Your north star metric is the single number that best captures the value your product delivers to customers as it grows — not your biggest number, but your most honest one. It should move only when customers get real value, correlate with revenue, and stay simple enough for engineering and the board to both understand it.
Quick Answer: A north star metric (NSM) is the one number that proxies customer value and predicts business growth — think Airbnb's Nights Booked or Slack's messages sent between teammates. You find it by tracing your product's actual value moment, not by defaulting to whatever number is already on a dashboard.
Most teams treat "picking a north star metric" as a naming exercise: gather the leadership team, argue for an hour, write a number on a slide. That produces a metric that sounds strategic and behaves like a vanity number — it goes up every week regardless of whether anyone's life got better. This guide is longer than the usual listicle because the real work is upstream of the metric itself: in the assumptions about value you're making before you ever open a spreadsheet.
What a North Star Metric Actually Is (And Isn't)
A north star metric is a single, trackable number representing the core value your product delivers to customers, chosen because it correlates with durable business outcomes like revenue and retention. It is not a KPI dashboard, not raw revenue, and not whatever metric already has a clean chart in your analytics tool.
The term traces back to growth teams at companies like Facebook and Airbnb in the early 2010s, but the underlying discipline is older. Alistair Croll and Benjamin Yoskovitz, in their book Lean Analytics, called this the OMTM — the One Metric That Matters — and argued every stage of a business has a different one, chosen deliberately rather than inherited from last quarter's report.
A north star metric earns its title only if it clears three bars:
- It's a leading indicator, not a lagging one. Revenue and churn tell you what already happened; a good NSM moves before those do, giving you time to act.
- It reflects customer value, not company convenience. "Signups" measures your funnel. "Sessions where a user completed the task they came for" measures value.
- A product team can actually influence it. If the number only moves because of macro market forces or a sales push, it isn't a product metric — it's a business result wearing a product costume.
Most teams that skip this filtering step for a deeper look at what makes a metric meaningful in the first place would benefit from a primer on building a genuinely data-driven product analytics practice before they touch the north star question at all — the metric only works if the surrounding measurement discipline does.
Why Most PMs Default to a Vanity Metric Instead
Most PMs choose a vanity metric — total signups, downloads, or cumulative registered users — because it's already instrumented, always trends upward, and looks impressive in a board deck. None of that means it reflects whether a single customer is better off. A real north star metric takes deliberate, often uncomfortable work to define.
Vanity metrics are seductive for structural reasons, not just laziness:
- They're cumulative, so they can never go down — a comforting property in a room full of stakeholders who want good news.
- They're already piped into your existing dashboard, so choosing them costs zero engineering time.
- They correlate loosely with effort (marketing spend, feature launches) even when they don't correlate with customer value, which lets a team claim credit without proving impact.
The contrast becomes obvious once you line up what well-known products optimize for against what they used to report publicly instead:
| Company | Vanity metric it could have chosen | Actual north star metric | Why the real one is better |
|---|---|---|---|
| Airbnb | Total listings created | Nights booked | Listings can sit empty; nights booked means a guest and host both found value |
| Slack | Total messages sent | Messages sent between teammates in the first weeks | Bot/system messages inflate volume without reflecting real collaboration |
| Facebook (early growth team) | Total signups | Users reaching a set number of friends within days of joining | Signups without connections churn fast; connected users stick around |
| Superhuman | App downloads | Users who'd be "very disappointed" without the product (Sean Ellis PMF survey) | Downloads say nothing about whether the product became essential |
| Spotify | Total streams | Time spent actively listening per active user | Total streams can be inflated by auto-play; active listening reflects real engagement |
A deeper breakdown of which numbers quietly mislead roadmaps is worth its own read — see this guide to spotting and retiring vanity metrics if your current dashboard is dominated by cumulative counts that only ever go up.
