Cohort Retention Analysis
See how well you keep customers. Each customer joins the cohort of their first purchase; the tool shows what share of every cohort comes back in each later period, as a heatmap and an average retention curve, with revenue per customer if you include amounts.
Updated · By Bibin XavierData 3775 × 4
| Customer_IDtext · 520 levels | Order_Datetext · 711 levels | Amountnumeric | Channeltext · 3 levels |
|---|---|---|---|
| C0109 | 2023-01-02 | 46.6 | App |
| C0461 | 2023-01-02 | 135.38 | App |
| C0512 | 2023-01-02 | 101.19 | App |
| C0461 | 2023-01-03 | 67.9 | App |
| C0511 | 2023-01-06 | 65.28 | Web |
| C0109 | 2023-01-07 | 59.7 | App |
| C0195 | 2023-01-07 | 73.29 | App |
| C0457 | 2023-01-07 | 19.37 | App |
Variables and options
Results
3775 rows · example dataRetention by cohort (% of customers active)
Cohort table
| Cohort | Customers | Month 0 | Month 1 | Month 2 | Month 3 | Month 4 | Month 5 | Month 6 | Month 7 | Month 8 | Month 9 | Month 10 | Month 11 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2023-01 | 31 | 100% | 48% | 42% | 48% | 45% | 35% | 23% | 10% | 16% | 16% | 16% | 13% |
| 2023-02 | 30 | 100% | 37% | 30% | 37% | 37% | 30% | 37% | 33% | 13% | 23% | 10% | 30% |
| 2023-03 | 26 | 100% | 46% | 38% | 35% | 38% | 35% | 27% | 23% | 23% | 12% | 4% | 35% |
| 2023-04 | 31 | 100% | 48% | 42% | 26% | 26% | 29% | 13% | 13% | 13% | 16% | 16% | 19% |
| 2023-05 | 25 | 100% | 40% | 32% | 40% | 8% | 24% | 20% | 40% | 28% | 16% | 16% | 24% |
| 2023-06 | 17 | 100% | 35% | 41% | 35% | 35% | 35% | 12% | 18% | 18% | 24% | 24% | 12% |
| 2023-07 | 35 | 100% | 54% | 40% | 43% | 37% | 34% | 29% | 31% | 20% | 17% | 29% | 23% |
| 2023-08 | 29 | 100% | 45% | 59% | 28% | 48% | 34% | 24% | 41% | 24% | 31% | 34% | 45% |
| 2023-09 | 26 | 100% | 42% | 50% | 42% | 31% | 27% | 27% | 31% | 19% | 23% | 27% | 23% |
| 2023-10 | 31 | 100% | 42% | 45% | 32% | 35% | 19% | 26% | 26% | 16% | 19% | 10% | 13% |
| 2023-11 | 23 | 100% | 61% | 52% | 22% | 43% | 35% | 22% | 13% | 26% | 22% | 22% | 13% |
| 2023-12 | 24 | 100% | 50% | 29% | 38% | 17% | 38% | 29% | 33% | 13% | 29% | 33% | 38% |
| 2024-01 | 29 | 100% | 34% | 48% | 34% | 34% | 34% | 34% | 34% | 31% | 21% | 24% | 17% |
| 2024-02 | 32 | 100% | 47% | 47% | 34% | 28% | 31% | 31% | 25% | 22% | 22% | 13% | |
| 2024-03 | 29 | 100% | 45% | 38% | 24% | 28% | 34% | 21% | 24% | 21% | 24% | ||
| 2024-04 | 32 | 100% | 44% | 41% | 31% | 34% | 22% | 16% | 25% | 16% | |||
| 2024-05 | 43 | 100% | 42% | 37% | 58% | 26% | 23% | 21% | 26% | ||||
| 2024-06 | 27 | 100% | 44% | 30% | 22% | 22% | 33% | 30% | |||||
| Weighted average | 520 | 100% | 45% | 41% | 36% | 32% | 30% | 25% | 26% | 20% | 21% | 20% | 24% |
Note. Month 0 is the period of the first purchase. Retention = customers with at least one order in the period ÷ cohort size. Empty cells lie after the end of the data. The average weights each cohort by its size.
Average retention curve
Interpretation
On average 44.8% of new customers buy again in the month after their first purchase, and 24.6% are still active 6 months later. The 2023-11 cohort retained best in its first month (60.9%) and the 2024-01 cohort worst (34.5%); look at what was different about acquisition or onboarding in those periods. A curve that flattens out shows a core of loyal customers; one that keeps falling points to a retention problem.
