North star metric: how to choose one (with examples and an input metrics tree)

A north star metric is the single number that best captures the value your product delivers to customers, chosen so that if it grows sustainably, the business grows too. To pick one, find the moment a user gets real value, count how often that happens in a fixed time window, and check that it predicts retention and revenue. You then break it into three to five input metrics that individual teams can actually move.

Key takeaways

• A north star metric measures customer value delivered, not revenue collected. Revenue is the result; the north star is the leading indicator.
• Good north stars usually combine a value action, a threshold and a time window, for example "weekly active buyers who place two or more orders".
• The input metrics tree is where the work happens. Teams own inputs (breadth, depth, frequency, efficiency), not the north star directly.
• Pair the north star with guardrail metrics so it cannot be gamed at the expense of trust, quality or unit economics.
• Pre-product-market-fit startups are often better served by a retention curve than by a formal north star.

What is a north star metric?

A north star metric (NSM) is the one product metric a company agrees best reflects the core value customers get from the product, used to align teams on long-term growth rather than short-term output.

The idea was popularised in growth circles by Sean Ellis, who described it as the metric that best captures the core value your product delivers. Amplitude later turned it into a fuller method, the North Star Framework, which pairs the metric with a set of input metrics. In Amplitude's description, a good north star expresses customer value, represents the product strategy and is a leading indicator of future success, rather than a lagging measure such as monthly revenue.

The practical point is focus. A company with twelve "top priority" metrics has none. A shared north star lets the growth team, the core product team and the platform team argue about how to grow the same number rather than about which number matters.

North star metric vs KPIs, OKRs and the one metric that matters

These terms get mixed up constantly. Here is how they relate.

A healthy setup: the north star stays stable for a year or more, team OKRs target input metrics, and guardrails are reviewed in the same meeting so nobody is surprised.

What makes a good north star metric?

Test every candidate against these questions. A metric that fails two or more is not your north star.

1. Does it reflect value the customer actually received? "Orders delivered" passes. "App opens" usually fails.
2. Does it lead revenue? If it rises this quarter, revenue and retention should follow in later quarters.
3. Can teams influence it through the product? If it moves mainly with marketing spend or the economy, it is a business metric, not a product north star.
4. Is it simple enough to say in one sentence? Every engineer should be able to explain it.
5. Does it have a time window? Weekly or monthly, chosen to match the natural usage frequency of the product.
6. Is it hard to game? If the cheapest way to grow it hurts users, you will eventually grow it that way.
7. Is it a count rather than a ratio? Ratios such as conversion rate can improve when the denominator shrinks. Ratios make great input metrics but poor north stars.

The three games: attention, transaction and productivity

Amplitude's framework suggests that most digital products play one of three "games", and the game shapes the north star.

Knowing your game stops a common mistake: a B2B workflow tool chasing "time spent in the app", which for a productivity product often means the product is slow, not loved.

How to choose your north star metric, step by step

1. Write down your product's core value in one sentence. Not the feature list; the outcome. "Customers get groceries at home in under 30 minutes without planning ahead."
2. Find the value moment. The action where that value is delivered and observable in your data: an order delivered, a report shared, a song played to the end.
3. List three to five candidate metrics built on that moment. Vary the threshold and window: "weekly orders delivered", "weekly customers with two or more orders delivered", "monthly delivered orders on time".
4. Run each candidate through the seven tests above. Cross out anything that is a ratio, a vanity count or out of the team's control.
5. Check it against history. Pull a year of data. Do users or cohorts with higher values of the candidate retain better and generate more revenue later? You are looking for a consistent relationship, not a perfect correlation.
6. Add guardrails. Decide which metrics must not degrade: delivery time, refund rate, contribution margin, complaint rate, uninstalls.
7. Pressure-test with leadership and two or three engineers. If the founders cannot agree, you do not have a metric problem; you have a strategy disagreement. Surface it.
8. Publish the definition precisely. Exact event, filters, time window, data source and owner. Ambiguity here produces three dashboards with three different numbers.
9. Review it once a year, or when the strategy changes. Not every quarter.

Before step 1, it is worth hearing customers describe the value in their own words. A short round of customer interviews often reveals that the value moment you assumed is not the one users care about.

Building the input metrics tree

The north star is an outcome. Teams cannot move it directly on a Tuesday afternoon. Input metrics are the levers they can move, and the tree shows how the levers connect.

A useful way to find inputs is to ask four questions about the north star:

• Breadth: how many users or accounts reach the value moment?
• Depth: how much value does each one get per session or order?
• Frequency: how often do they come back to it?
• Efficiency: how quickly and reliably do they get there?

