The north-star metric: pick one number. Defend it. That is the whole point.

What a north-star metric actually is

A north-star metric is the one number your team agrees to put first. It is almost never revenue itself. Revenue is the result. Your north star is the customer behavior that creates the revenue — listens on Spotify, nights booked on Airbnb, payments processed on Stripe. Pick one. Make every other team metric serve it. That is the whole framework in five sentences.

The framework exists because most growth teams measure too much. Dashboards balloon to fifty KPIs. Every team optimizes its own slice. The company hits its individual targets, and the integrated business stalls anyway. The north star is the explicit decision to put one metric above all the others. When two projects compete for the same week, the question stops being "which one has higher ROI." The question becomes "which one moves the north star faster." That single shift removes more meeting hours than any project tool I have ever seen ship.

The metric is not a slogan. It is structurally specific. Spotify does not track "engagement." It tracks time spent listening. Airbnb does not track "bookings." It tracks nights booked. Read that gap again. A booking can get cancelled. A night booked means the guest showed up, the host got paid, and the platform earned its rate. Same word, very different metric. The work is in the noun phrase, not the slogan.

Claim: Facebook disclosed daily active users (DAUs) of 526 million on the cover of its 2012 S-1 filing. Source: Facebook S-1, SEC EDGAR (February 2012). Context: DAU was the public-facing north star Facebook organized internal product work around for roughly a decade. The number sat above revenue in the prospectus because investors had already accepted that revenue would follow from DAU growth, not the other way around.

Where the framework came from

Sean Ellis brought the phrase into growth practice around 2010, while he was leading early growth at Dropbox. He later founded GrowthHackers. Amplitude turned the idea into a working framework in 2018, in a piece John Cutler wrote called the North Star Playbook. Brian Balfour at Reforge, Andrew Chen, and Lenny Rachitsky have extended the ideas in essays since. The roots run deeper too. Eric Ries pushed a similar concept in The Lean Startup. Clayton Christensen's Jobs to Be Done work gets you to the same place from a different door.

Ellis used the phrase to describe the one metric that should anchor a growth team's roadmap. His argument was simple. Growth teams that optimize for activation, while product optimizes for retention, while marketing optimizes for sign-ups, all hit their numbers — while the company makes less money. The local wins cancel each other out. Pick one metric. Route every team's work through it. The local wins start to compound instead.

John Cutler's 2018 playbook at Amplitude is what made the framework usable in a real product org. He introduced the "input metric tree" — the north star written as a formula, the product of three to five inputs, each one owned by a single team. That one idea turned a vague slogan into a tool a PM or growth lead could actually run a Monday standup against. Most product teams still use the tree shape today, even if they have never read the original playbook.

Brian Balfour's 2019 Reforge essays added scaffolding around the metric. He called it the four-fits framework. A chosen north star only works when four kinds of fit are aligned: product-market, market-model, model-channel, and channel-product. Andrew Chen made another move in The Cold Start Problem (2021). He argues that for network-effect products, the north star almost has to be a frequency-of-use metric. Sessions. Matches. Conversations. Not total users — because total users does not tell you whether the network is actually working.

Claim: Amplitude's North Star Playbook has been downloaded more than 100,000 times since publication and is cited as the reference product-analytics treatment of the framework in essays by Reforge, Lenny's Newsletter, and First Round Review. Source: Amplitude product blog (December 2018, updated 2023). Context: The playbook moved the framework from a marketing-team idea into a product-analytics standard adopted by Slack, Disney+, Atlassian, and dozens of other product organizations.

"If your north star is the metric that, when it moves, signals to your team that customers are getting more value — and your business is making more money — then everything else falls in line." John Cutler, Amplitude — North Star Playbook (2018)

Real examples from companies that disclosed theirs

Public S-1 filings, investor letters, and recorded founder interviews show how the biggest software companies have defined their north stars. The pattern is the same across every example. Each company picked a leading indicator of revenue. Each picked something the customer actually does, not something the company collects. Below is what each one publicly told the market the number was.

