การวิเคราะห์กลุ่มผู้ใช้และเส้นโค้งการรักษาผู้ใช้

Total downloads feel good but hide the truth. A growing total can mask the fact that everyone leaves after a week. Cohort analysis reveals what is really happening.

This lesson teaches you to read retention honestly.

A cohort is a group of users who share a starting event — usually the week or month they signed up. You then track how each cohort behaves over time.

This separates new-user noise from real retention.

Retention is the percentage of a cohort still active after N days. Day-1, Day-7, and Day-30 retention are the classic checkpoints.

Plot retention over time and you get a curve. It drops fast at first, then ideally flattens. A curve that flattens above zero means you have a core of users who stay.

A curve that hits zero means no real retention.

The best products show a curve that flattens and sometimes ticks back up as dormant users return — the smile curve. This signals product-market fit.

Aim to flatten the curve before scaling spend.

A cohort table has cohorts as rows and periods as columns. Each cell is the retention for that cohort at that age.

The real power: compare cohorts over time. If newer cohorts retain better than older ones, your changes are working. If worse, something regressed.

Cohorts turn product changes into measurable cause and effect.

Slice cohorts further by:

• Acquisition channel
• Platform
• Whether they hit the aha moment

This reveals which users to acquire more of and which onboarding paths to fix.

Retention depends on what counts as active. For a daily app it might be opening the app; for a weekly tool, completing a key action.

Choose a definition that reflects real value, not just app opens.

Cohort insights drive iteration:

• Low D1: onboarding problem
• Steep D1 to D7 drop: weak habit formation
• Curve never flattens: missing core value

Diagnose, change, then watch the next cohort.

Put it together:

• Group users into signup cohorts
• Define a meaningful active event
• Track D1, D7, D30 retention
• Look for a flattening curve
• Compare and segment cohorts to guide changes

Retention curves tell you the truth about your product.

Test your cohort analysis knowledge.

You learned cohort-based retention analysis:

• Cohorts group users by signup period to reveal true retention
• Retention curves should flatten above zero
• The smile curve signals product-market fit
• Compare and segment cohorts to measure changes
• Define active by real value, then act on the diagnosis

Cohorts cut through vanity metrics to the truth.