UserValueMetrics [rule] v1.0.0

UX experiment success metrics must reflect actual user value and behavior — task completion, error rate, retention, or outcome measures — not what is easiest to instrument (clicks, page views, session duration).

Severity

warning

Applies When

selecting success metrics for any UX experiment or product change evaluation

Verify By

For each metric, ask: 'Does a positive movement here mean a user actually got value?' Reject metrics that could improve while user outcomes worsen.

UserValueMetrics [rule] v1.0.0

UX experiment success metrics must reflect actual user value and behavior — task completion, error rate, retention, or outcome measures — not what is easiest to instrument (clicks, page views, session duration).

Severity

warning

Applies When

selecting success metrics for any UX experiment or product change evaluation

Verify By

For each metric, ask: 'Does a positive movement here mean a user actually got value?' Reject metrics that could improve while user outcomes worsen.

Examples

• Good: task completion rate, error rate on key flows, 30-day retention, activation metric (first meaningful action).
• Good: time-to-complete for a core task (shorter = more efficient, but validate it's not just skipping steps).
• Bad: 'clicks on this button increased 30%' — users may be clicking because it's confusing, not because it's working.
• Bad: 'page views increased' after adding a modal — the modal creates extra views without delivering value.
• Bad: 'session duration increased' — could mean users are lost, not engaged.

Rationale

Convenience metrics (clicks, page views) can improve while user success decreases, producing false-positive experiment results. Teams that optimize for convenient metrics systematically mislead themselves.

Severity

warning

Applies When

selecting success metrics for any UX experiment or product change evaluation

Verify By

For each metric, ask: 'Does a positive movement here mean a user actually got value?' Reject metrics that could improve while user outcomes worsen.