Most product organizations measure what is easy: tickets closed, story points delivered, features shipped. These are output metrics. They tell you how busy the system is, not whether it is producing value.

An outcome metric describes a change in customer behavior or business health. It is harder to measure and far more useful for decisions.

The three-layer model

We use three layers, each with a different audience and cadence:

  • Business outcomes (quarterly): revenue retention, expansion, activation, support cost per account.
  • Product outcomes (monthly): the specific behavior change the product is designed to create.
  • Delivery health (weekly): lead time, failure rate, and reliability — the system's ability to produce change safely.

Teams own product outcomes. Leadership owns business outcomes. Delivery health is a shared constraint, never a performance ranking.

A good product outcome is a sentence

Write outcomes as a change in behavior with a population and a timeframe: "Increase the percentage of new workspace admins who invite a second member within 14 days from 28% to 40% by Q1." This is testable and specific. "Improve onboarding" is not.

Pair every outcome with a guardrail metric. If activation rises but week-four retention falls, you have optimized the wrong thing. Guardrails keep local optimization from damaging the whole system.

Velocity is a property of the system, not a virtue of the team. Measure it to improve flow, never to rank people.

Retire metrics deliberately

Metrics accumulate like unused dashboards. Once a quarter, ask of each metric: what decision does this change? If the answer is "none," archive it. A smaller set of trusted measures is more powerful than a comprehensive set nobody believes.

Connect the narrative

Numbers alone do not create alignment. Each review should connect three things: the customer problem, the bet the team made, and the evidence observed. When the evidence contradicts the bet, say so explicitly. The organization learns more from a well-documented miss than from a vague success.

Output metrics will always be easier to collect. The discipline is to keep them in their proper place: as inputs to the real conversation about whether customers are better off.