How to cite this tool
Using these results in a thesis, journal article, report or class assignment? Please cite this free tool so others can find it. It keeps the site free.
APA 7
Xavier, B. (2026). Cohort Retention Analysis [Web application]. FreeStatisticalTools.com. https://freestatisticaltools.com/cohort-retention-analysis/
BibTeX
@software{xavier_cohortretention_2026,
author = {Xavier, Bibin},
title = {Cohort Retention Analysis},
publisher = {FreeStatisticalTools.com},
year = {2026},
url = {https://freestatisticaltools.com/cohort-retention-analysis/},
note = {Web application}
}Suggested sentence for your methods section: “The analysis was carried out with the free Cohort Retention Analysis tool on FreeStatisticalTools.com (Xavier, 2026).”
What is cohort retention analysis?
Cohort analysis groups customers by the period of their first purchase and tracks what share of each cohort is active in later periods. The tool builds the cohort table and retention heatmap, a size-weighted average retention curve, and, with an amount column, revenue and cumulative revenue per customer by cohort.
When to use cohort retention analysis
Use cohort analysis to track retention, repeat purchase and revenue per customer over time, compare the quality of customers acquired in different months or campaigns, and see whether product or onboarding changes improved retention.
Assumptions
- A customer belongs to the cohort of their first order in the data; customers who bought before the data starts are misclassified, so start the extract at a clean date if possible.
- Recent cohorts have fewer periods of history; compare cohorts at the same age.
- Retention counts any activity in the period, not consecutive activity.
How to set up your data
One row per order or activity, with a customer ID and a date, and optionally an amount for revenue views.
What cohort analysis shows
A cohort is a group of customers who share a starting point, most often the month of their first purchase or sign-up. Cohort analysis follows each group over time and measures what share of its members are still active in each later period. The result, a cohort table or retention heatmap, is one of the most useful views of a business, because it separates two things that total sales mix together: how many new customers you acquire, and how well you keep them.
Total revenue can grow even while retention is getting worse, as long as acquisition grows faster. A cohort table makes that visible. It also shows whether a product change, an onboarding improvement or a different acquisition channel produced customers who stay longer, because you can compare cohorts at the same age.
How this tool builds the cohort table
- Each customer is assigned to the cohort of their first order in the data (by month, week or quarter).
- For every later order, the tool computes the period number: period 0 is the cohort period itself, period 1 the next month (or week, or quarter), and so on.
- For each cohort and period, it counts the customers with at least one order in that period and divides by the cohort size. Period 0 is always 100%.
- With an amount column, it can instead show the revenue per customer in each period, or the cumulative revenue per customer, which is a simple view of lifetime value as cohorts age.
- The weighted average row combines cohorts, weighting by cohort size and using only cohorts old enough to have reached each period. The retention curve plots this average together with the most recent cohorts.
Reading the heatmap
- Rows are cohorts, with their size in brackets. Read along a row to follow one cohort as it ages.
- Columns are periods since the first purchase. Read down a column to compare cohorts at the same age: if month-3 retention is higher for recent cohorts than for older ones, retention is improving.
- Diagonals represent calendar periods. A darker or lighter diagonal points to something that happened in a particular month for all cohorts, such as a seasonal peak, a promotion or a service outage.
- The lower-right triangle is empty because recent cohorts have not yet reached those ages.
The shape of the retention curve matters as much as its level. A curve that drops steeply and then flattens shows that a core of loyal customers stays; that flat level is the long-term retention rate. A curve that keeps declining towards zero means there is no stable core and every cohort eventually leaves.
Worked example with the built-in data
The example contains orders from 520 customers in 18 monthly cohorts from January 2023 to June 2024. On average 45% of new customers buy again in the month after their first purchase, 36% are active in month 3, and 25% in month 6, after which the curve flattens at around 20–25%. Month-1 retention ranges from 34% (January 2024 cohort) to 61% (November 2023 cohort, possibly customers acquired during a pre-Christmas campaign who returned in December). Switching the heatmap to cumulative revenue per customer shows how quickly each cohort pays back its acquisition cost.
Preparing the data
- One row per order or activity, with a customer ID and a date; an amount column is optional and enables the revenue views.
- Start the extract at a clean date. Customers who bought before the data begin will appear as “new” in the first months, so ignore the first one or two cohorts if your data do not start at the launch of the business.