Here is a worked tree for a fictional quick-commerce grocery app whose north star is weekly customers with two or more delivered orders:

North star: Weekly customers with 2+ delivered orders
│
├── Breadth: New customers completing a first order within 7 days of install
│     ├── Install-to-signup rate
│     └── Signup-to-first-order rate
│
├── Frequency: Share of last week's customers who order again this week
│     ├── Reorder from history usage
│     └── Repeat purchase of top 50 staples
│
├── Depth: Items per order from categories beyond the first one bought
│     └── Cross-category discovery on the home screen
│
└── Efficiency: Orders delivered within the promised time
      ├── Store pick-pack time
      └── Out-of-stock substitutions accepted

Guardrails: contribution margin per order, refund rate, rider attrition

Then give each input an owner. This is what turns the tree into a plan:

Input metrics also make prioritisation easier. When two projects compete, ask which input each moves and by roughly how much, then use a structured method such as RICE (see our guide to product prioritization frameworks). The north star and its inputs should appear in your product roadmap and in the success metrics section of every PRD.

North star metric examples from well-known companies

Companies rarely publish their internal north star, so most lists online repeat each other. The table below sticks to examples that are either published by the company itself or so widely cited that they are part of the standard teaching on the topic.

Amplitude's own story is instructive: it first used weekly querying users, then moved to weekly learning users after finding that retention tracked more closely with people sharing insights than with people running queries. That is step 5 of the process above in action.

Illustrative north star metrics for Indian companies

The examples below are not these companies' official metrics. They are plausible north stars, written to show how the method applies to business models familiar to Indian PMs. Use them as discussion prompts, for example in a product teardown or an interview exercise.

Notice that several of these deliberately bake in quality ("on time", "not returned", "successful"). That is a cheap way to reduce gaming before you even add guardrails.

North star metric anti-patterns

1. Revenue as the north star. Revenue is the outcome you want, but it lags and it can be grown in ways that destroy value (price hikes, aggressive upsells). Keep it as a business KPI.
2. Vanity counts. Downloads, registered users and page views go up and to the right while the product quietly fails.
3. Ratios. Conversion rate improves if you cut low-intent traffic. Use ratios as inputs.
4. A metric nobody can move. If no team can name an initiative that would shift it within a quarter, it is too far from the product.
5. Several north stars. "Our north stars are GMV, DAU and NPS" means you have not chosen. Large multi-product companies can have one per product, not three per product.
6. No guardrails. Optimising time spent without watching complaints, uninstalls or notification opt-outs leads straight to dark patterns.
7. Changing it every quarter. That is an OKR, not a north star. Frequent changes tell teams the metric is negotiable.
8. Confusing activation with the north star. Facebook's early "seven friends in ten days" insight is often quoted as a north star; it is better understood as an activation threshold, an input that feeds a broader engagement metric.

Should an early-stage startup have a north star metric?

Before product-market fit, the most important question is whether anyone keeps using the product. A cohort retention curve that flattens tells you more than any north star. Many founders do well to track one simple value action ("weekly teams who completed a report") and treat it as a working north star, without the full tree.

Once retention is stable and more than one team is working on growth, formalise the north star and its inputs. That is usually also when it becomes worth writing the definition down in your metrics doc and your glossary so new hires use the same language.

Frequently asked questions

What is a north star metric in simple terms?

It is the one number that best shows whether customers are getting real value from your product. If it keeps growing, the business should grow too. Airbnb's commonly cited example is nights booked.

What is the difference between a north star metric and a KPI?

A KPI is any metric the business tracks, from cash burn to support response time. The north star is one specific metric, chosen because it captures customer value and predicts long-term growth. All north stars are KPIs; very few KPIs are north stars.

Can a company have more than one north star metric?

A single product should have one. Large companies with genuinely separate products, such as a payments business and a lending business, can have one per product. If one product has several, the team usually has not made the strategic choice yet.

Is revenue a good north star metric?

Usually not. Revenue lags behind value and can be increased in ways that hurt customers, such as price rises that drive churn. Pick a value metric that predicts revenue, and keep revenue as a business KPI and guardrail.

How often should you change your north star metric?

Rarely: review it once a year or when the strategy genuinely changes, such as a new business model. Amplitude itself changed its north star from weekly querying users to weekly learning users after studying what drove retention, which is the right kind of reason.

Talk it through with people who have done it

Choosing a north star is as much a debate as an analysis, and it helps to hear how other PMs and founders have handled it. Product Circle's weekly AMA is a good place to ask exactly that kind of question, and our membership is built for founders and PMs working through these decisions. If you would like to be part of it, see how to join.

• Metrics
• Product strategy
• Growth