Cohort retention curves

Cohort retention curves is a one-off analytical project (can be made regular): based on your data (product events, billing, analytics) we build retention curves by arrival cohorts and the needed breakdowns, show the shape of retention (fast drop, plateau, a 'smile' of returns), compare cohorts and sources, and highlight critical churn periods. We deliver a report with charts and observations. Honestly about the nature, this is key: the curves show PATTERNS and correlations ('the cohort from channel X retains worse'), but NOT a proven cause — why needs separate checking (interviews, experiments); correlation ≠ causation. Honestly about data: sufficient volume and correctly collected events are needed — on small cohorts the curves are statistically noisy, and the definition of 'activity/return' affects the shape, so we fix it in advance; dirty data will distort the result (garbage in, garbage out). Honestly about the essence: this is analysis and visualization of retention, NOT raising retention and NOT implementing changes — we show how and where you lose, but retention work (product, onboarding, communications) is done by you or your specialist; a curve by itself retains no one. Requirements: access to data and an agreed definition of activity/return. An important boundary: this is building retention curves, not a full cohort analysis of all metrics (683/675 are broader), not predictive churn (separate) and not retention work. If there are few users/data, the curves are premature. Picture this: instead of 'retention seems so-so but unclear when it breaks' you see the curve shape and the critical period. The base price starts from 12,000 ₽ per analysis; it depends on data volume and the number of breakdowns.

• You do not see how retention changes over time after arrival.
• You do not know in which period the main drop happens.
• You do not compare retention of cohorts and sources.
• You make retention decisions without curves.

• Preparing data (events, billing, analytics)
• Agreeing on the definition of activity/return
• Building retention curves by cohorts
• Breakdowns by sources, plans, segments
• Highlighting critical churn periods
• A report with charts and observations
• Comparing cohorts with each other
• Reviewing results with you

• The shape of retention and where it breaks/stabilizes is visible
• You understand which cohorts and sources retain better
• Critical churn periods are in plain sight
• Data for retention decisions (implementation — separately)

• We clarify the activity definition, cohorts, breakdowns; collect access
• We prepare data, build curves, compare cohorts
• We compile a report with observations, review with you

• Will retention curves raise retention?

  Not directly. They show the shape and critical churn periods, but retention work (product, onboarding, communications) is done by you or your specialist. A curve is a base for decisions, not retention itself.

• Will you say why a cohort breaks?

  We show the pattern and correlations (where and when it breaks), but not a proven cause — why exactly needs checking (interviews, experiments). Correlation is not causation.

• How is this different from cohort analysis?

  Retention curves focus specifically on retention over time. A full cohort analysis is broader (behavior, returns, revenue by cohorts). If you need a broad breakdown, that is a separate, larger service.