- Define activity consistently: for an app, a session or a purchase; for a shop, an order. Exclude refunds and cancelled orders.
- Choose the period to match the purchase cycle: weekly cohorts for apps and food delivery, monthly for most e-commerce, quarterly for infrequent purchases.
Common retention metrics
- N-period retention (what this tool shows): share of the cohort active in period N. Customers can skip periods and come back.
- Rolling or unbounded retention: share active in period N or any later period, which is always higher.
- Churn rate: 1 − retention, often used for subscriptions where an inactive customer has definitively left.
- Repeat purchase rate: share of a cohort that has made at least a second purchase by a given age.
Turning insight into action
Find the period with the steepest drop: if most customers are lost between period 0 and 1, the priority is the first repeat purchase (onboarding emails, a second-order incentive, better first experience). If the decline is gradual, focus on engagement and loyalty programmes. Compare cohorts acquired through different channels or campaigns by filtering your data before pasting; channels that bring cheap customers who do not return can be less profitable than expensive channels with high retention. For individual-level predictions of who will return, use the customer lifetime value tool; for segments of current customers, use RFM analysis.
Limitations
- Small cohorts make percentages noisy; a cohort of 20 customers moves 5 points with one customer.
- Seasonality affects both acquisition and retention; compare cohorts from the same season when possible.
- Cohort tables describe; they do not explain why retention changes. Combine them with survey data or experiments.
Reporting
Report how cohorts and activity are defined, the period used, the number of cohorts and customers, and average retention at key ages, ideally with the heatmap. Example: “Across 18 monthly cohorts (520 customers), an average of 44.8% of new customers purchased again in month 1, 35.8% in month 3 and 24.6% in month 6.”
Glossary of cohort terms
- Cohort: customers grouped by the period of their first purchase or sign-up.
- Cohort age: the number of periods since the cohort’s first period; period 0 is the acquisition period.
- Retention rate: the share of a cohort active in a given period.
- Churn rate: one minus retention; in subscriptions, the share of customers who cancel in a period.
- Revenue per customer: revenue in a period divided by the original cohort size, so that churned customers count as zero.
- Cumulative revenue per customer: the running total of revenue per customer, a simple empirical lifetime value curve.
- Payback period: the cohort age at which cumulative margin per customer exceeds the acquisition cost.
- Survivorship bias: the error of looking only at customers who are still active, which overstates engagement.
Key references: Fader and Hardie (2007), “How to project customer retention”, Journal of Interactive Marketing; Gupta, Lehmann and Stuart (2004), Journal of Marketing Research.
How to use this tool
- Enter your data. Paste columns from Excel, Google Sheets or SPSS with variable names in the first row, or upload a CSV or Excel file. The example data shows the expected layout.
- Choose the variables. Select the customer id, order or activity date and amount (optional).
- Set the options. Fill in cohort period, heatmap shows and periods to show.
- Read the results. Results update automatically. Read the plain-English interpretation, then use Copy table to paste APA-style tables into Word and Copy for the results sentence.
How to report cohort retention analysis in APA style
Report the cohort definition and period, the retention table or heatmap and the average retention at key periods (for example month 1, 3 and 6).
Frequently asked questions
What is a cohort in retention analysis?
A group of customers who made their first purchase (or signed up) in the same period, such as the same month.
How is retention calculated?
Customers in the cohort with at least one order in the period divided by the cohort size. Period 0 is always 100%.
How do I read a retention heatmap?
Read along rows to follow one cohort over time, down columns to compare cohorts at the same age, and along diagonals to spot calendar events.
Why does the weighted average matter?
It combines cohorts in proportion to their size and uses only cohorts old enough to reach each period.
Monthly or weekly cohorts?
Match the purchase cycle: weekly for apps and food delivery, monthly for most e-commerce, quarterly for infrequent purchases.
What does a flattening curve mean?
A stable core of loyal customers. A curve that keeps falling towards zero indicates a retention problem.
Related tools
See also
Learn this method step by step in the free course Customer and Marketing Analytics, with exercises and a certificate.
Browse all categories
Data cleaning and preparation · Describe and explore · Compare means · Nonparametric tests · Correlation and regression · Bayesian statistics · Categorical data · Reliability and agreement · Multivariate analysis · Structural equation modeling · Econometrics and time series · Panel data and causal inference · Marketing research · Marketing analytics · Survival, clinical and meta-analysis · Study planning and calculators · PLS-SEM