North-star metrics disclosed by public-company filings, founder interviews, and on-record executive talks

Company	North-star metric	Source	Why it works for this business
Facebook (Meta)	Daily active users	2012 S-1, repeated quarterly through 2018	Network value compounds with daily session frequency; DAU is the leading indicator of ad inventory
Airbnb	Nights booked	2020 S-1, Brian Chesky quarterly letters	Each night is a complete value-exchange unit — guest stays, host paid, platform takes its rate
Spotify	Time spent listening	Daniel Ek public statements; Spotify Investor Day 2022	Time predicts subscription retention better than logins; advertisers price audio inventory in listening time
Stripe	Total payment volume processed (TPV)	Public Stripe Sessions keynotes; investor materials	Stripe's revenue is a percentage of TPV; the metric is mathematically linked to revenue
Slack	Daily active users at 2,000+ messages sent	Stewart Butterfield interviews; First Round essay (2017)	The 2,000-message threshold was Slack's identified retention inflection — below it, users churned
HubSpot	Weekly active users running marketing automation workflows	Dharmesh Shah; HubSpot investor presentations	Workflows are the sticky feature; WAUs running them retain at 90%+ while broader WAUs churn
Dropbox	Files saved to Dropbox (originally) → paid sharing events	Sean Ellis interviews; Dropbox S-1 (2018)	Files were the early proxy for stickiness; sharing events became the metric that predicted team conversion
Duolingo	Daily active users completing a lesson	Luis von Ahn earnings calls; Duolingo S-1 (2021)	Streak-driven product; daily lesson completion captures the habit loop the business depends on
Notion	Weekly active users editing a page	Ivan Zhao public talks; Notion product blog	Editing distinguishes engaged users from passive readers; editing predicts paid conversion at the team level
Peloton	Total workouts per member per month	2019 S-1; quarterly investor letters through 2022	Workouts per member is the leading indicator of subscription retention and refer-a-friend behavior

Look at what these metrics are not. None of them is total users. None is total revenue. None is total signups. Each one captures the moment the customer actually gets the value the product was built to deliver. A listen. A stay. A workout. A payment processed. The growth team's job, once the metric is set, is simple to describe: find the inputs that make that one event happen more often, to more people, faster on day one.

Claim: In its 2018 S-1, Dropbox reported that paying users who shared at least one file converted at materially higher rates and retained 60% longer than paying users who did not — which is why Dropbox's north star evolved from "files saved" to "paid sharing events" as the company matured. Source: Dropbox S-1, SEC EDGAR (February 2018). Context: The shift illustrates a rule that applies to every scaling company — the right north star changes as the business model matures, and the discipline is to recognize the shift before revenue confirms it.

The four properties of a working north-star metric

A metric is north-star quality when four things are true. It is a proxy for value the customer actually receives. It happens again and again across the customer base. One team or pod owns it. And the team can move it inside one quarter. The Amplitude playbook adds a fifth that the best operators lean on hard: the metric should break cleanly into a formula of three to five inputs the team can directly attack.

Value-proxy. The metric tracks what the customer experiences, not what the company collects. "Subscriptions started" is what the company collects. "Hours of music played in week one" is what the customer experiences. The first one only predicts the second if onboarding actually works. The second one predicts revenue regardless of the customer's entry point.

Repeatable. The behavior has to happen over and over across the customer base. A metric that fires once per customer — account created, first purchase — is a leading indicator at best, an acquisition-cohort metric at worst. North stars count behaviors that repeat. Stays. Sessions. Workouts. Listens. Messages sent.

Owned. A specific team or pod has to be on the hook for moving the number. A north star that no one owns is a north star no one optimizes. The trick that makes ownership workable is the input tree. Growth owns activation rate. Product owns session frequency. Lifecycle owns reactivation. The north star is the product of all three — and now each team knows the exact input they own.

Movable. The team has to be able to move it inside one quarter or two. A metric that only moves with macro forces — the economy, the season, a rival cutting price — is not a north star. It is a market indicator. The team needs to feel the line between the work it ships on a Tuesday and the chart it stares at on Friday.

Decomposable. The strongest north stars come with a formula. Airbnb's "nights booked" can be broken into: Guests × Searches per Guest × Search-to-Book Rate × Nights per Booking. Each input gets an owner. Each gets a baseline, a target, and a tripwire. When the north star stalls, the team checks which input drifted and ships work against that one input — instead of guessing.

How to identify your north-star metric

Here is the framework I use. It is borrowed from Amplitude's North Star Playbook, Brian Balfour's Reforge essays on metric architecture, and Lenny Rachitsky's interview archive of 60-plus growth leads. The work takes about a week of focused effort, plus a month of checking the metric against actual cohort data. Skip the stress-test step at your own risk — that one is the most common reason teams pick a metric they end up regretting.

