Choosing a north-star metric

#What this question is really testing

"What is the north-star metric for this product?" "How would you measure the success of our app?" "If you could track only one number, which would it be?" These questions come up in metrics rounds, at the end of product sense rounds, and sometimes as a warm-up.

The interviewer is not checking whether you know the company's internal metric. You almost certainly don't, and companies rarely publish it. They are checking three things:

• Can you connect what a product does for people to a number?
• Can you see how that number could mislead or be gamed?
• Can you defend your choice when pushed, without caving or digging in?

#What a north-star metric is

A north-star metric is the one number that best shows whether users are getting the product's core value, chosen so that when it grows, the business grows with it over time.

Three parts of that definition do the work.

• Users getting value. It counts something users do because the product helped them, not something the company does to them. "Emails we sent" is not a north star. "Lessons learners finished" might be.
• Core. It reflects the main reason the product exists, not a side feature.
• Over time. It should lead revenue and retention, not just mirror this month's revenue.

A north star does not replace other metrics. Teams still track revenue, cost, quality and reliability. The north star is the one number everyone agrees to move. The others keep it honest.

#The usual wrong answers

• Revenue. It is a result of value, it lags, and it can be raised in ways that hurt users, such as price rises or more ads. Revenue is a business goal, not a north star.
• Downloads or registered users. These only go up. A product can have 50 million sign-ups and be quietly dying.
• Daily active users, with no qualifier. Closer, but "opened the app" says little. A push notification that gets someone to open and close the app counts the same as a real session.
• Time spent, for a tool. For a maps app or a tax app, less time often means a better product.

#The Value Moment method

This kit uses one method for every "which metric?" question: north stars, feature goals and trade-off calls. It has five steps.

1. Name the value. One sentence: who gets what from the product. For a marketplace, name both sides.
2. Find the moment. The specific action that shows a user just got that value. Not "opened the app" but the thing that happened because the app worked.
3. Make it a number. Decide what you count (actions, users, teams), over what window (day, week, month) and with what qualifier. Match the window to how often the value naturally happens.
4. Stress-test it. Ask four questions. Does it rise only when users get more value? Could a team push it up in a way that hurts users? Does it lead to revenue and retention? Will it move within weeks, so a team can steer by it?
5. Guard it. Name one to three guardrail metrics that would catch the damage if a team chased the north star too hard.

#Worked example: a fitness class app

Pulse is a fictional app where people book drop-in fitness classes at local studios. Pulse earns a fee on each booking.

Name the value. Members find and attend classes they enjoy without buying a membership at one studio. Studios fill spots that would otherwise sit empty.

Find the moment. The member shows up and takes the class. A booking is only a promise. Attendance is the value for both sides: the member worked out and the studio filled a spot.

Make it a number. Classes attended per week, across all members. Weekly fits because most people exercise on a weekly rhythm, and a daily count would swing every weekend.

Stress-test it.

• Rises with value? Yes. More attended classes means more workouts for members and more paid spots for studios.
• Can it be gamed? A team could buy attendance with deep discounts. The number rises while the business loses money.
• Leads revenue and retention? Pulse earns per booking, and it is reasonable to assume members who attend often are the ones who stay. In an interview, say this out loud as an assumption.
• Moves within weeks? Yes, it is counted weekly.

Guard it.

• Revenue per attended class, to catch growth bought with discounts.
• Late-cancel and no-show rate, to protect the studios.
• Average class rating, to catch growth that comes from low-quality studios.

Final answer: "Weekly attended classes, guarded by revenue per class, no-show rate and class rating."

Notice what the method did. It moved the answer from bookings, the obvious choice, to attended classes, because it asked where the value actually lands. That one move is often what separates a strong answer from an average one.

Key idea The north star sits at the value moment, the action that shows a user just got what they came for. Find the moment first and the metric usually follows.

#Patterns by product type

Different kinds of product tend to land on different shapes of north star. Treat this table as a starting point for your own reasoning, not as the metrics any real company uses.

For work tools, count teams or accounts rather than people when a company is the buyer. A company renews because a team relies on the tool, not because one person logs in.

#How to say it in the room

Keep the answer short and announce the structure first.

You I'll name the value this product creates, find the moment a user gets it, turn that into a number, test it for weak spots, and add guardrails. Then I'll commit to one metric.

Then do it out loud in about two minutes. Close with the metric and its guardrails in one sentence.

Expect the follow-up "What could go wrong with that metric?" and answer it from your stress-test. Expect, too, that the interviewer may suggest a different metric. Don't fold at once and don't dig in. Say what their metric captures that yours misses, then decide.

You Bookings would show demand a few days earlier, which is useful. I'd still steer by attendance, because a booking that ends in a no-show costs the studio. I'd track bookings as a leading input to attendance.

Try it Pick an app on your phone. Run the five steps out loud in under three minutes. Then answer the hardest question about your metric: how would a team cheat it?

The next lesson shows how to break a north star into the smaller numbers teams actually move: Metric trees: breaking a goal into parts. For finishing a design answer with metrics, see Close with how you would measure it.

#Recap

• A north star is the one number that best shows users getting core value, chosen so that business growth follows.
• Revenue, downloads and unqualified daily users are the usual wrong answers, each for a clear reason.
• The Value Moment method: name the value, find the moment, make it a number, stress-test it, guard it.
• Finding the moment is where strong answers pull ahead, for example attendance rather than bookings.
• Always finish with one metric plus one to three guardrails, and know how your metric could be gamed.