Choosing a North Star Metric for a Consumer Product

A north star metric is supposed to be the single number that tells you whether the product is working. Most companies pick one in an offsite, put it on a dashboard, and quietly stop looking at it within a quarter. Usually because they picked the wrong kind of number.

Done properly it's the most useful of all your product metrics: it resolves arguments, it makes prioritisation tractable, and it lets people make decisions without escalating. Done badly it's decoration.

What it has to do

A useful north star metric satisfies four conditions. Most candidates fail at least one.

1. It reflects customer value. It goes up when users get more of what they came for, not when they're extracted from more efficiently.
2. It leads revenue. It moves before the money does. A metric that lags revenue tells you what already happened, which is history rather than steering.
3. Teams can influence it. If no individual team's work visibly moves it, it can't guide decisions. Everyone will keep optimising their local metric and cite the north star in the preamble.
4. It's hard to game. Any metric under pressure will be gamed, usually without anyone intending to. The question is whether gaming it requires harming the user.

That last point is where most consumer metrics fail, and it's the one worth being paranoid about.

The metrics that go wrong

Time in app. The classic trap. You can increase it by making the product more valuable, or by making it slower, more addictive, and harder to leave. Both look identical on the dashboard. If your product's honest promise is to save someone time, this metric is pointing backwards.

Registered users. Cumulative, never goes down, therefore always encouraging and never informative. A number that cannot deliver bad news is not a metric, it's reassurance.

Sessions per user. Sometimes right — for a habit product, frequency is genuine value. Often wrong, because it rises when people can't find what they need on the first try.

Revenue. Tempting, since it's what actually matters. But it lags badly, it can be pumped for a quarter at the cost of the year, and no product team can trace their sprint to it. Revenue is the outcome; your north star should be the thing that causes it.

What good ones look like

The pattern in strong consumer north star metrics: they count a quality-qualified action, not an action.

Not "songs played" but songs played to completion. Not "listings viewed" but nights actually booked. Not "messages sent" but conversations with a reply. The qualifier is doing the work — it's the difference between measuring activity and measuring value delivered.

For most consumer products the shape is: number of users doing [core valuable action] at [meaningful frequency]. Weekly active users who complete a workout. Monthly buyers who reorder. It captures breadth and depth in one number, and it's difficult to inflate without genuinely helping someone.

One number is not enough

The north star is a direction, not a dashboard. Around it you need a small set of counterweights — metrics that catch the damage a single-minded push would cause.

If your north star is engagement, watch churn and support volume. If it's transaction volume, watch return rates and refunds. If it's activation, watch week-four retention, because you can always activate more people by lowering what activation means.

Three or four counterweights, reviewed alongside the main number, every time. The job of a counterweight is to say "yes, but" — and the review is worthless if nobody is expected to raise one.

Deriving yours

Don't brainstorm it. Work backwards from evidence.

1. Name the value. In one sentence, without jargon: what does a user get from this? If several people give different answers, that's the real problem and no metric will paper over it.
2. Find its observable trace. What does a user do when they get that value? That action is your candidate.
3. Test it against retention. Split old cohorts by whether they did the action. If the retained and churned groups look the same, the action doesn't represent value. Try again. This is the same technique as finding an activation metric, applied to the ongoing behaviour rather than the first one.
4. Test it against revenue. Does it lead? Look at whether cohorts with a higher rate of the action monetise better later.
5. Try to game it. Genuinely — spend twenty minutes as an adversary. How would you move this number without helping anyone? If the answer is easy, pick something else.

This is product analytics work rather than a workshop. It takes a couple of weeks, and it presumes instrumentation you can trust. If nobody believes the dashboard, fix that first — a north star built on unreliable events is worse than none, because it's confidently wrong.

Making it actually operate

Most north star metrics fail at adoption rather than selection. The number is fine; nothing changes because of it.

What makes it real:

• Every roadmap item names its relationship to it. Including "this doesn't move it, and here's why we're doing it anyway" — that's a legitimate answer for compliance work, technical debt, and table-stakes features. What's not legitimate is never asking.
• It's reviewed on a fixed cadence with the counterweights, whether or not it moved. Metrics reviewed only when they look good stop being metrics.
• One number, company-wide. Teams can still hold their own product KPIs underneath it, but if marketing, product and engineering each have their own north star, you have three departments and no direction.
• It changes when the strategy does. A north star for a company chasing product-market fit is not the one for a company scaling a proven model. Changing it deliberately is healthy; changing it because it stopped flattering you is not.

What it's really for

The point isn't the measurement. It's that a good north star makes saying no cheaper.

Without one, every feature request is a debate about opinions, and those are won by seniority and persistence. With one, the question becomes "how does this move the number," which is answerable, and answerable by anyone in the room.

That's the actual return: not a nicer dashboard, but a hundred small decisions a month that point the same way without anyone having to escalate them. If your team is relitigating the same priorities every sprint, this is often what's missing — happy to talk it through.