How to choose a North Star metric and the inputs that move it
A seven-step guide to picking one North Star metric, breaking it into input metrics with owners, and pairing it with a guard metric.

Someone has asked you to pick “the one metric” for your product or team. This guide is for the product manager who got that request and wants to answer it with more than a guess. It takes seven steps.
Who it is for
Product managers asked to name one metric for a product or a team.
What you will have at the end
One North Star metric, three to five input metrics with an owning team each, and a guard metric.
What a North Star metric is
A North Star metric is one number that captures the value customers get from your product. It leads revenue. It is not revenue.
That difference matters. Revenue tells you that customers paid. A North Star tells you whether they are getting what they paid for. If the second number is healthy, the first one has a reason to follow.
One number is not enough on its own. Under it sit a few smaller numbers that teams can move directly. Beside it sits a guard metric that keeps the main number honest. The steps below build all three.

The five tests a candidate must pass
Keep this list next to you. Every candidate metric gets checked against it in step 3.
- It reflects value delivered to the customer.
- It can be measured weekly.
- The team can influence it.
- It is not revenue itself.
- It is not a vanity number such as sign-ups or page views.
Write the value sentence
Finish this sentence: “Customers get value when ...”
Write what the customer does or gets, not what you earn. If you cannot finish the sentence without talking about payment, you are describing revenue. Start again.
Keep it to one sentence. If it needs two, the product may be doing two jobs, and you should pick the one that matters most for this team.
Draft three candidate metrics
Write three numbers that could stand for the value sentence. Three is enough to force a choice. With only one candidate you are just defending it.
Include the obvious one. It is usually the easiest number to count, and often the weakest. Seeing it next to better options makes that clear.
Test each candidate against the list
Take the five tests above and run each candidate through all of them. Write down pass or fail for every test, with one line of reasoning.
Be strict on the last two. A number that looks like growth but only counts sign-ups or page views fails, however good the chart looks. Revenue fails too, even though everyone cares about it.
Pick one
Choose the candidate that passes the most tests and sits closest to your value sentence. If two pass everything, take the one that is easier to explain to a new teammate.
If none pass, do not lower the bar. Go back to step 1 or step 2. The value sentence is often the real problem.
Break it into input metrics
A North Star is too big for one team to push on directly. Under it, define three to five input metrics that a team can move. Four questions help you find them:
- Breadth: how many customers?
- Depth: how much do they use it?
- Frequency: how often?
- Efficiency: how well does it work for them?
You do not need one of each. You need three to five inputs that, if they all improved, would move the North Star.
Give each input an owning team
Every input gets one owning team. One team, not two. When two teams share an input, each can assume the other is handling it.
Check that the team can change its input through its own work. If it cannot, the input is too far from the team, and you should redefine it.
Review the North Star and its inputs weekly
Look at the North Star and every input once a week, at the same time. That is why the weekly measurement test matters: a number you cannot read weekly cannot be reviewed weekly.
Ask which input moved, and why. Then check the guard metric, described next.
Add a guard metric
Pair the North Star with a guard metric so the North Star cannot be won by hurting customers. Any single number can be pushed up in ways that cost you elsewhere. The guard metric watches for that.
Pick one number that would get worse if customers were being harmed to lift the North Star. Put it on the same weekly review. If the North Star rises while the guard metric turns the wrong way, the rise does not count.
Worked example: a meal-kit subscription
Made up
This company is invented for the example. It sells a subscription meal kit, and the numbers below are not real data.

Step 1, the value sentence. Customers get value when they cook good meals at home without planning or shopping.
Steps 2 to 4, candidates and tests. Three candidates, checked against the list:
| Candidate | What the tests show | Result |
|---|---|---|
| Subscribers | Counts people, not value. | Rejected |
| Boxes delivered | Better, but a box can sit in the fridge uncooked. | Not chosen |
| Households cooking two or more kit meals a week | Measures the thing in the value sentence: meals cooked at home. | Chosen |
The chosen metric needs one more decision before it can be measured: how the company knows a meal was cooked. Pick one signal, such as a recipe marked done in the app, and keep it fixed so the weekly numbers stay comparable.
Step 5, input metrics. One input for each question:
Breadth
New households that cook their first meal
Depth
Meals cooked per household
Frequency
Weeks in a row with cooking
Efficiency
Share of deliveries on time and complete
Step 6, owners. Each of the four inputs gets one owning team: growth owns first meals, the product team owns meals per household and weeks in a row, and operations owns deliveries that arrive on time and complete.
Step 7, the weekly review. Each week the team reads the North Star, the four inputs and the guard metric together.
Guard metric
Refund rate. If more households are cooking but refunds are climbing, something is going wrong for customers, and the team should look at that before celebrating.
What you should have now
- A value sentence that starts “Customers get value when ...”
- One North Star metric that passed all five tests.
- Three to five input metrics that a team can move directly.
- One owning team for each input.
- A guard metric reviewed beside the North Star.
- A weekly review on the calendar.
If you can say all of that in one short paragraph, you have your answer to “the one metric”. If you cannot, go back to step 1 and rewrite the sentence.


