Marcus Andrey Study
Ler em português Library

Framework · Measurement · 2 min read

Goals-Signals-Metrics

A three-step chain from a user goal to observable signals to trackable metrics, so you measure what matters instead of what is easy to count.

  1. GGoal

    What users should achieve, in plain words. No numbers yet.

    "Operators book quickly and correctly."
  2. SSignals

    Behaviour you could observe if the goal is met, or missed.

    Fewer corrections, faster completion
  3. MMetrics

    A number for each signal that you can track over time.

    Median booking time; % edited in 24h
Goals-Signals-Metrics works from what matters to what you can count, never the other way round.

Use it when

  • Someone asks "Which metrics should we track?" and the first answers are page views and clicks.
  • You are writing OKRs or a roadmap and need key results that reflect real user outcomes.
  • You need to justify every event in a tracking plan before it goes to data protection review.

How to use it

  1. Write the goal in plain words. Describe what users should be able to achieve, without numbers. Keep it to one sentence about the user, not the product. Output: one goal statement.
  2. List the signals. Ask: "If this goal were met, what would we see users do differently? And if it failed?" Write 2–3 observable behaviours. Output: a short list of signals.
  3. Turn each signal into a metric. Choose a number you can track over time, such as a median time, a percentage or a rate. Prefer metrics you can get without tracking individuals, such as aggregated analytics or support tickets. Output: 1–2 metrics per signal.
  4. Cut to the few that matter. Keep the metrics that would change a decision. Drop the rest. Output: 2–4 metrics with an owner and a source.
  5. Set a baseline and review. Measure the current value before you change anything, then review on a fixed rhythm. Output: a baseline value and a review date.

Worked example

An internal operations platform wants operators to book shipments faster without more mistakes.

Level Content
Goal Operators book shipments quickly and correctly
Signals Fewer corrections, faster completion, fewer abandoned bookings
Metrics Median booking time; % bookings edited within 24h; abandonment rate

Notice that "% edited within 24h" measures correctness, which a pure speed metric would miss.

Watch out for

  • Starting from the dashboard. Picking metrics because they already exist skips the goal. Write the goal first, even if the data comes from the same dashboard.
  • Vague goals. "Better UX" produces no signals. Make the goal specific enough that you could see it happen.
  • Too many metrics. Ten metrics for one goal means nobody watches any of them. Keep two to four.
  • Metrics about people. "Bookings per operator" turns a usability metric into performance monitoring. Measure the system.

Template

Copy this into your metrics doc and fill one block per goal.

Goal:      (what users should achieve, no numbers)

Signals:   (what we would see if it is working, or failing)
  1.
  2.

Metrics:   (a number per signal)        Source:        Baseline:
  1.
  2.

Owner:                 Review rhythm:
Source
Google (Rodden, Hutchinson & Fu)
Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close