How to measure user experience

Without measurement, design decisions rest on opinions. One stakeholder says "this form is confusing", another says "it looks fine to me". Measuring UX introduces evidence: the form abandonment rate is 68 %, average completion time is double the industry benchmark, and three out of five users click the wrong field first.

Beyond settling internal debates, UX metrics let you establish baselines, set concrete targets ("reduce checkout time by 30 %") and track progress sprint by sprint. Companies like Google, Spotify and Airbnb embed UX metrics in their quarterly OKRs, treating them with the same rigour as financial metrics.

Quantitative metrics answer "how much" and "how often". They are collected at scale, support statistical comparisons and are ideal for measuring the impact of changes between versions.

The System Usability Scale (SUS) is a 10-question survey that yields a score from 0 to 100. A score above 68 is considered acceptable; above 80, good. The Net Promoter Score (NPS) measures willingness to recommend the product on a 0-to-10 scale and classifies users as detractors, passives or promoters.

Beyond surveys, behavioural metrics capture what users actually do. Task success rate measures the percentage of users who complete a flow without assistance. Time on task quantifies how long they take. Error rate records how many mistakes are made per task. And abandonment rate reveals the exact point at which users stop trying.

Quantitative metrics tell you what is happening; qualitative methods tell you why. A think-aloud usability test asks real users to complete tasks while verbalising their thoughts. With five participants you can identify roughly 85 % of usability problems, according to Nielsen’s model.

In-depth interviews explore motivations, frustrations and mental models. Open-ended post-task surveys ("What was the hardest part of this process?") capture immediate perceptions. And diary studies record the experience over days or weeks, revealing patterns that a one-off test cannot detect.

The UX measurement tooling landscape has matured significantly. For behavioural analytics, Hotjar and Microsoft Clarity offer heatmaps, session recordings and conversion funnels. For remote usability testing, Maze and UserTesting let you recruit participants, set tasks and collect success and timing metrics automatically.

For standardised surveys like SUS or NPS, tools such as Typeform or SurveyMonkey handle distribution, collection and analysis at scale. Mixpanel and Amplitude cover product analytics with cohorts, funnels and retention analysis. And for qualitative research synthesis, Dovetail and Condens help organise, tag and synthesise findings from interviews and tests.

The HEART framework, developed by Google’s research team, organises UX metrics into five dimensions: Happiness (subjective satisfaction), Engagement (depth of use), Adoption (new users), Retention (returning users) and Task success (completion of key flows). For each dimension you define goals, observable signals and concrete metrics.

The UX Scorecard is a lighter variant that assigns scores to each major product flow across dimensions like efficiency, satisfaction and error rate. It is updated after each research round and lets you visualise how the experience evolves over time. Both frameworks share one principle: do not measure everything — measure what matters for your business objectives.

Measuring UX once is useful; measuring it continuously is transformative. A sustainable programme combines quantitative metrics collected automatically (analytics, periodic NPS) with scheduled qualitative research (one usability test per sprint or per month).

The first step is to define which flows are critical to the business and which metrics represent their health. The second is to automate data collection wherever possible. The third is to set a review cadence: a monthly dashboard the product team reviews alongside business metrics. And the fourth is to close the loop — every finding should produce an action, whether a design change, an A/B experiment or a deeper investigation.