Cohort analysis for subscription apps: revenue, LTV and retention by cohort

The cohorting_date selector picks how customers are grouped. new_customers uses each customer's cohort date, the earlier of when they were first seen and their first transaction. initial_conversions uses their first trial start, purchase or one-time purchase. new_paying_customers uses their first payment: a paid purchase, a renewal or a one-time purchase. A customer who never reaches that event is not in that cohort. Rows are cohorts at the chosen resolution, month by default, and the first column is the cohort size.

Columns are months of each customer's own age, counted from that customer's own start, not calendar months. Month 0 is their first month. The cohort_measure selector picks the cell: revenue, revenue_net_of_taxes, proceeds, realized_ltv (cumulative revenue), realized_ltv_per_customer (cumulative revenue divided by cohort size, the default), retained_subscriptions (subscriptions with paid access at the end of that month) or subscriptions_set_to_renew (those among them set to renew).

Money means purchases, renewals and one-time purchases minus refunds on the refund date, in US dollars at the purchase-date rate. Free trials carry no money and ad revenue is not included. proceeds subtracts the store commission, and revenue_net_of_taxes equals gross because stores report no tax. Sandbox purchases, granted access and Family Sharing purchases are excluded. Months a cohort has not reached are blank, and cells that are still filling are marked incomplete.

You can filter by app, store, product, product duration, offering, country, platform and app version, but not segment. Find it in the dashboard under Analytics > Charts, or call GET /v2/projects/{project_id}/charts/cohort_explorer with selectors such as {"cohorting_date":"new_paying_customers","cohort_measure":"revenue"}. The response has the same shape as RevenueCat's Charts API: values with cohort, period and value, and periods[0] as the cohort size.

Illustrative numbers for a January cohort of 500 new customers.

By month 2, the 500 customers have paid $2,500, which is $2,500 ÷ 500 = $5.00 each.

Cohorts tell you whether the customers you get today are worth more or less than the customers you got last quarter. A change to your paywall, price or onboarding shows up as a new row that sits above or below the rows before it, long before an average does.

Use it to judge payback on ad spend. If a cohort of customers from one campaign has paid back its cost by month 3, you can keep spending. If it flattens below cost, stop. Pair the realized columns with Prediction Explorer when you need to estimate months that have not happened yet.

How to read it

• Read across a row for one cohort's life, and down a column to compare cohorts at the same age.
• The newest cells are incomplete. Do not read them as a drop.
• Cohort on new paying customers to leave out people who never pay. A new customers cohort includes free users, so its LTV per customer is lower.
• A product, store or offering filter narrows the money and subscriptions but not the new-customer count, so per-customer values fall.
• Realized LTV is money already received. It is not a forecast.