The Anatomy of a Real North Star Metric
A defensible north star metric passes a value test, a sensitivity test, and an ownership test — it moves when and only when customers get more value, it's sensitive enough to detect real changes within weeks, and a cross-functional team can trace their work to it. Metrics that fail any one of these three collapse under scrutiny within a quarter.
The Value-Moment Test
Every product has a moment where a customer actually receives what they came for — Kerry Rodden's HEART framework at Google (Happiness, Engagement, Adoption, Retention, Task success) was built precisely to make that moment measurable instead of assumed. Your north star candidate should sit as close to that moment as instrumentation allows.
Finding that moment is really a Jobs to Be Done exercise disguised as an analytics problem: you're asking what job the customer hired your product to do, and your metric should count how often that job gets done well. If you haven't done that mapping yet, working through a full Jobs to Be Done framework exercise first will surface candidate metrics your team hasn't considered — most teams find their eventual NSM is a direct restatement of the core job, not a separate invention.
Input Metrics: The Tree Beneath the Star
A single north star number is meaningless to a team without a supporting tree of input metrics — the 3-6 levers a squad can actually pull that plausibly move the star. Amplitude's widely referenced North Star Playbook frames this as splitting output (the star) from input (what teams do daily), so no one confuses "watching the number" with "doing the work."
- Input metrics should be owned by a specific team, not shared ambiguously across the org.
- They should be leading indicators of the star itself, verified with real data, not assumed by analogy to another company's tree.
- Counter-metrics matter as much as input metrics — a growth lever that quietly tanks retention or NPS needs a metric watching it too.
How to Choose Yours: A Step-by-Step Process
Choosing a north star metric is a five-step process: map the value moment, draft several candidates, stress-test each one against the value/sensitivity/ownership bars, validate the strongest candidate against real retention data, and only then lock it in with a supporting input-metric tree. Skipping any step is how teams end up re-litigating the metric every two quarters.
- Map the customer journey and locate the "aha" moment — the point where a new user's behavior visibly changes because they got value. A full customer journey mapping exercise makes this moment visible instead of guessed at, especially the emotional dip right before it that most teams miss because they only look at the happy path.
- Draft three to five candidate metrics, deliberately including at least one "obvious" vanity option so the group can see the contrast in the room, not just in theory.
- Score each candidate against the value, sensitivity, and ownership tests from the section above — a simple weighted scorecard works better than a debate, because it forces everyone to name their assumptions out loud.
- Validate the leading candidate with a real cohort analysis, checking whether users who hit a high value on this metric actually retain and monetize better than users who don't. A rigorous cohort analysis for product managers is the difference between "this metric feels right" and "this metric is provably predictive."
- Pressure-test with a live experiment before fully committing — change something that should move the candidate metric and confirm it actually does, using sound experiment analysis and statistics for PMs rather than eyeballing a chart that moved for unrelated reasons.
Chamath Palihapitiya's often-cited account of Facebook's early growth team — that new users who added roughly seven friends within ten days were dramatically more likely to stick around — is popular precisely because it shows steps 1 and 4 in action: a value moment, hypothesized and then confirmed against real retention data, not assumed from a gut feeling in a meeting.
North Star Metrics by Business Model
The right north star metric depends heavily on your business model, because "value delivered" looks structurally different across a subscription product, a two-sided marketplace, and an ad-supported consumer app. Copying a metric from a company with a different model is one of the most common mistakes teams make.
| Business model | Typical north star candidate | What it captures | Common trap |
|---|---|---|---|
| B2B SaaS | Weekly/monthly active accounts completing a core workflow | Ongoing, repeated value to the buying organization | Chasing seat count instead of usage depth |
| Two-sided marketplace | Completed transactions (e.g., nights booked, rides completed) | Both sides of the marketplace found a match | Optimizing supply-side listings without demand-side match rate |
| Consumer social | Users reaching a connection or content threshold in early days | Network value kicking in before novelty wears off | Total signups or installs, which say nothing about retention |
| Content / media | Time actively engaged per active user | Whether content is genuinely holding attention | Total page views inflated by auto-refresh or infinite scroll |
| Productivity tool | Core workflows completed per active user per week | Repeated, durable task success | Feature adoption counted once, never re-measured |
Notice that none of these are single universal metrics like "engagement" — each is a specific, falsifiable behavior. If your candidate metric can't be stated as a specific countable event, it isn't ready yet.
