Retention rate = (customers at end − customers acquired during) ÷ customers at start
The subtraction is the whole formula. Without it you are measuring growth and calling it retention.
It is the inverse of logo churn
Retention rate and logo churn are the same measurement stated two ways: 95% retention is 5% churn. Teams report both and occasionally compute them differently, which is how a deck ends up with 95% retention and 7% churn on adjacent slides.
Pick one as canonical and derive the other.
Why it diverges from revenue retention
A customer is one unit regardless of size, so retention rate weights a $500-a-year account exactly like a $500,000 one.
That divergence is informative rather than a flaw:
- High customer retention, low revenue retention: the leavers were big.
An enterprise problem.
- Low customer retention, high revenue retention: the leavers were small.
Often a self-serve or trial-conversion problem, and frequently acceptable.
Neither is visible from one number. Report retention rate beside net dollar retention or the pair says less than either alone.
A worked example
Start the quarter with 500 customers. End with 540. During the quarter you acquired 65.
Retention = (540 − 65) ÷ 500 = 475 ÷ 500 = 95%
Skipping the subtraction gives 540 ÷ 500 = 108%, which is not a retention rate. It is growth, and reporting it as retention is the most common error in this metric.
Choices to state
- What counts as a customer. Account, contract, or billing entity. A parent
with six subsidiaries is one or six, and the choice moves the number.
- Whether a pause counts as churn. Seasonal customers who suspend and
return will otherwise churn and reappear every year.
- The window. Monthly retention annualised by multiplication is wrong for
the same compounding reason as churn.
The traps
- Omitting the subtraction.
- Mixing segments with very different behaviour into one rate.
- Reading it alone, without revenue.
- Changing the customer definition after a pricing or packaging change without
annotating the series.
Where this sits
Retention is Measurement in the Tenbound Pipeline Architecture Standard, and it is the pillar where a Market failure surfaces last. Customers acquired outside the ICP retain worse, whatever the customer success motion does afterwards.