Google's framework for choosing user-centred metrics across five dimensions — Happiness, Engagement, Adoption, Retention, Task Success — so you measure UX quality, not just raw usage.
▸ Try the interactive toolThe HEART framework, developed at Google, is a structured approach to selecting user-centred metrics across five dimensions of user experience: Happiness, Engagement, Adoption, Retention, and Task Success. It exists to help teams measure UX quality and product health beyond pure usage counts — to capture how well the product actually serves users, not just how much it's used.
HEART's real contribution is paired with the Goals–Signals–Metrics process: for each relevant dimension, you state the goal (what good looks like), identify the signal (the user behaviour that indicates it), and define the metric (how you'll measure that signal). Not every dimension applies to every product or feature — the framework's discipline is choosing which of the five matter for what you're measuring, then deriving real metrics rather than reaching for whatever's easy to count.
Happiness · Engagement · Adoption · Retention · Task success.
For each relevant dimension: state the goal → find the signal (user behaviour) → define the metric.
Teams default to measuring whatever's easy to count (usage, page views) rather than what actually reflects UX quality — HEART forces a more deliberate, user-centred choice.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Counting usage only | Raw usage misses happiness, task success, and real UX quality. | HEART measures the dimensions of experience, not just activity. |
| Easy-to-count metrics | Teams measure what's convenient, not what matters. | Goals–Signals–Metrics derives the right metric from the goal. |
| Applying all five everywhere | Forcing every dimension onto every feature dilutes focus. | HEART's discipline is choosing the dimensions that actually apply. |
| No link to a goal | Metrics with no stated goal can't be interpreted. | Starting from the goal makes each metric meaningful. |
Of the five — Happiness, Engagement, Adoption, Retention, Task Success — pick the ones relevant to what you're measuring. A new feature might emphasise Adoption and Task Success; a mature product, Retention and Happiness.
For each chosen dimension, articulate what success looks like in plain terms. The goal is the anchor that makes the eventual metric meaningful.
Find the observable user behaviour that indicates the goal is being met — the thing users do when the experience is good. Signals bridge abstract goals and concrete metrics.
Turn the signal into a specific, measurable metric. This is the number you'll track — derived deliberately from the goal, not grabbed because it was easy.
Track the metrics and read them against the goals they came from. A metric divorced from its goal is just a number; tied to its goal, it tells you about the experience.
A team measured a feature purely by usage — how many times it was used — and concluded it was successful because the count was high. But the usage number said nothing about whether users were actually succeeding at their task or were happy with the experience; high usage could even mean users struggling and retrying.
Applying HEART with Goals–Signals–Metrics, they chose Task Success and Happiness as the relevant dimensions, stated goals (users complete the task efficiently; users are satisfied), found signals (task completion without errors; positive feedback), and derived real metrics. The fuller picture showed that high usage was partly users retrying after failures — a task-success problem the raw count had hidden as apparent success.
The deliverable is, for each chosen dimension, a goal → signal → metric chain — a deliberate, user-centred set of measures.
| Dimension | Measures | Example signal |
|---|---|---|
| Happiness | Satisfaction, attitude | Survey scores, feedback |
| Engagement | Depth of involvement | Frequency, intensity of use |
| Adoption | Uptake of new | New users of a feature |
| Retention | Continued use | Return rate over time |
| Task success | Effectiveness | Completion rate, error rate |
HEART exists to stop teams from mistaking activity for quality. Its Goals–Signals–Metrics discipline ensures you measure what good experience actually looks like — derived from a goal — rather than whatever happens to be easy to count.
The two disciplines that make HEART valuable are selection and derivation. Selection: not every dimension applies to every product or feature, so the framework forces a deliberate choice of which of the five matter here — resisting the urge to measure all five badly. Derivation: the Goals–Signals–Metrics chain ensures each metric traces back to a goal, so it can actually be interpreted, rather than being a convenient number with no meaning attached. Together these guard against the most common UX-measurement failure — grabbing usage counts because they're available and reading them as quality. Used well, HEART captures whether users are happy, succeeding, and coming back, which is what 'product health' actually means.
Activity isn't experience. Use HEART to measure the dimensions that reflect real quality.
Convenient numbers aren't meaningful ones. Derive metrics from goals via Goals–Signals–Metrics.
Not every dimension applies everywhere. Choose the relevant ones deliberately.
A metric divorced from its goal can't be interpreted. Always start from the goal.
HEART supplies the product-level UX metrics that sit in the hierarchy (Tool 02).
The 'E' in HEART connects to the breadth/depth engagement work (Tool 10).
HEART metrics can become input metrics in the tree (Tool 05).
Deriving metrics from goals mirrors outcome-based OKRs (Module 2).
Pick a feature and one HEART dimension that matters for it (say, Task Success). State the goal, identify the signal (what users do when it's working), and define the metric.
Compare your derived metric to the raw usage count you might have grabbed instead. What does yours capture that the count misses?
If your goal-derived metric reveals something the usage count hid — like users succeeding versus just clicking — you've seen why HEART measures experience rather than activity.