Start studying
Module 14 · Data Science in Practice Module contents
Module 14  ·  Card 6 / 12intermediate

Retention and Churn Metrics

Retention is the share of users who are still active after a given time, and churn is the share who leave.

Over the same period, churn equals 1 minus retention, so 90% monthly retention means 10% monthly churn.

For example, a service that starts a month with 1,000 subscribers and loses 50 has a churn rate of 5% for that month.

Small monthly churn compounds: 5% a month leaves only about 54% of a cohort after a year.

Churn must be defined carefully, such as cancelled or inactive for 30 days, since each definition gives different numbers.

Learn in depth ↓
Module 14 · Data Science in Practice · 12 concepts
In depth  /  15 lines

Retention and Churn Metrics, in more detail

Retention and churn measure whether customers stay, which drives the value of most subscription and app businesses.

Winning customers is expensive, so keeping them is often the cheapest growth available.

Small differences in churn grow into large differences in customers over time.

Retention after n days is the share of a cohort still active on day n.

Churn rate over a period is the customers lost divided by the customers at the start.

Customers who join and leave within the same period complicate the count, so definitions must say how they are treated.

Subscription businesses usually define churn by cancellation, while apps usually define it by inactivity.

Inactivity needs a window, such as 30 days without opening the app, which is a judgement call.

Revenue churn counts lost revenue rather than lost customers, since large customers matter more.

Net revenue retention adds upgrades from the customers who stay, and can exceed 100% when they grow.

Monthly churn compounds, so 3% a month loses about 31% of customers in a year.

Average customer lifetime is roughly 1 divided by the churn rate, so 5% monthly churn implies about 20 months.

Churn is usually highest in the first weeks, which is why early retention gets so much attention.

In practice retention is reported by cohort, so improvements in new cohorts are visible.

The usual mistake is comparing churn figures that use different definitions, such as cancellations against inactivity.

Want the full technical article? Wikipedia ↗
Test yourself

One question on this card. Every option explains itself once you pick, right or wrong.

Question 1
A subscription service keeps 95% of its subscribers each month. Roughly what share of a cohort is left after 12 months?
Data Science in Practice

12 concepts in this module, then a graded test with every answer explained.

Study this module →
↑↓ move↵ openesc close