One number that only improves when customers genuinely benefit. Choosing it badly points the whole company at the wrong thing.
10 min read · Level 3
The wrong north star actively causes harm, because a team will optimise it. Choose revenue and you will optimise for extraction; choose time-on-site for a tool whose value is saving time and you will make the product worse on purpose. The only safe test is whether the number can move without a customer being better off — if it can, someone will eventually move it that way, and they will be doing their job.
A north star metric is the single number that best captures the value customers get from your product. It aligns the whole team on one definition of progress.
The test: it should only go up when customers are genuinely better off.
Because a team without a shared definition of progress optimises locally — engineering ships features, marketing drives signups, sales closes deals, and none of it necessarily adds up.
And because the wrong north star actively causes harm. Choose revenue and you will optimise for extraction. Choose time-on-site and you will optimise for wasting people's time.
The scheduling product. Candidate north stars:
Signups — increases with marketing spend regardless of value. ❌ Revenue — can rise while customers get worse outcomes, through price rises or lock-in. ❌ Time in app — for this product, less time is better. Actively harmful. ❌ Rotas published per active team per week — only rises when teams genuinely use it to do the job. ✅
That last one has the right property: nobody can move it without customers actually getting value. Marketing cannot inflate it. A price rise does not touch it. Engineering can only move it by making the product better at its job.
Note the structure: an action, per customer, per time period. That shape stops it being gamed by simply having more customers.
First-time founders often pick revenue because it seems the most serious. Revenue is an outcome and it lags — it can rise for a quarter while the product deteriorates, and it tells engineering nothing about what to build.
The second mistake: picking something the team cannot influence. A north star that only moves with market conditions is demoralising and useless.
The third: changing it frequently. Its whole value is alignment over time. Change it when the business genuinely changes, not quarterly.
In one sentence, from the customer's point of view. The metric comes from this.
The thing a customer does when they are getting that value.
This prevents it being inflated simply by acquiring more customers.
Could someone increase this while making customers worse off? If so, choose differently.
Every function should be able to see how their work affects it.
Its value is alignment over time. Revisit annually, not quarterly.
Once you have a team and product-market fit. A solo founder pre-PMF does not need one.
Do not set one pre-PMF — you do not yet know what value you deliver, and a wrong north star at that stage points everyone confidently in the wrong direction.
A raw total rises simply by acquiring more customers, which makes it indistinguishable from a growth metric. Expressing the north star as an action per customer per period isolates whether each customer is genuinely getting more value — the thing you actually wanted to know.
"Rotas published per active team per week" cannot be inflated by acquisition. "Total rotas published" can, and will be.
The reliable way to test a candidate metric is to ask how someone could increase it while making customers worse off. If a plausible answer exists, the metric is unsafe — not because your team is cynical, but because incentives compound quietly over years.
Run the perverse-incentive test before adopting any headline metric. It takes five minutes and prevents years of drift.
Create a free account to read the rest and apply it to your own startup.