Choosing a North Star Metric You Can Actually Move

Most north star metrics I have seen in the wild are posters, not targets. They live on a slide, they get quoted in the all-hands, and nobody on the team can tell you what they did last week or why. A north star metric is supposed to be the single number a growth team organizes around, the thing that tells everyone whether the product is winning. When it is chosen badly, it does the opposite: it points people at work that feels important and moves nothing.

I have picked, inherited, and quietly retired a fair number of these. At Chegg I owned a landing system of 200+ pages and ran experimentation through Optimizely, which meant I was answering a very concrete version of this question every week: of all the numbers I could chase, which one actually reflects value delivered, and which one can this team move with the levers in front of it? Those are not always the same number, and the gap between them is where most growth programs quietly lose the plot.

This post is about closing that gap. What a north star metric is and is not, why it is not the same thing as revenue, what makes a good one, and how to decompose it into the input metrics your teams own day to day so the number on the slide connects to the work in the sprint.

What a north star metric actually is

A north star metric is the one measure that best captures the value your product delivers to customers, chosen because moving it reliably moves the business behind it. It is a proxy. You cannot put “customers got value” on a dashboard, so you find the observable behavior that stands in for it most honestly, and you point the team at that.

The word “single” matters. The whole reason to name a north star is to force prioritization across a group of people who otherwise optimize their own local numbers. Marketing wants traffic, sales wants pipeline, support wants ticket volume down. A north star metric is the thing that sits above all of those and lets you say, credibly, that a given piece of work either serves the number or does not. If you cannot use it to kill a project, it is not doing its job.

What it is not: it is not a scorecard, it is not the only number you watch, and it is not a vanity figure that only ever goes up. Registered users, total downloads, cumulative signups, page views. These climb no matter what you do because they never subtract. A metric that cannot go down cannot tell you when you are getting worse, which makes it useless as a compass. If your north star only rises, you have chosen a trophy, not an instrument.

Why a north star is not revenue

The most common mistake is picking revenue itself. It feels rigorous, because revenue is what the business ultimately runs on, so surely the truest number to chase is the money. The problem is that revenue is a lagging indicator, and a north star is meant to lead.

A good north star metric leads revenue by capturing delivered customer value before that value shows up as money. When customers get more of what the product promises, they buy, renew, expand, and refer, and revenue follows weeks or months later. If you steer by revenue directly, you are steering by the wake behind the boat. You find out you were off course long after the correction would have helped, and by then the causes are tangled up with pricing changes, seasonality, and deals that closed for reasons unrelated to product quality.

Revenue is also too shared and too downstream for one team to own. Half a dozen functions touch it, which means no single growth team can point at it and say “we moved that.” I make the fuller version of this argument in own the number, not the task: the thing you steer by has to be close enough to your levers that your work visibly bends it. Revenue rarely is. A well-chosen north star is upstream of revenue and correlated with it, which gives you both an early signal and a number your team can actually be accountable for.

The four properties of a good north star

When I evaluate a candidate metric, I hold it against four tests. A number that fails any of them will eventually mislead the team.

It reflects real customer value. The metric should go up only when customers are genuinely better off. This is the honesty test. Ask whether the number could rise while customers get a worse experience. If the answer is yes, you have a proxy that can be satisfied without delivering anything, and teams will eventually satisfy it the cheap way.

It is a leading indicator. It should move before revenue does, and changes in it should predict changes in the business. This is what makes it useful for steering rather than reporting. If the metric only confirms what the P&L already told you, it is a lagging number wearing a north star costume.

Your team can actually influence it. This is the property that gets skipped, and it is the one this whole post is named after. A north star you cannot move on a reasonable timescale is a source of guilt, not direction. The levers you control (onboarding, activation, key-action frequency, retention mechanics) need to connect to the metric within a horizon short enough that a quarter of work produces a visible change.

It is simple enough to rally around. Everyone from an engineer to the CEO should be able to state it in one sentence and understand why it matters. If explaining the metric requires a paragraph and a footnote, it will not organize behavior, because people cannot orient toward a number they cannot hold in their head.

Notice that a metric can be a perfect measure of value and still be a bad north star, because your team cannot move it fast enough to learn from. The art is finding the number that scores well on all four at once, not the one that is theoretically purest.

What good ones look like across business types

The right north star depends on how your product creates value, so the examples that get quoted famously are useful mostly as patterns to reason from.

For a consumption product, the value shows up in usage. Something like credits consumed, messages sent, or core actions completed per active user tracks whether people are actually getting the thing they came for. I have worked on credit-based monetization where consumption was the truest early signal of value, well ahead of the revenue it eventually produced.

For a collaboration or team product, the value compounds when a whole team adopts it, so weekly active teams (not seats, teams) tends to beat raw user counts. One committed user is fragile; a team that has wired the product into its workflow is durable, and the metric that counts teams reaching a real activity threshold captures that durability.

For a marketplace or supply product, you want the number that represents a completed exchange of value. The nights-booked style of metric is the classic here: not listings, not searches, but the transaction where both sides got what they came for. It sits right at the point where value is delivered, which is exactly where a north star belongs.

The pattern across all of these is the same. Find the moment the customer gets the value your product exists to deliver, then count how often that moment happens across your active base. That is almost always a better north star than any total, any cumulative count, or revenue itself.

The metric tree beneath the star

A north star metric on its own does not tell anyone what to do on Monday. It becomes operational when you decompose it into the input metrics that feed it, the smaller numbers that individual teams actually own and move. I think of this as the metric tree, and building it is where a north star earns its keep.

