Which Metric Should This Business Care About
By the end of this lesson, you should be able to: apply two tests that kill most proposed North Star metrics, say why a marketplace and a subscription product need different ones, and work out whether a number is any good when nobody will hand you a benchmark.
The same quarter, scored four ways
Last lesson we pinned down how to define a metric precisely. That leaves the harder question: which one should Basket steer by.
Here's Basket's last eleven weeks, scored against four reasonable candidates.
| Candidate North Star | First week | Last week | Change |
|---|---|---|---|
| Active users | 8,603 | 17,300 | +101.1% |
| GMV | $243,742 | $420,413 | +72.5% |
| GMV per active user | $28.30 | $24.30 | −14.2% |
| Checkout conversion | 20.6% | 15.4% | −25.5% |
Pick GMV and you open the review announcing you grew the business by nearly three quarters. Pick checkout conversion and you open by explaining a decline. Nobody's lying. The company did all four of these at once.
Look at the red line. It tracks flat until the middle of April, then turns down and keeps going. Something happened to Basket in mid-April. Hold that thought. Finding out what it was is the job of the rest of this track.
So "what is our North Star" isn't an administrative question. It decides what the company notices.
Two tests every North Star has to pass
Most candidate metrics die on one of these two questions. Ask them in this order.
Test 1. If this number goes up, is the business genuinely better off?
Active users passes in theory. Now look at Basket again. Active users nearly doubled while GMV per active user fell 14%. All the growth is more people, and each one is worth less than the last. That's still growth, but it's the opposite of "our users became more valuable," and a metric that can't tell those apart is a weak steering wheel.
Checkout conversion fails the same test from the other side. Conversion can rise because you switched off a marketing channel and only the motivated visitors are left. The rate improves, the business shrinks. A rate on its own never passes test 1.
Test 2. Can the team actually move it?
A North Star nobody can influence is a weather report. Revenue fails this for most product teams, because pricing, sales and the economy swamp anything the product ships in a quarter. The fix isn't to abandon revenue. Hand each team the input they own, and keep revenue as the thing those inputs ladder into.
When an interviewer asks for a North Star, name the metric, then volunteer the guardrail that stops you gaming it. "Completed orders per active user, with delivery time as a guardrail so we don't win by promising deliveries we can't make." That sentence shows you've thought past the headline.
The pattern behind both tests: the wrong metric is the one you can buy. You can buy installs. You can buy signups. You can buy GMV with a big enough discount. You can't buy someone coming back next week because the product was good. When two candidates are close, pick the one that's harder to purchase.
Basket is a live example. Somebody bought a lot of users this quarter. Two of the four metrics rewarded it, one shrugged, and one caught it.