations measure your marketing. Completed jobs measure the customer getting what they came for.
• It leads revenue. Revenue itself is a lagging summary and cannot guide a decision made today. The metric should move first and predictably.
• Teams can move it. If no roadmap item plausibly changes the number within a quarter, it is a board metric, not a north star.
• It decomposes. You should be able to break it into inputs that different teams own, so the shared goal turns into distinct work.

Business type	Weak choice	Stronger choice
Marketplace	Registered sellers	Completed transactions per week
Collaboration tool	Monthly active users	Weekly active teams with 3+ contributors
Media	Pageviews	Weekly returning readers
Fintech	Accounts opened	Accounts with a recurring deposit
Ecommerce	Sessions	Second orders within 90 days

Counts of actions beat counts of people

Active-user metrics have a well-known weakness: they treat a user who does one thing a month identically to one who does forty. They also make it tempting to increase the number with notifications that generate opens rather than value. Counting the valuable action itself — messages sent, nights booked, invoices paid — sidesteps both problems and correlates better with revenue.

Where a people-count is unavoidable, qualify it. "Weekly active teams with at least three contributors" is a person-count with a value threshold attached, which restores most of the lost information.

Before adopting a metric, ask how you would increase it by 30% next quarter if you were dishonest. If the answer is easy, the metric is wrong.

Guardrails

Every single metric can be gamed by sacrificing something it does not measure. Pair the north star with one or two guardrails that must not degrade: a quality measure such as refund rate or support contacts per order, a health measure such as retention plateau, and a financial measure such as gross margin.

Decomposition makes it operational

A north star that cannot be broken down is a poster. Decompose it into a small tree: for "completed transactions per week", the branches are active buyers, transactions per buyer, and completion rate — each of which has a clear owner and its own inputs. Teams then work on branches while the shared number stays visible, and the arithmetic makes trade-offs explicit: a marketing plan that grows buyers while depressing completion rate can be evaluated rather than argued about.

Revisit it

The right metric depends on stage. Early on, the constraint is proving anyone wants the product, so a usage metric is right. Later the constraint becomes monetisation or efficiency, and the north star should shift with it. Review annually, change it deliberately with an explanation, and never change it quietly mid-quarter — the credibility cost of a moving target exceeds the benefit of the better number.