Tags: analytics commerce concept

North Star Metric

Date: 2026-08-17

---

One number nominated to represent whether the product is delivering value, used to align teams that would otherwise optimise against each other. It works as a coordination device and fails as a control system — because the moment it becomes the target, people find the cheapest route to moving it.

---

A North Star metric is a single metric chosen to represent the value customers get from the product, which every team’s own metrics are expected to feed.

What it’s actually for

Not measurement. Coordination.

WITHOUT                              WITH

acquisition team → sessions          everyone → "weekly purchasing customers"
merchandising    → add-to-cart
email team       → open rate         acquisition can't win by buying
CRO team         → conversion rate     traffic that never buys
retention        → repeat rate       email can't win by sending more
                                     CRO can't win by pushing a discount
each optimises locally, and            that destroys margin
several of them can win while
the business shrinks

The failure it fixes is real and common. Six teams hitting six targets while revenue falls is the normal state of a large organisation, and a shared number is the cheapest available fix.

What makes one work

A good north star sits between what you do and what you earn:

too far upstream          too far downstream
    sessions        →     ★ the north star     →     revenue
    page views                                       profit
    signups                                          share price

  moves easily, means      moves with what you       moves too slowly,
  nothing                  do AND predicts money     too many other causes

Properties worth insisting on:

• It reflects value delivered to the customer, not effort spent by you. “Orders delivered on time” beats “orders placed”
• It’s a leading indicator of revenue, verifiably — check the historical correlation rather than assuming it — Leading and Lagging Indicators
• A team can actually move it within a quarter
• It’s a rate or a count of people, not a total. Totals rise with traffic and hide per-user decline — Metric Design
• It’s hard to move by cheating. This is the property most often skipped and the one that decides whether it survives contact with incentives

Examples with the right shape for commerce: weekly purchasing customers, repeat purchase rate within 90 days, contribution margin per active customer.

Goodhart’s law, which is not optional

“When a measure becomes a target, it ceases to be a good measure.” Every north star degrades this way; the question is how fast and how visibly.

north star: "weekly purchasing customers"

the cheap routes to moving it
  · aggressive discounting            → customers up, margin down
  · free-shipping threshold removed   → orders up, contribution negative
  · a £1 add-on product promoted      → "purchasing customers" up 20%,
                                        revenue flat, fulfilment cost up
  · counting a subscription renewal   → definition change, not growth
    as a weekly purchase

Each of those is a rational response by someone being measured on the number. None of them requires bad faith — which is why “we’ll trust people not to game it” isn’t a control.

The counterweight is a small fixed set of guardrails, published alongside it and treated as vetoes rather than as context: contribution margin per order, return rate, and a service-quality measure. Same structure as an experiment’s guardrails, same logic — Guardrail Metrics, Contribution Margin.

The input tree

The number itself is useless to most teams — nobody can act on “weekly purchasing customers” directly. What makes it operational is decomposing it into inputs each team owns:

weekly purchasing customers
   │
   ├── new customers purchasing
   │     ├── qualified traffic          ← acquisition
   │     └── first-purchase conversion  ← CRO, merchandising
   │
   └── returning customers purchasing
         ├── active base                ← retention, lifecycle
         └── repeat purchase rate       ← merchandising, CRM

The tree is the actual deliverable, more than the top number. It shows each team what they own, how it connects, and — importantly — where two teams share an input and need to agree.

When it’s the wrong tool

• Businesses with genuinely different segments. One number across B2B and B2C, or across wholesale and retail, is an average that describes neither — Simpson’s Paradox
• Early or exploratory work, where what “value” means isn’t settled. A north star chosen too early locks in a wrong model
• As a target with a number attached. “Get north star to 45,000 by Q3” converts a coordination device into exactly the incentive Goodhart describes. Direction, not quota
• Where it replaces the financials. It’s a proxy. Contribution margin and cash are the actual outcomes, and the north star is only useful while it still predicts them — which needs re-checking annually

Where it interacts

• Vanity Metrics — the failure state; a north star that only ever goes up has become one
• Metric Design — the definition has to be written down and stable, or the number isn’t comparable across the quarters it’s meant to align
• Overall Evaluation Criterion — the experiment-level equivalent, and the same logic: one number, chosen in advance, to stop the result being negotiated
• Leading and Lagging Indicators — the property that makes a north star worth having rather than just a smaller version of revenue