The Assumption Problem: Why Analytics Falls Apart Without a Logged Hypothesis
Most north star metric failures don't originate in the dashboard — they originate weeks earlier, in an assumption someone made out loud in a meeting and nobody wrote down. By the time the metric looks wrong in a retro, the team is reconstructing "what we thought would happen" from memory, and memory reliably favors whichever story makes the team look right.
This is the quiet failure mode underneath most of the mistakes in this article:
- A PM assumes "engagement" means time-in-app, ships against that belief for two quarters, and only discovers in a postmortem that the team actually meant something narrower.
- An analyst assumes a cohort is comparable to last month's, runs the comparison, and nobody logged the assumption that a pricing change happened in between.
- A team picks an input metric assuming it drives the star, watches it rise for a quarter, and only later realizes nobody ever tested the causal link — they just assumed it, once, and it calcified into "known fact."
None of these are data problems. They're unexamined-assumption problems, and unlike data, an assumption that never gets written down at the moment it forms is nearly impossible to audit later — you're left asking three people what they remember believing eight weeks ago.
The point isn't the tool; it's that an assumption caught the moment it's made is falsifiable, and an assumption recalled three months later during a metric debate almost never is.
Key Takeaways
- A north star metric proxies customer value, not company activity — it should only move when a customer genuinely got more value from your product.
- Vanity metrics survive because they're cumulative and already instrumented, not because anyone decided they're the right thing to watch.
- Every real NSM passes three tests: it's a leading indicator, it reflects value rather than convenience, and a product team can actually influence it.
- The metric needs a supporting tree of input metrics and counter-metrics owned by specific teams, or the star becomes a number people watch rather than act on.
- The right metric depends on your business model — a marketplace, a SaaS product, and a consumer social app each need structurally different candidates.
- Choosing the metric is a five-step process ending in an experiment, not a single meeting where the loudest voice wins.
- Most analytics mistakes trace back to an assumption nobody logged — capture the hypothesis the moment you form it, not when you're explaining a chart in a retro.
Frequently Asked Questions
What's the difference between a north star metric and a KPI?
A north star metric is the single number a whole company aligns around because it best represents customer value and predicts growth; a KPI is any tracked indicator, and most organizations have dozens. Every north star metric is a KPI, but almost no KPI qualifies as a north star — most measure activity or output rather than value delivered.
Can a company have more than one north star metric?
In practice, yes, especially at multi-product or multi-business-unit companies where a single number would flatten genuinely different value propositions into one misleading average. The tradeoff is real: more than one or two dilutes the alignment benefit that makes a north star metric useful in the first place, so most single-product companies should resist splitting it further than necessary.
How often should you change your north star metric?
Rarely, and only when your business model or core value proposition has genuinely shifted — most companies should expect their north star metric to hold for a year or more, not a quarter. Frequent changes usually signal the original metric wasn't validated properly in the first place, not that the business evolved that fast.
Is monthly active users (MAU) a good north star metric?
Usually not on its own, because MAU counts anyone who touched the product at all, including users who opened it once and got no value. A stronger version narrows it to users completing a specific value-generating action, which is closer to why companies like Slack and Airbnb built their star around a specific completed behavior rather than raw activity.
What's a good north star metric for an early-stage startup?
Early-stage teams are usually better served by a product-market fit signal — like the share of users who'd be "very disappointed" without the product, per Sean Ellis's PMF survey methodology — than a full north star metric, because there isn't yet enough usage volume or business model clarity to validate a durable value metric. Treat the PMF signal as a placeholder until growth gives you enough data to validate a real candidate.