North star metric: what it is and how to choose yours

A north star metric gives everyone on a team the same answer to "is the product getting more useful to more people?" It is not a replacement for revenue or for every other number you track. It is the number you reach for when two priorities compete.

A good one passes most of these tests:

• It reflects value delivered to customers, not value extracted from them. Time spent in an app can rise because the app is confusing; projects shipped with it rarely do.
• It leads revenue. When it rises this quarter, revenue should tend to rise later.
• The team can move it. Product, engineering, marketing and support should each be able to name something they could do to push it up.
• It is measurable often. Weekly is ideal. A metric you can only calculate once a year cannot steer anything.
• It is simple to say. If it takes a paragraph to explain, people will not use it to make decisions.
• It is hard to game. Counting sign-ups invites low-quality sign-ups. Counting activated accounts is harder to inflate.

Most good north star metrics combine a unit of value (a project, a transaction, a report) with a sign it was real (a collaborator, a completion, a return visit) and a time frame (weekly, monthly).

Revenue, churn and profit are lagging indicators. They are accurate, but they report the past. By the time monthly revenue dips, the customers who caused it stopped getting value weeks or months earlier.

A north star metric should be a leading indicator: a measure of behaviour that happens before the money moves. Customers who get value keep paying, expand and refer others. Customers who stop getting value leave at the next renewal.

This is why revenue itself usually makes a poor north star, even though it is what the business ultimately needs. It moves too late to act on, and it can be pushed up in ways that damage the product, such as raising prices or cutting a free tier, which feel like progress until churn catches up.

Keep watching the lagging numbers. They confirm whether your leading indicator is the right one. If the north star metric has been climbing for two quarters and revenue has not followed, the metric is measuring something customers do not value enough to pay for, and it needs changing.

A north star metric is too broad for any one team to own. Input metrics break it into parts a team can work on directly. A common way to split it is into breadth, depth, frequency and efficiency.

For a collaboration tool whose north star is weekly active projects with at least one collaborator, the tree might look like this:

Weekly active projects with at least one collaborator
├── Breadth: new projects created per week
│   └── sign-ups who create a first project within 7 days
├── Depth: share of projects with a second member
│   ├── invites sent per new project
│   └── invite acceptance rate
├── Frequency: projects with activity on 2 or more days a week
│   └── notifications that bring collaborators back
└── Efficiency: time from sign-up to first shared project

Each branch should be something a team can change with a specific piece of work: a shorter onboarding flow moves efficiency; better invite emails move depth.

Keep the tree small: three to five input metrics will do. If you have fifteen, you do not have a north star, you have a dashboard.

1. Write down the value in one sentence. What does a customer get when the product works? "Teams ship work together in one place" is a start.
2. Find the moment that value happens. Which action in your event data shows it actually occurred? This is often close to your activation event (see the aha moment guide).
3. Add a quality bar and a time frame. "Projects" becomes "projects with at least one collaborator, active this week".
4. Check it against retention and revenue. Using past data, confirm that accounts scoring high on the candidate metric retain and expand more than those scoring low.
5. Draft the input tree. If you cannot name three levers that move it, it is either too abstract or too narrow.
6. Agree it and publish it. Put it at the top of the weekly review with its input metrics underneath, and give each input metric an owner.
7. Review it yearly. Products change. A metric that suited a single-player tool may not suit it once teams are the main buyers.

• Choosing a vanity metric. Total sign-ups, page views or downloads only ever go up and say little about value.
• Picking a ratio. Conversion rates and percentages can improve while the business shrinks, for example when fewer, better-qualified visitors arrive. Prefer counts, and track ratios as inputs.
• Using revenue. It lags, and it is easy to raise in ways that hurt customers.
• Having several north stars. Two headline metrics that pull in different directions put you back where you started.
• Changing it every quarter. Teams stop trusting it. Change it deliberately, when evidence shows it no longer predicts retention or revenue.
• Forgetting the guardrails. Pair the north star with a few measures that must not get worse, such as support volume, error rate or refund rate, so it cannot be hit at the product's expense.
• Measuring it by hand. A metric someone recalculates in a spreadsheet when they remember will not steer anything.

OpenHelm has data tables that agents read and write, with charts, so a north star and its input metrics can sit next to the work meant to move them. Each agent carries goals with machine-checkable criteria, and every run is judged by a separate evaluator against the job's outcome contract, so progress is measured against the numbers rather than against an agent's own account of what it did. The journey analyst also maps your key journeys from GA4 events and files a task wherever users stall. Anything risky still waits for your approval, so the metric steers the agents without handing them the keys. Keeping the north star next to the work also makes it easier to notice when a busy week of activity did not move it at all. See the platform and user funnels.