Level 1 · One north star, a few guardrails

What a north star is

A north star metric is the one number that best captures the value a business creates, and that every team can move. It sits at the top of the KPI sheet and settles arguments: if a change grows it without breaking anything else, the change was good.

For an online store, the candidates are volume (sessions, orders) or money (gross sales, net sales, net revenue). Money beats volume, because ten orders of $20 are not worth one order of $500. Among the money numbers, the honest one is what the store keeps:

gross sales     2,410,185.52    quantity × price paid, booked orders, 2025
− discounts        47,953.28    1.99% of gross
= net sales     2,362,232.24
− refunds         140,869.57    5.96% of net sales
= net revenue   2,221,362.67

Why one number is not enough

Any target, once people are paid to hit it, gets hit in the cheapest way available. This is Goodhart’s law: when a measure becomes a target, it stops being a good measure. For net revenue the cheap ways are known in advance:

• Buy sales with discounts. A sitewide 25% coupon raises orders this month and gives away margin. In 2025 discounts were 1.99% of gross sales; a jump to 5% would show up before the margin report does.
• Sell what comes back. Pushing products with high return rates inflates sales today and refunds next month. The return rate catches it.

Guardrail metrics

A guardrail is a metric you do not try to grow. You only watch that it stays inside a band while the north star moves. Good guardrails share three traits: they react quickly, they measure a cost or a harm the north star hides, and each has an owner who must explain a breach.

Typical store guardrails: discount rate, return rate, gross margin %, customer support contacts per order, page load time. Pick two to four. A sheet with fifteen “key” metrics has no key metrics.

Common mistakes

• A north star that can be bought. Sessions and sign-ups rise with paid traffic whether or not anyone buys.
• A guardrail as the goal. Minimising the return rate rewards selling less; it belongs beside the goal, not above it.
• No guardrails at all. Then the first time revenue jumps, nobody can say whether it was earned.

Interview note

“What would be the north star for an e-commerce business, and what would you watch alongside it?” is a classic product analytics question. Name a value metric, say how it could be gamed, and name the guardrail that catches each way.