Take weekly active teams as the star. Underneath it sit inputs like new teams activated, the rate at which activated teams stay active week over week, and the rate at which dormant teams come back. Each of those decomposes again. Activation breaks down into signup-to-first-action, first-action-to-habit, and the specific setup steps that predict a team sticking. Now you have numbers small enough that a team can own one, run experiments against it, and see movement inside a sprint. Activation in particular deserves its own treatment, which is why I keep activation metrics as a separate discipline rather than folding them into the star.

The tree does two things. It tells you where the constraint is, because you can look down the branches and find the input that is leaking, and it gives every team a line of sight from their daily work up to the number the company cares about. A designer improving an onboarding step can see exactly which input they are pushing and how that input rolls up. This is also how the north star connects to planning: the roadmap should be organized around the inputs in the tree, not around a wish list of features, which is the argument I make in detail in building a growth roadmap.

One metric or a small set

People argue about whether you should have one north star or several, and the honest answer is one primary metric with a small supporting set. The single number is non-negotiable for the alignment reason: the entire point is to force a shared priority, and two north stars is zero north stars, because the moment they conflict you are back to everyone arguing their local optimum.

But one number in isolation is dangerous, which is why the metric tree beneath it and the guardrails around it exist. The tree gives you the inputs to diagnose and act on. The guardrails, which I will come to next, keep the pursuit of the star from breaking things it should not. So the shape is a pyramid: one metric at the top that everyone can name, a handful of input metrics below it that teams own, and one or two counter-metrics off to the side that keep everyone honest. That is a small enough set to hold in your head and rich enough to actually run a program on.

Goodhart’s law and the guardrail metric

“When a measure becomes a target, it ceases to be a good measure.” Goodhart’s law is the single most important thing to understand about north star metrics, because the entire premise of naming one is to make it a target, which is exactly the condition under which it starts to lie.

The mechanism is simple. Once a team is pointed hard at a number, it will find the cheapest way to make that number move, and the cheapest way is often not the way you intended. Point people at messages sent and they may fragment one message into five. Point them at weekly active users and they may hound people with notifications until the number rises and the goodwill collapses. The metric goes up, the value goes down, and the dashboard looks great while the product quietly gets worse.

The defense is to pair the north star with a counter-metric or guardrail, a number that captures the thing you must not sacrifice in pursuit of the star. If the north star is active usage, the guardrail might be a retention or satisfaction measure that would fall if you were juicing activity with dark patterns. If the star is conversion, the guardrail might be refund rate or downstream retention of the customers you converted. When I lifted conversion 34% on the Chegg landing system, the number I watched just as closely was what happened to those converted users afterward, because a conversion win that produces churn is not a win, it is a transfer of the problem downstream. The guardrail is what lets you push the star aggressively without wrecking the business behind it. Lifecycle mechanics are a common place for guardrails to live, which is part of why I treat lifecycle and repeat revenue as inseparable from the growth number rather than a separate concern.

Getting the org to actually agree on it

A north star metric only works if the organization genuinely shares it, and shared is a higher bar than announced. I have seen leadership declare a north star that the teams politely ignored because it was chosen in a room they were not in, against levers they did not control.

Getting real alignment takes a few things. The metric has to be co-owned, which means the people expected to move it were part of choosing it and believe it reflects their work. It has to be visible, on a dashboard everyone reads, updated often enough that people build a feel for it. And leadership has to actually use it to make decisions, because the fastest way to teach a team that the north star is theater is to override it whenever it is inconvenient. When engineering can see that the experiment they shipped moved the number three points, the metric stops being a management artifact and becomes something they own. That transparency is what turns a group of people with separate incentives into a group pointed at the same target.

When to revisit or change it

A north star metric is not permanent, but it should be stable, and holding both of those at once is a matter of judgment. Change it too often and it never organizes anything, because a target that moves every quarter is just a mood. Refuse to ever change it and you will keep steering by a number that has stopped describing your business.

The signals that it is time to revisit are worth naming. The metric has saturated, so it keeps climbing while the business stalls, which means it has decoupled from value. Or the company’s strategy has moved, from acquisition to retention, from single-user to team adoption, and the old star measures a stage you have left. Or you have simply learned that your proxy is not as honest as you thought, that it can rise while customers are worse off. When one of those is true, change it deliberately, explain why, and give the new metric time to bed in. What you should never do is quietly swap the number because the current one is embarrassing this quarter. The point of a north star is to be uncomfortable when the news is bad. A metric you abandon the moment it stings was never really your north star.

The short version

A north star metric is the single proxy for delivered customer value that your team organizes around. It is not a vanity total and not a scorecard.
It is not revenue. Revenue lags and is too shared to own; a good north star leads revenue by capturing value before it becomes money.
A good one passes four tests: it reflects real customer value, it is a leading indicator, your team can actually move it, and it is simple enough to rally around.
Choose by finding the moment your product delivers value and counting how often it happens: credits consumed, weekly active teams, completed transactions.
Decompose the star into a metric tree of input metrics that individual teams own, so daily work connects to the number.
Keep one primary metric, not several, supported by a few inputs and one or two guardrails.
Pair the star with a counter-metric to defend against Goodhart’s law, because any target you push hard will get gamed unless something guards the thing you must not sacrifice.
Get genuine org agreement by co-owning, making it visible, and using it to make real decisions.
Revisit it when it saturates, when strategy shifts, or when you learn the proxy is dishonest, but never swap it just because the news is bad.

I am Deepanshu Grover, a Growth Product Manager in Paris. If your team is chasing a metric it cannot actually move, connect on LinkedIn or get in touch.