How to Choose a North Star Metric

The danger with a North Star Metric is not having no metric; it is choosing one that makes the team feel focused while hiding the truth. A useful metric represents recurring customer value the product creates — and needs guardrails around it so the team does not learn to game its own scoreboard.

What a North Star Metric is — and what it is not

A North Star Metric is a single focus metric intended to represent recurring value the product creates for customers, translated into a number the team can align around. Done well, it connects product usage, customer value, and business direction — so that when it moves in the right direction, real things are happening, and when it stagnates, real things are broken.

The framing traces back to Sean Ellis and Morgan Brown’s Hacking Growth, which treats the North Star as the top of a small tree of metrics rather than as a dashboard. Two distinctions the honest version enforces:

• It is not the only metric. A team that reads a single number stops seeing the shape of the business.
• It is not a dashboard replacement. The North Star gives direction; the dashboard tells the team what to work on this week.
• It is not automatically revenue, signups, traffic, or active users. Any of those can be the right North Star for a specific product — and any of them can be exactly the wrong number to organise around.

Why North Star Metrics fool startups

The failure modes are consistent enough to name. Each one describes a metric that grows while the underlying business does not.

• The metric is easy to grow but not meaningful — the team pushes on it because it moves, not because it matters.
• The metric tracks activity without value — a “session” or an “action” that the buyer does not actually value.
• The metric ignores retention or quality — the number goes up while the same users leave a week later.
• The metric is too lagging — it moves after the decisions that mattered are already made.
• The metric rewards the wrong behaviour — the team optimises for the number in ways the buyer would not thank them for.
• The metric hides segment differences — a healthy segment carries a struggling one, and no one notices until it stops carrying.
• The metric is copied from another company — a famous NSM from a different product and different stage.
• The metric is chosen for investors instead of operating truth — a number that sounds impressive at a board meeting and useless on a Tuesday.

Naming these before choosing the metric is cheaper than surviving them afterwards.

The North Star decision framework

This is an author’s editorial framework — not an external standard. Nine inputs I work through with founders before we agree on a metric.

1. Customer value — the specific value the product delivers, in the buyer’s terms, not the team’s.
2. Recurrence — how often that value is delivered; a metric that ignores repetition rewards a one-time win.
3. Product behaviour — what a user actually does in the product when the value is delivered.
4. Business model — the way the startup makes money; the metric has to be reconcilable with that.
5. Stage — early-discovery, MVP, first revenue, growth, multi-product; the appropriate NSM changes.
6. Segment — which buyers the metric represents; averages over dissimilar segments lie.
7. Input metrics — the actions the team can influence weekly that move the NSM over months.
8. Guardrails — the metrics that catch a team from optimising the NSM while damaging the business.
9. Review cadence — the moment the team commits to revisiting whether the metric still tells the truth.

The table below is the one I use to keep the metric honest by startup type. It is not a ranking or a benchmark — it is a set of trade-offs to weigh.

Start with customer value, not internal activity

The strongest defence against a misleading NSM is to start from what the buyer would say the product does for them. Teresa Torres’ Continuous Discovery Habits keeps the conversation honest because it insists that discovery is a rhythm — the team keeps talking to buyers, keeps updating what “value delivered” looks like, and only then translates that into a metric. Signups, traffic, installs, and page views are not automatically vanity, but they often need caution: they measure activity that may or may not be tied to value.

Two disciplines protect the choice. First, name the value in the buyer’s terms — the outcome they would describe to a colleague, not the feature the team ships. Second, ask what behaviour in the product happens when that value is delivered — a workflow completed, an artefact created, a task automated, a purchase repeated, a conversation had. The NSM should reflect that behaviour, not the surrounding noise.

Match the metric to the business model

A metric that ignores how the startup makes money will drift from the business. The illustrative pairings below are hypothetical, not case studies:

• B2B SaaS — a retention-flavoured NSM (accounts / seats using the core workflow) usually beats raw signup counts.
• Product-led SaaS — an activation-flavoured NSM (weekly active users completing the qualifying action) can lead paid conversion by weeks or months.
• Marketplace — successful transactions in the segments the team can serve tend to beat aggregate GMV, which can mask cold starts.
• E-commerce — repeat purchase behaviour tends to reveal product-market fit that first-purchase revenue can hide.
• Subscription product — actively renewing users tied to real feature usage is more honest than “paying customers” alone.
• Usage-based product — consumption on real customer workloads is more honest than total consumption, which can be inflated by test data or team-driven runs.
• Agency or service-enabled startup — the operating NSM is often engagement quality (repeat retainers, referral rate) rather than volume of leads.
• Developer tool — successful integrations still running after a set window separates real adoption from installs.
• Content or community product — members returning to contribute or consume reflects the underlying loop; traffic alone rarely does.

