HEART Framework: How to Measure User Experience

Google's HEART framework measures user experience across five dimensions — Happiness, Engagement, Adoption, Retention, and Task success — and pairs them with a Goals-Signals-Metrics process that turns each dimension into a number you can track. You choose only the categories that fit your product; you rarely need all five.

Quick Answer: HEART is Google's model for measuring UX in five categories — Happiness, Engagement, Adoption, Retention, and Task success. Use the Goals-Signals-Metrics method to translate each category you pick into a trackable metric, normalized as a rate or ratio. Skip any category that doesn't map to a real goal.

Created by Google researchers Kerry Rodden, Hilary Hutchinson, and Xin Fu, HEART was designed to bring rigor to a slippery question: is the experience actually good, and is it getting better? This guide walks each dimension the way a UX researcher applies it, then shows the Goals-Signals-Metrics process that stops you from tracking numbers you can't act on.

Why User Experience Needs Its Own Metrics Beyond Revenue

User experience needs its own metrics because revenue and signups are lagging, blended numbers that tell you the business moved without telling you why. HEART gives the experience its own scoreboard — one that can improve or degrade weeks before the money reflects it.

Think about a redesign. Sign-ups might hold steady for a month while satisfaction quietly climbs and support tickets fall. Revenue is the last thing to move, so if it's your only gauge, you're steering by a mirror that shows where you've already been.

For a PM-turned-founder this is familiar territory, but the trap is subtler at startup scale. You have fewer users, noisier data, and more pressure to point at one flattering chart. Experience metrics are how you catch a problem while it's still cheap to fix.

They also separate two very different kinds of stall. Revenue that flatlines could mean the market is saturated, the pricing is wrong, or the product simply isn't good enough to keep — and each demands a different response. HEART's dimensions let you tell "people don't like it" apart from "people like it but won't pay," which a revenue chart alone never can.

HEART complements your business metrics rather than replacing them. The AARRR pirate-metrics framework tracks how users move through your funnel, and a single north star metric keeps the whole team pointed at one outcome — but neither tells you whether the experience itself is any good. That is the gap HEART fills.

One more distinction matters: HEART works at any altitude. You can apply it to a whole product, a single feature, or a redesign of one flow. That flexibility is why teams reach for it when a launch needs its own success definition, not just a share of the company dashboard.

The Five HEART Dimensions at a Glance

The five HEART dimensions are Happiness, Engagement, Adoption, Retention, and Task success — one attitudinal category and four behavioral ones. Together they span how users feel and what they actually do, which is why the model reads as a spectrum rather than a checklist.

The table below maps each dimension to the question it answers and where its data usually comes from. Read the middle column first — it's the fastest way to spot which categories your team actually needs.

The takeaway is permission to be selective. If a dimension's question isn't one your team urgently needs answered this quarter, that's your cue to leave it out — the sections below treat each category as optional, not mandatory.

Happiness: Attitudinal Signals from Surveys

Happiness measures how users feel about your product — their satisfaction, perceived ease of use, and willingness to recommend it. It is the one attitudinal dimension in HEART, which means it usually comes from asking people, not from watching what they do.

Because it's self-reported, Happiness fills a gap the behavioral categories can't. A user can complete every task and still find the product frustrating, confusing, or joyless. Only a direct question surfaces that feeling.

Common Happiness signals include:

• Satisfaction ratings collected after a key interaction or on a recurring basis.
• Perceived ease of use, often a short "how easy was that?" prompt tied to a specific task.
• Likelihood to recommend, the attitudinal question behind net-promoter-style surveys.

Track the trend, not the snapshot. A single satisfaction score in isolation means little; the same score falling over three releases is a signal worth acting on. Segment it too — power users and first-timers often feel very differently about the same change.

Where and when you ask shapes what you learn. A short in-the-moment prompt fired right after a task captures how that flow felt while it's fresh, while a periodic relationship survey captures how someone feels about the product overall. The two answer different questions, and blending them into a single score hides more than it reveals.

The honest caveat with Happiness is bias. People misremember, answer to be polite, and respond in non-representative samples. Treat it as one input triangulated against behavior, not as a verdict on its own.

Engagement: Depth and Frequency of Use

Engagement measures how involved your active users are — the frequency, depth, and intensity of their interaction over a period of time. It answers "when people do use this, how much are they using it?" rather than "how many people are there?"

The critical technique is to measure Engagement per user, as an average or rate, not as a raw total. A rising total number of actions might just mean your user base grew; a rising average of actions per user means the experience itself is pulling people in deeper.

