Data and Business Intelligence Glossary Terms

A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
V
W
X

What is churn?

Also known as

Attrition

Churn is the loss of customers, subscriptions, or recurring revenue over a period of time. It’s the mirror image of retention: if 100 customers start the month and 5 cancel, churn is 5% and retention is 95%.

The word “churn” on its own is ambiguous. Before you compare your number to anyone else’s — or to last quarter’s — you need to say which churn you mean.

Customer churn vs. logo churn vs. revenue churn

  • Customer churn counts people or accounts that stopped paying, divided by the number who could have. A team with 100 customers that loses 5 has 5% customer churn regardless of what those customers paid.
  • Logo churn is the same idea applied to accounts in a B2B context — one “logo” is one company, no matter how many seats or contracts it holds. It’s customer churn under a different name.
  • Revenue churn counts money instead of accounts. Losing five $50/month customers and losing one $250/month customer are identical on a revenue basis and very different on a customer basis.

Which one is right depends on the decision. Customer churn tells you about product fit and onboarding; revenue churn tells you what happened to the business. Small-business-heavy companies usually see customer churn far above revenue churn, because the accounts that leave are the cheap ones. If the two diverge in the other direction, you’re losing your biggest accounts, which is a much louder alarm.

Gross vs. net

Gross churn only counts what you lost. Net churn subtracts expansion — upgrades, seat growth, usage overage — from the losses in the same period. Net churn can be negative, meaning the existing base grew even though some customers left. The retention-side versions of these are gross revenue retention and net revenue retention, and most companies report both: GRR shows how leaky the bucket is, NRR shows whether the bucket fills itself.

Getting the number out of your data

Churn is deceptively hard to compute because the denominator, the period, and the definition of “canceled” all have to be pinned down first — a downgrade isn’t a cancellation, a failed payment isn’t always a churn, and annual contracts only get the chance to churn once a year. Break it down by segment (plan, size, acquisition channel, signup month) rather than watching a single company-wide average, since aggregate churn hides the cohorts that are actually failing.

For the calculation itself — SQL patterns, the data you need, and the pitfalls in each one — see the churn rate metric guide.

Related terms

Put it to work

Was this helpful?

Thanks for your feedback!