Pair the North Star with input metrics — and guardrails

The NSM gives direction; input metrics tell the team what to act on this week. Croll and Yoskovitz’s Lean Analytics is useful precisely because it treats “one metric that matters” as the top of a hierarchy — with a small number of inputs the team can genuinely move, not a dashboard of thirty. Illustrative inputs, without benchmark numbers:

• Activation steps completed inside the first meaningful window.
• Onboarding completion for a defined segment.
• Qualified usage — an action the team can honestly say represents intent.
• Retained users or accounts as a cohort, not a running total.
• Created projects, workflows, or artefacts that outlive the first session.
• Invite or collaboration actions, when the product genuinely benefits from more people in it.
• Repeat purchase or repeat use, on a cadence that matches how buyers naturally return.

Guardrail metrics are what keep the team from cheating itself. Even a well-chosen NSM can be pushed in ways that damage the business; guardrails catch that. Practical examples:

• Retention by cohort — is the number growing while retention decays?
• Quality signals — task completion, error rate, refund rate.
• Churn — voluntary and involuntary — and the reasons behind each.
• Support burden — is the metric growing because the product is more useful, or because the team is heroically saving accounts?
• Margin and unit economics — the metric may move while unit economics tighten uncomfortably.
• Customer-satisfaction signals — CSAT, NPS-shaped notes, review sentiment, hand-collected verbatims.
• Abuse or spam signals for products where those are possible.
• Sales quality — are the deals closing profitable, or are we buying growth we cannot serve?
• Segment health — is any segment being sacrificed to make the aggregate look healthier?
• Revenue quality — durable ARR versus one-off contracts marked as recurring.

NSM by startup stage, and how to test whether the metric is lying

Stage changes the sensible answer. In the discovery stage qualitative learning is often more useful than any NSM; declaring one too early forces the team to score before there is a game. At the MVP stage an activation- or meaningful-use metric tends to be more informative than scale. At the early-revenue stage, quality and repeatability of the first deals matter more than volume; a revenue NSM can drift into cash-collection theatre if learning is not attached. At the growth stage the NSM needs segmentation and guardrails to stay honest as the business gets more complex. At the multi-product stage one metric can become too simplistic — the team may need one NSM per product line, held together by a broader business commitment.

Testing whether the current metric is lying is a small discipline done regularly. Eric Ries’ Lean Startup underlines the frame: measure things that lead to validated learning, not things that make the team feel busy. Practical checks:

• What behaviour does the metric reward? Would that behaviour make a specific buyer happier? If not, the metric may not be pointing at value.
• Inspect segments — an average across three ICPs can hide a struggling one; break the metric down by the segments the business actually cares about.
• Compare metric movement to customer conversations — if the number is climbing but sales calls sound the same or worse, the number is not describing the reality the team feels.
• Check whether revenue or retention improves qualitatively — the qualitative smell test often beats one more chart.
• Look for gaming — if a team member can plausibly explain how they could move the metric without producing customer value, the metric is exposed.
• Review what the team stops doing because of the metric — a metric with no falsification condition is decoration.
• Identify leading vs lagging signals — a lagging NSM without leading input metrics is a scoreboard the team cannot influence.
• Revisit after major GTM or product changes — the same metric can quietly stop describing the business after a repricing, a new segment, or a product expansion.

Common mistakes worth refusing plainly: copying another startup’s NSM without their conditions; choosing signups because they are easy; choosing revenue too early without the learning context to interpret it; ignoring retention; ignoring the buyer / user split (the buyer signs, the user uses — one metric may not describe both); hiding weak activation behind aggregate usage; tracking a metric no team member can influence; changing the NSM every week; and treating the NSM as investor storytelling rather than operating truth. If you want a calm review of whether your NSM reflects real customer value, has useful input metrics, and is protected by retention, quality, and revenue guardrails, that is what my marketing audit and consulting engagement covers. Longer write-ups on positioning, growth, and content strategy live in Insights, and the AI Lab contains practical templates worth using once the metric shape is stable.

Frequently Asked Questions

Is there a best North Star Metric for startups?

No. A useful North Star Metric reflects the specific customer value the product delivers, the way that value repeats, the business model, and the stage. A metric that fits one product and stage rarely fits another. The right choice is a decision — not a lookup.

Should our North Star Metric be revenue?

Sometimes — for products whose revenue directly reflects delivered value at the cadence value repeats. For many early-stage products, revenue alone lags behind the behaviour that actually predicts a durable business. A revenue-only NSM can quietly turn into cash-collection theatre if learning is not attached.

Are signups always vanity?

No. Signups are vanity when they are disconnected from value or retention. A signup that comes with genuine intent and reliably converts to durable usage can be a leading signal worth watching — usually as an input metric, not as the North Star itself.

How often should we change the North Star Metric?

On evidence, not on a calendar. Revisit when the stage changes, when a major GTM or product decision happens, or when the metric stops moving with the reality the team feels. Changing the NSM every week is not honesty; it is drift.

Can we have more than one North Star Metric?

At the earliest stages, no — the point of one number is to force a decision. Later, at multi-product stage, one metric can become too simplistic; some companies use one NSM per product line under a broader business commitment. Complexity should be earned, not defaulted to.