Typical Engagement signals depend entirely on what your product is for:

• Frequency — sessions or visits per active user per week.
• Depth — actions taken per session, or number of features touched.
• Intensity — content created, messages sent, or documents edited per user.

The nuance worth internalizing: Engagement is most meaningful when usage is voluntary. For a consumer app people choose to open, more engagement is a good sign. For internal or mandated enterprise software, high "engagement" can just mean the workflow is inefficient — so Adoption and Retention may tell you more. Picking the right engagement metrics for depth and frequency starts with being honest about whether your users have a choice.

Don't confuse Engagement with Retention, either. Engagement is about intensity within a period; Retention is about returning across periods. A user can be intensely engaged this week and gone the next.

Adoption: New Users Reaching First Value

Adoption counts new users who start using a product or feature within a given time period — and, done well, it measures how many of them actually reach value, not just how many create an account. It is the dimension you reach for at a launch.

Adoption shines for new products, new features, and redesigns, because those are exactly the moments when "are people picking this up?" is the urgent question. A dormant feature nobody discovers has an Adoption problem long before it has a Retention one.

Useful Adoption signals include:

• New accounts or activations created in the last 7, 14, or 30 days.
• Feature uptake, such as the share of eligible users who tried a newly shipped capability.
• Setup completion, the percentage of new users who finish onboarding and reach a first meaningful action.

Anchor Adoption to value, not to registration. Counting signups flatters you; counting new users who reach the "aha" moment tells you whether the product actually landed. The gap between those two numbers is often where onboarding goes to die.

Because both Adoption and Retention rely on counts of unique users over time, they share a hazard: a metric can rise simply because your overall user base is growing. That is why normalization — expressing the number as a rate or share rather than a raw count — matters so much, a point the Goals-Signals-Metrics section returns to.

Retention: Users Who Stay Active

Retention measures the rate at which existing users remain active over time — the direct antidote to a leaky bucket. Where Adoption watches newcomers arrive, Retention watches whether they stick around instead of quietly disappearing.

It's usually expressed as the share of users active in one period who are still active in a later one, or inversely as churn. Cohort analysis is the natural home for it: group users by when they started, then watch how each cohort's activity decays or flattens.

Retention is often the HEART dimension that sits closest to product-market fit. A flattening retention curve — users who keep coming back without being prodded — is one of the clearest experience-level signals that you've built something people genuinely need, which is a large part of what product-market fit actually means in practice.

Two failure modes are worth naming:

• Confusing Retention with Adoption. A surge of new signups can mask a base of veterans steadily churning; only cohorts separate the two.
• Averaging across cohorts. A blended retention number can look stable while every recent cohort gets worse, because older, loyal users prop up the mean.

The shape of the curve carries the message. A curve that keeps sliding toward zero says users try the product and drift away; a curve that flattens onto a plateau says a stable core has turned it into a habit. In HEART terms, that plateau is the retention signal you most want to see rise across successive cohorts.

The comeback case matters too. Users who lapse and later return are a distinct, valuable segment, and measuring the retention metric that captures user comebacks keeps you from writing off accounts that were only dormant, not dead.

Task Success: Efficiency, Effectiveness, and Error Rate

Task success covers the classic behavioral UX metrics — efficiency, effectiveness, and error rate — for the parts of your product where users are trying to get something specific done. It is the most traditional dimension, drawn straight from usability research.

Its three components are distinct:

• Efficiency — how long a task takes, or how many steps it requires.
• Effectiveness — the percentage of users who complete the task successfully.
• Error rate — how often users make mistakes or hit dead ends along the way.

Task success fits utilitarian flows, not open-ended ones. Checkout, search, onboarding, and configuration are task-focused: there's a clear goal and a clear success state, so completion and error rate are meaningful. For open-ended browsing or discovery, "success" is fuzzier, and forcing a task metric onto it can mislead.

A checkout flow makes the dimension concrete. Effectiveness is the share of users who reach the confirmation screen, efficiency is how long that takes, and error rate is how often a mismatched field or a declined payment bounces them back a step. Because the success state is unambiguous, you can measure it in a moderated test before launch and then confirm it against live logs afterward.

You can source Task success two ways. Behavioral logs give you completion and time-on-task at scale, while moderated usability testing explains the why behind a high error rate that logs alone can't diagnose. The two together are far stronger than either alone.

The Goals-Signals-Metrics Method to Pick What to Track

The Goals-Signals-Metrics method is the process that turns a HEART dimension into an actual number. HEART tells you which categories of experience exist; Goals-Signals-Metrics translates a category you've chosen into something you can put on a dashboard. It runs in three ordered steps.

