Measure across happiness, engagement, adoption, retention, and task success

What it is

Experience is not one additive score. At minimum, five questions need separate answers: happiness (is affect positive), engagement (how deeply and often people use it), adoption (whether a new capability is actually taken up), retention (whether people return), and task success (whether intended work gets done). These dimensions name different failures. Completing the task while disliking the product, using it intensely then never returning, or shipping a feature nobody takes up all disappear inside a single index. The taxonomy exists so those failures stay visible.

Why it happens

Each dimension can degrade through a distinct mechanism. Task success usually fails from broken paths, missing feedback, or capability thresholds. Happiness fails from emotional load, unfairness, or tone. Engagement moves with supply, habit loops, and interruption cost. Adoption fails at discovery, comprehension, and first value. Retention fails from substitutes, unformed habits, or after-the-fact regret. Time scales differ too: task outcome is visible inside a session; adoption needs a first-use window; retention needs a return cycle. Collapsing them into one index assumes they move in the same direction at the same speed. Product decisions routinely trade across them: more onboarding can raise adoption and lengthen the task; fewer steps can raise success and cut time-on-task. Multiple dimensions exist so one mechanism’s gain cannot hide another’s loss.

Studying it

Write the product questions as a dimension list first, then attach at least one attitudinal or behavioral indicator to each, on the same users and window so they can sit side by side. Inspect cross-dimension correlation: near-collinearity means the extra axes add no information; low correlation plus distinct failure stories means the split is doing work. Intervention studies should pre-declare which dimensions matter and which may stay flat or fall, so teams cannot pick the column that rose. Qualitative material explains splits: high success with low happiness should point to a named affective source, not be averaged away.

Where it stops holding

The five axes cover common product questions; they are not natural kinds, and they do not require every product to collect all five. In internal tools, “adoption” may mean an entitlement being switched on, which is not the same construct as consumer feature uptake. In strongly task-oriented systems, happiness can weigh less than success and error cost. Averaging the five into a composite recreates the problem the taxonomy was meant to split. Attitudinal measures carry response burden and social desirability; behavioral measures inherit event-definition choices. Sitting on one dashboard does not make their units comparable.

Applying it

• Open the measurement plan with one sentence per question, not with a list of available fields.
• Hang at least one operable metric on every dimension still being asked, and label it as attitude or behavior.
• Review releases on splits: “success up, happiness down” must be allowed as an explicit agenda item; a composite score must not be the only report.
• If two dimensions move together in size and direction for a long stretch, merge or retire one.

Related

• Same group: Q6.01.2 A single dimension drives one-sided optimization · Q6.01.3 Not every product needs every dimension
• Adjacent: Q6.07 Experience measurement frameworks · Q6.02 Goals–Signals–Metrics
• Search terms: multidimensional UX measurement · happiness engagement adoption retention task success · metric taxonomy