North Star Metric: How to Choose Yours (With Examples by SaaS Type)
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.