Engagement Metrics

• Define an active user with explicit event criteria
• Compute DAU/MAU and stickiness and interpret them
• Use engagement buckets to segment users by habit strength

• 'Active' means nothing until you say which events count — an email-open is not a product session.
• Stickiness (DAU/MAU) is the single most powerful growth diagnostic because it equalizes for size.
• Engagement segments predict retention and revenue better than demographic segments.
• Session metrics reveal whether users are dropping in for seconds (grab-and-go) or investing minutes.

A gym tracks who actually shows up, how often and for how long, versus who merely has a card. DAU/MAU is 'shows up this month vs this day'; stickiness is 'how many members come on an average day'.

• Define active: e.g. active = unique user_id with ≥1 'session_start' event that day.
• DAU = unique actives per day; MAU = unique actives per 28-day rolling window (or calendar month — pick one and stay consistent).
• Stickiness = DAU / MAU, typically 10–25% strong, debated for every product.
• Depth: sessions per active user, actions per session, median session length.
• Breadth: count of distinct events/features a user touched, bucketed (power user curve).

• A news app reports DAU but flags that 60% of 'active users' is push-opened sessions — engagement depth contradicting the headline number.
• A collaboration tool segments users by stickiness bucket (≤1, 2–4, 5+ days/week) and finds the 5+ bucket has 8x retention at 90 days.
• A gaming studio tracks session length distributions and ships a 'quick session' mode after spotting 40% of sessions under 3 minutes.

• Document the active-user definition in the metric dictionary and version it.
• Use a rolling 28-day MAU so comparisons aren't distorted by month length, and state the window.
• Report DAU/MAU per cohort, not just as a company total, to separate product shifts from acquisition mix.
• Complement ratios with absolute counts — a 20% stickiness on 1M users is different from a 20% on 10M.

• Counting passive events (app opened in background) as active.
• Mixing calendar-month MAU with daily DAU and getting a stickiness number that drifts with month length.
• Celebrating DAU growth caused by a viral campaign while cohort stickiness is falling.
• Defining 'feature used' as merely seeing a screen — breadth should require an interaction.

• When stickiness jumps, split DAU by acquisition channel — a campaign spike inflates DAU without real habit.
• Reconcile the active definition by re-running it on a 1-day window in a sandbox with a known user.
• If MAU is suspiciously flat while DAU moves, check your dedupe key (user_id vs device_id).
• Compare stickiness by platform; a mobile+web product blends two different engagement norms.