Cohort analysis

Cohort analysis is a one-off analytical project (can be made regular): based on your data (analytics, product events, billing, CRM) we build cohorts and retention/repeat curves, compare cohort behavior over time, highlight where and at which step churn happens, which sources/segments give more 'durable' customers. We deliver a report with charts and observational conclusions. Honestly about the nature of the analysis, this is key: cohorts show patterns and correlations ('this cohort drops off in month 2'), but NOT a proven cause — why exactly needs separate checking (interviews, experiments); correlation ≠ causation. Honestly about data: sufficient volume and correctly collected events/history are needed — on small numbers cohorts are statistically noisy and conclusions premature; if data is incomplete or dirty, the analysis will be distorted (garbage in, garbage out). Honestly about the essence: this is analysis and observations, NOT raising retention and NOT implementing changes — we show what happens with cohorts and where to dig, but retention work (product, onboarding, pricing) is done by you or your specialist; the analysis itself does not retain customers. Honestly about requirements: access to data and agreed definitions (what counts as activity, churn, a cohort) are needed. An important boundary: this is one-off cohort analysis, not continuous metric monitoring (separate), not predictive models (churn/LTV — separate services) and not retention work. If there are very few customers/data, the analysis is premature — we say so honestly. Picture this: instead of 'it seems new customers leave fast, but unclear when and who' you see by cohort exactly where you lose and whom to focus on. The base price starts from 15,000 ₽ per analysis; it depends on data volume and the number of breakdowns.

• You do not understand which customer cohorts stay and which drop off.
• You see overall churn but not at which step and in which group.
• It is unclear which sources/segments give more durable customers.
• You make retention decisions without cohort data.

• Collecting and preparing data (analytics, events, billing, CRM)
• Agreeing on definitions (activity, churn, cohort)
• Building cohorts and retention/repeat curves
• Comparing cohort behavior over time and by segments
• Highlighting churn points and more 'durable' sources
• A report with charts and observational conclusions
• Reviewing results with you
• Recommendations on where to dig (implementation — separately)

• You understand which cohorts stay and which leave and when
• Churn points and more valuable sources/segments are visible
• Data for retention decisions (implementation — separately)
• A base for hypotheses and experiments

• We clarify questions, data, cohort definitions; collect access
• We prepare data, build cohorts and curves, look for patterns
• We compile the report, review conclusions and directions with you

• Will cohort analysis raise retention?

  Not directly. It shows which cohorts drop off and when and where to dig, but retention work (product, onboarding, pricing, win-back) is done by you or your specialist. The analysis is a base for decisions, not retention itself.

• Will you say exactly why a cohort leaves?

  We show patterns and correlations (where and when you lose), but not a proven cause — why exactly needs separate checking (interviews, experiments). Correlation is not causation.

• Is there enough data for analysis?

  Sufficient volume and correctly collected events/history are needed. On small numbers cohorts are statistically noisy and conclusions premature; with dirty data the analysis will be distorted. If data is scarce, we honestly say it is too early.