1. List the customer outcome that creates value.Write down the moment a customer actually gets the value your product was built to deliver. Not the moment of purchase — the moment of value receipt. For Spotify it is listening, not subscribing. For Airbnb it is staying, not booking. For Stripe it is a payment processed, not an account opened. The verb matters more than the noun.
2. Find the leading indicator that predicts it.Run a cohort analysis. Look at every behavior in the product that happens before the value-receipt event. The behavior with the strongest tie to 6-to-12-month retention is your best candidate. Mixpanel and Amplitude both ship cohort-correlation tools for this. In a warehouse, the query is a window function over user events, grouped by signup cohort.
3. Test for the four properties.Confirm the candidate is a value proxy, repeats across the base, can be owned by a single team, and can be moved inside one quarter. If any of the four is missing, the metric is a vanity proxy. It will move when the business is stalling, or sit flat when the business is healthy.
4. Stress-test against perverse incentives.Ask what the team would do if they only saw this number on a wall. Three failure shapes show up most often. Bad for customers, like dark patterns that inflate sessions. Bad for revenue, like growing free users at the expense of paid. Or simply not sustainable, like one-off promo blasts. Any of the three means the metric is wrong. Goodhart's law applies. Once a measure becomes a target, it stops being a good measure if it can be gamed.
5. Break the metric into an input formula.Write the north star as a formula of three to five inputs the team can move directly. Users times sessions per user times actions per session is a common shape. If you end up with more than five inputs, the north star is probably too high-level. If you end up with fewer than three, the metric is probably already an input metric for something else.
6. Assign each input to a team.Make each input the responsibility of exactly one team or pod. That team owns the baseline, the target, and the experiments. Inputs without owners do not move. The rule holds across every growth org I have audited in the last five years.
7. Set a quarterly target and a tripwire.The quarterly target is the direction the team is moving the input. The tripwire is the floor that, when you hit it, triggers a hard look at the metric itself — not at the team's tactics. Hitting the tripwire means the north star may have stopped predicting value. That is a board-level conversation, not a sprint-planning one.
8. Re-examine the metric every 12 to 18 months.North stars change as the company matures. The number that matters at $5M in revenue is almost never the same number that matters at $50M. Brian Balfour's Reforge data shows the average scale-stage software company changes its north star every 14 months. Planning for the change is what separates teams that compound from teams that get stuck defending a metric that stopped working a year ago.

Claim: Across roughly 50 paid-media and analytics audits we run per year at RGM, the most common north-star-metric failure is a metric that satisfies properties 1 through 3 but fails property 4 — the team picked a value proxy they cannot actually move within a quarter, usually because the input variables sit outside the growth team's control (pricing, product roadmap, third-party platform rules). Source: Real Growth Matters Inc., internal audit data, 2024-2026. Context: The fix is rarely to find a different metric — it is to redraw the org chart so the team accountable for the north star also owns the inputs. We have written about this in our piece on CAC payback and the LTV ratio.

The three failure modes most teams hit

Teams pick north stars that look great on a slide and then break under the weight of actual operating decisions. Three failure modes account for almost every regret I see in audit. Catching them before the team commits saves a quarter — sometimes more — of work spent on the wrong thing.

Failure mode 1: vanity proxy

The metric tracks something easy to count but uncorrelated with actual customer value. Total users. Total page views. Total signups. Total downloads. All of these move when marketing spends more, and none of them move when the product gets better. The quick test: does the metric improve when retention improves? If it can rise while retention falls, you have a vanity proxy.

Failure mode 2: gameable metric

The metric rewards behavior that hits the number without delivering value. The classic example is "weekly active users" defined as anyone who opens the app. That definition gets gamed by push notifications driving zero-engagement opens. Slack saw this coming and added a depth threshold: 2,000 messages sent. Any metric that has no quality or depth constraint is going to get gamed by the team trying to hit it.

Failure mode 3: outside the team's control

The metric depends on inputs the growth team cannot touch. "Revenue per customer" is owned by pricing, product, sales, and finance. The growth team's work moves it indirectly at best. North stars that sit downstream of three or more functions almost always go stale, because no one team has the authority to ship the experiments that would actually move it. The fix is to pick a metric upstream of the bottleneck — usually a usage metric the product and growth teams jointly own.

