Skip to main content
The HEART framework was developed at Google by Kerry Rodden, Hilary Hutchinson, and Xin Fu as a way to measure user experience with quantitative metrics at scale. It’s useful when you need to track UX quality across a large product without relying solely on qualitative research 💛

​
The five dimensions

Happiness - subjective satisfaction. Typically measured through surveys: NPS, CSAT, or app store ratings. The one dimension that requires asking users directly rather than observing behaviour. Engagement - depth and frequency of interaction. How often do users come back? How much of the product do they use in a session? Metrics like sessions per user per week, features used per session, or time spent in core workflows. Adoption - uptake of new features or by new users. What percentage of users have tried the new flow? How quickly are new sign-ups reaching key milestones? Useful for measuring whether launches actually land. Retention - users returning over time. The percentage of users who come back after day 1, day 7, day 30. The most reliable signal that your product has genuine value. If retention is flat, no amount of acquisition will save you 💡 Task success - can users actually accomplish what they came to do? Completion rates, error rates, time on task. Often measured through usability testing or by instrumenting specific flows.

​
The Goals-Signals-Metrics structure

HEART isn’t just a list of dimensions - it comes with a method. For each dimension you care about, you define:
  • Goal - what are you trying to achieve for users in this area?
  • Signal - what user behaviour would indicate you’re achieving or failing at that goal?
  • Metric - how do you measure that signal at scale?
This structure stops teams from picking metrics that are easy to measure rather than meaningful. The goal comes first; the metric follows 🙌

​
When to use it

HEART is particularly well-suited for large products where a single metric can’t capture the full user experience - Google Maps, a complex SaaS dashboard, a platform with multiple user types. For early-stage products, pirate metrics or a focused OMTM is usually simpler and more actionable. Lesson learned: most teams using HEART skip the Goals-Signals-Metrics structure and just pick five metrics they already track. That defeats the purpose. The value is in the rigour of working backwards from what you’re trying to achieve.