Step 1 — Goals. For each dimension you're using, articulate what success looks like in plain language. What are users trying to accomplish, and what does a good experience let them do? Rodden and her co-authors are deliberate about doing this first, because teams find it far easier to agree on goals than to jump straight to metrics — and stakeholders who skip the alignment step often discover they were optimizing different things.

Step 2 — Signals. Ask how each goal would reveal itself in user behavior or attitudes. What action would a user take, or what feeling would they report, if the goal were being met? Name failure signals too. Good signals are sensitive to the goal and specific to it, and it helps to note where the data would come from — logs for behavior, surveys for attitude.

Step 3 — Metrics. Refine each signal into a specific metric suitable for tracking over time. This is where normalization earns its keep: express metrics as rates, ratios, or per-user averages so they aren't secretly just tracking overall growth. A raw count of "active users" rises with the company; a percentage tells you whether the experience is improving.

The grid below shows the three steps applied to one hypothetical feature, so you can see how a vague ambition tightens into a trackable number.

The lesson the table encodes is that the metric is a consequence of the goal, not a starting point. Teams that begin with "what can we easily measure?" end up with tidy dashboards that answer no real question.

The signal step is where most grids go wrong. A signal has to be both sensitive — it moves when the experience changes — and specific — it moves for that reason and not a dozen others. "Total logins" fails both tests: it drifts with marketing campaigns, seasonality, and pricing changes, so it can't cleanly tell you whether the experience improved. A signal you can't attribute is a signal you can't act on.

Two rules keep the whole method honest. First, you don't need all five HEART categories — pick only the dimensions with a goal behind them, and a focused two or three beats a hollow five. Second, treat the metric as the last step, never the first. For a fuller walkthrough of building the grid, see the deep dive on the HEART framework's goals-signals-metrics process.

This is also where experience measurement loops back into validation. The goals you set in this grid are, in effect, hypotheses about the experience you promised users — and instrumenting them is how a platform like Edmired helps you confirm that a shipped product actually delivers what your earlier discovery work said it would.

Key Takeaways

• HEART measures UX in five dimensions. Happiness, Engagement, Adoption, Retention, and Task success — one attitudinal category plus four behavioral ones — created by Google researchers Kerry Rodden, Hilary Hutchinson, and Xin Fu.
• You rarely need all five. Choose only the dimensions that map to a real goal; a focused two or three categories beat a hollow full set that nobody acts on.
• Goals-Signals-Metrics turns a dimension into a number. State the goal in plain language first, identify the behavior or attitude that signals it, and only then define the metric — in that order.
• Normalize behavioral metrics. Express Engagement, Adoption, and Retention as rates, ratios, or per-user averages so they track the experience rather than your overall growth.
• Engagement is per user, and context-dependent. Measure intensity per active user, and remember it's most meaningful when usage is voluntary rather than mandated.
• Retention is your closest read on need. A flattening retention curve is one of the strongest experience-level signals that you've built something people genuinely come back for.
• HEART complements, not replaces, business metrics. It sits alongside funnel and north-star metrics, giving the quality of the experience its own scoreboard the money can't see yet.

Frequently Asked Questions

What does HEART stand for in the HEART framework?

HEART stands for Happiness, Engagement, Adoption, Retention, and Task success. Google researchers Kerry Rodden, Hilary Hutchinson, and Xin Fu introduced it as a way to measure user experience at scale, spanning one attitudinal dimension (Happiness) and four behavioral ones. Each category is paired with the Goals-Signals-Metrics process to become measurable.

Do I need to use all five HEART metrics?

No. HEART is a menu, not a mandate. You choose only the dimensions that map to an actual goal for your product or feature, and most teams use two or three at a time. Forcing all five produces metrics nobody acts on; a smaller, goal-backed set is far more useful than a complete but hollow dashboard.

What is the difference between the HEART framework and AARRR?

HEART measures the quality of the user experience across five dimensions, while AARRR tracks how users move through your business funnel — acquisition, activation, retention, revenue, and referral. HEART tells you whether the experience is good; AARRR tells you where users convert or drop off. They complement each other rather than competing, and many teams run both.

How does the Goals-Signals-Metrics process relate to HEART?

HEART names the categories of user experience worth measuring; Goals-Signals-Metrics is the method for turning any chosen category into a trackable number. You state the goal for a dimension, identify the behavior or attitude that signals the goal is being met, then refine that signal into a specific, normalized metric — in that order, never metric-first.

Is engagement or retention more important in the HEART framework?

Neither ranks above the other; they answer different questions. Engagement measures how intensely active users interact within a period, while Retention measures whether users return across periods. A product can have strong engagement and weak retention, or the reverse. Which matters more depends on your goal — pick the dimension that reflects the experience you're trying to improve.