North-star metric versus adjacent concepts

The phrase "north-star metric" has picked up overlapping meanings with at least five neighboring frameworks. Operators who know the differences argue more productively about which one their team actually needs. Below are the distinctions that matter when picking the right framework for where the team sits today.

How north-star metric compares to OMTM, OKRs, KPIs, MAU, and AARRR pirate metrics

Framework	Scope	Time horizon	Best fit
North-star metric	One company-level outcome that predicts long-term value	12–18 months, sometimes longer	Scale-stage companies aligning multiple teams on a shared target
One Metric That Matters (OMTM)	The single metric for the company's current stage	3–9 months, explicitly stage-dependent	Early-stage startups that expect the metric to change as they cross PMF
OKRs (Objectives + Key Results)	Quarterly objectives with 3–5 measurable key results each	One quarter, sometimes annual	Larger orgs coordinating across multiple teams and functions
KPIs (Key Performance Indicators)	Multiple per team, tracking specific functional outcomes	Continuous, dashboard-driven	Every operating company tracks dozens of KPIs in service of the north star
MAU / DAU	Aggregate engagement count	Continuous	Sometimes a north star (Facebook, Duolingo), often a vanity proxy without behavioral constraints
AARRR pirate metrics	Funnel-stage breakdown: Acquisition, Activation, Retention, Referral, Revenue	Continuous funnel view	Decomposing where in the funnel the north star is failing and which AARRR stage to attack

These frameworks work together. They are not alternatives. A scale-stage company should run on a north star, break it down into OKRs, watch its functional KPIs, and pull out AARRR as the diagnostic lens when the north star stalls. The mistake is to treat them as either-or. Picking AARRR instead of a north star leaves the team with five funnel metrics and no anchor for the work.

When to change your north-star metric

North stars change. The number that defines a product-led-growth company at $5M ARR is almost never the number that defines the same company at $50M ARR. The triggers for the change are predictable. Operators who plan for the shift capture more upside than the ones who keep defending an outdated metric. Below are the three triggers that should prompt a fresh look, and the cost of ignoring them.

The business model expanded. Dropbox started with files saved as its north star. When the company moved toward team workspaces, the metric that predicted long-term value became paid sharing events — a different behavior, in a different part of the product, owned by a different team. Any company that adds a marketplace on top of a SaaS product, or B2B on top of a B2C base, almost always needs a new north star to capture the new value exchange.

The metric is being hit while revenue stalls. When the team is hitting the north star and the business is not growing, the metric has stopped predicting value. The Slack-style fix is to add a behavioral qualifier — active users at X actions per week — instead of throwing the metric out. The deeper fix is to ask whether the company has crossed into a stage where a different value exchange is now the binding constraint.

The leading indicator drifted. Cohort analysis can show you that the behavior most predictive of retention has shifted. Maybe two years ago, weekly logins were the strongest signal. Now it might be workspace invites. The team should refresh the cohort-correlation analysis once a year and update the metric when the regression shifts under it.

Claim: Brian Balfour and Casey Winters, writing in published Reforge essays, argue that the average tech company in the scale-up phase changes its north-star metric every 14 months on average, and that companies that do not re-examine the metric annually are likely to be running on an outdated definition by the time the next funding round closes. Source: Reforge published essays (2019-2024). Context: The cadence implies that "set and forget" is the wrong model. The discipline is to re-examine, not to change for the sake of change — most reviews end with the existing metric reaffirmed, but the act of re-examining catches the cases where it has stopped working.

How the north-star metric fits with adjacent concepts

The north-star metric is the anchor of a metric system, not the whole system. The frameworks around it — cohort analysis, unit economics, funnel breakdown, jobs-to-be-done framing — do the diagnostic work when the north star moves the wrong way. A team that runs only on the north star is flying blind on the inputs. A team that runs only on the surrounding metrics has no anchor for the work. The two layers are built to work together.

For commerce companies, the north star usually settles on contribution-margin-positive repeat purchases per cohort, paired with CAC payback period and the LTV ratio as the unit-economics check. For B2B SaaS, it usually lands on weekly active users running the workflow that predicts paid conversion, paired with AARRR funnel breakdown when activation stalls. For marketplaces, it is volume of completed value-exchange events, paired with cohort retention curves to show whether the matching is actually working.

