Ask five people at a startup what winning looks like this quarter and you will often get five answers: traffic, signups, MRR, NPS, feature launches. A North Star Metric (NSM) exists to collapse those five answers into one, a single measure of the value your product delivers to customers, which the whole company can align behind.

Chosen well, it is the most clarifying management tool a founder can adopt. Chosen badly, it institutionalizes the wrong goal at company scale. Here is how to choose well.

What a North Star Metric actually is

An NSM is the metric that best captures the value customers receive from your product, measured in a way that predicts durable business results. The classic illustrations:

A messaging platform: messages sent within active teams, not registered accounts.
A streaming service: hours watched, not subscriptions sold.
A marketplace: nights booked, not listings created.
A collaboration suite: weekly collaborating teams, not seats sold.

Notice the pattern: every good NSM counts value delivery events, not money collected and not accounts opened. Revenue is the lagging shadow of value; signups are the leading promise of it. The NSM sits between them, at the moment value actually changes hands.

Why not just use revenue?

Three reasons:

Revenue lags. A SaaS can grow MRR for two quarters after the product started failing customers, annual contracts and inertia hide the decay. An NSM based on usage turns down first.
Revenue is not actionable for most teams. A support engineer cannot move MRR directly; they can move "weekly active workspaces."
Revenue can be grown against the customer, aggressive upsells, dark-pattern renewals. An NSM anchored to customer value cannot: growing it requires making customers better off, which makes it safe to maximize.

Revenue stays on the dashboard, as the check that the NSM monetizes. The NSM leads; revenue confirms.

The five tests of a good NSM

Run every candidate through these:

Value test: does it increase only when a customer genuinely receives value? ("Emails sent" fails if spam counts; "emails that get replies" is closer.)
Leading test: does it predict retention and revenue 3–6 months out? Validate on your own cohorts: users high on the metric in month 1 should retain and expand more.
Actionability test: can teams influence it within weeks through product, marketing and success work?
Breadth test: does it aggregate across your whole active base (a rate or volume), rather than describing a niche?
Anti-gaming test: if every team maximized this number ruthlessly, would customers be better off? If not, tighten the definition (add a quality qualifier: active, weekly, successful, retained).

Examples by SaaS type (illustrative)

Communication / collaboration tools: weekly teams exchanging N+ messages; documents edited by 2+ people per week. Value = collaboration happening.
Analytics / BI products: weekly active organizations viewing reports fed by live data; queries run that get consumed by a human. Value = decisions informed. ("Dashboards created" fails the value test, an unviewed dashboard delivers nothing.)
Scheduling / workflow automation: meetings booked; workflows executed successfully per week. Value = the job completing.
Developer tools / API products: weekly active projects making successful production calls. ("API keys issued" is a promise; "successful calls from production" is delivery.)
E-commerce / checkout SaaS: GMV processed, or orders completed through the platform per week.
Content / CMS platforms: posts published and read, publication multiplied by audience, because value needs both sides.
Marketplaces: completed transactions per period (with both sides retained).

Note the recurring qualifiers: weekly, active, successful, 2+ users. The precision is the metric.

From North Star to input metrics

An NSM you cannot decompose is a poster, not a tool. Break it into 3–5 input metrics, the operational drivers teams actually work on. Illustration for "weekly active organizations viewing live-data reports":

New organizations activated per week (acquisition × activation)
% of organizations with a live data connection (setup quality)
Reports viewed per active organization (depth of habit)
Week-over-week organization retention (durability)

The NSM = the product of its inputs, approximately. Teams own inputs; leadership watches the star. Quarterly goals get set on inputs ("raise live-connection rate from 55% to 70%"), which keeps goals actionable while guaranteeing they aggregate into customer value.

A worked selection, start to finish

Illustrative walkthrough for a scheduling SaaS choosing its star:

Candidates: signups per week, meetings booked per week, weekly active users, revenue.
Five tests applied: signups fail the value test (a signup who never books received nothing). WAU is vague, active doing what? Revenue fails the leading test. "Meetings booked per week" passes value (a booking is the job done), passes leading (validated on cohorts: users with 3+ bookings in week 1 show 3x month-3 retention, illustratively), passes actionability (onboarding, reminders, integrations all move it), passes breadth, and mostly passes anti-gaming, tightened to "meetings booked and attended" to exclude junk bookings.
Star chosen: meetings attended via the platform per week.
Inputs assigned: new users reaching first booking within 3 days (owner: onboarding), bookings per active user (owner: product), booking-page conversion rate (owner: growth), 4-week user retention (owner: lifecycle).

Total elapsed time for a rigorous selection: about two weeks, mostly spent validating the leading test on historical cohorts. Skipping that validation step is how companies end up ceremonially tracking a number that predicts nothing.

Common failure modes

Choosing a vanity star: registered users, page views, downloads. Fails the value test; grows while the business dies.
Choosing revenue as the star: fails the leading and actionability tests (see above).
Compound metrics nobody understands: a weighted index of seven signals cannot rally a company. If you cannot explain the NSM to a new hire in one sentence, simplify.
Two stars. Companies with two north stars have zero. If you serve two genuinely distinct value loops (e.g., a marketplace's supply and demand), pick the binding constraint as the star and track the other as a guardrail.
Never revisiting. The right NSM at 100 customers may be wrong at 10,000. Re-run the five tests yearly; change the star rarely and loudly.

A note on guardrail metrics: alongside the star, name two or three numbers that must not degrade while you maximize it, typically revenue, churn, and a quality signal (support tickets per active account, error rates). Guardrails are what let you push hard on the NSM without anxiety: any input experiment that lifts the star while tripping a guardrail gets rolled back, no debate needed. The star tells you where to run; the guardrails tell you when you have cut a corner.

Making it operational

An NSM only works if it is seen: on the dashboard everyone opens, reviewed in the weekly growth meeting, decomposed into inputs with owners, and cross-checked against revenue monthly. The moment it lives in a slide instead of a live dashboard, it stops steering anything.

That is the practical bar: your North Star and its input metrics, computed from real product and billing data, on the same screen as the AAARRR funnel they summarize, visible to the whole team every day. Exactly the kind of single-screen discipline Growth Pilot's cockpit was designed to make effortless for founders.