The principle behind all of this is older than the phrase "north-star metric." Clayton Christensen's jobs-to-be-done framing says customers hire products to make progress on a specific job. The north-star metric is the company-level measurement of how often that progress is actually happening, repeatably, at scale.

Quick answers about the north-star metric

What is a north-star metric in plain English?

The one number a growth team is trying to move above all others. It measures the value a customer receives in a way that predicts revenue. Examples: Spotify's time spent listening, Airbnb's nights booked, Stripe's payment volume processed.

Who first used the phrase?

Sean Ellis, while working at Dropbox and later founding GrowthHackers, is widely credited with popularizing the phrase in growth practice around 2010. The framework was formalized for product analytics by Amplitude's John Cutler in the North Star Playbook in 2018.

How is it different from a KPI?

A KPI is one of many performance indicators a team tracks. A north star is the single metric the company optimizes above all others. Most companies have dozens of KPIs and one north star. The difference is hierarchy.

What is the easiest test for whether a metric is north-star-quality?

Ask: if this number doubles, would the business be obviously healthier in 12 months? If yes, it is a candidate. If no — if the number can double without the business getting healthier — it is a vanity proxy.

How often should the north star be reviewed?

Annually at minimum. Reforge essays from Brian Balfour and Casey Winters report that the average scale-up tech company changes its north star every 14 months. Most reviews reaffirm the existing metric; the act of reviewing catches the cases where it has stopped working.

Do we still need other metrics if we have a north star?

Yes. The north star is the single optimization target. The input metrics that break it, the leading indicators that predict it, the lagging indicators that confirm it — all are tracked in service of the north star, not in competition with it. Dashboards do not shrink; the hierarchy does.

Frequently asked

What is a north-star metric?

A north-star metric is the single business outcome that, when optimized, drives long-term value better than any other measurement available to the team. The term was popularized by Sean Ellis around 2010 and formalized in Amplitude's North Star Playbook in 2018. A north star is not revenue itself. It is the leading indicator of the revenue that follows.

Who invented the north-star metric framework?

Sean Ellis, then at Dropbox and later founder of GrowthHackers, is widely credited with popularizing the phrase in growth-team practice around 2010. The framework was later codified by Amplitude in the North Star Playbook in 2018 and extended by John Cutler, Brian Balfour at Reforge, Andrew Chen, and Lenny Rachitsky in published essays through 2024.

What is an example of a north-star metric?

Facebook used daily active users as its north star through its early growth years and disclosed the number on the cover of its 2012 S-1 filing. Airbnb tracks nights booked. Spotify tracks time spent listening. Stripe tracks payment volume processed. The pattern is consistent: each number is a leading indicator of revenue, not revenue itself.

How is a north-star metric different from a KPI?

A KPI is one of many performance indicators a company tracks. A north star is the single metric the company optimizes above all others. Most companies have dozens of KPIs and one north star. The difference is hierarchy, not category. The north star sits at the top of the metric tree and every other KPI feeds into it.

How is north-star metric different from OMTM (One Metric That Matters)?

OMTM, from Alistair Croll and Benjamin Yoskovitz's 2013 book Lean Analytics, is the metric that matters most for a company's current stage. A north star is meant to remain stable across multiple stages. In practice the two overlap heavily — many teams use the terms interchangeably — but OMTM explicitly assumes the metric will change as the company matures, while a north star is the longer-horizon anchor.

What are the four properties of a good north-star metric?

The metric must be a proxy for customer value, repeatable across the customer base, owned by a specific team or pod, and movable by the team's actions in the relevant time horizon. Amplitude's North Star Playbook adds a fifth: the metric should break cleanly into a multiplicative formula of three to five input metrics each team can directly affect.

Can a company have more than one north-star metric?

Most experienced operators argue against it. Two north stars create competing targets that confuse the team. The exception is two-sided marketplaces. Airbnb arguably tracked both nights booked and host-supply hours during its scale years. Even there, leadership eventually picks one as primary and treats the other as a constraint to monitor.

When should a company change its north-star metric?

When the metric stops predicting long-term value. Common triggers: the company enters a new business model, the product expands into a category the metric does not measure, or the metric is being hit while revenue stalls. Reforge essays from Brian Balfour and Casey Winters argue the average tech company changes its north star every twelve to eighteen months in the scale-up phase.