What is Retention Rate?
Formula
Retention Rate = 100% − Churn Rate (same period, same basis)
- Churn Rate
- The loss rate on the identical unit, period and denominator — customer churn gives customer retention, gross revenue churn gives gross revenue retention
- Same basis
- Both figures must share a definition. A customer retention rate cannot be derived from a revenue churn rate
Worked example
A monthly subscription business with 2.0% monthly customer churn, viewed over increasing horizons.
| Step | Value |
|---|---|
| 1Monthly customer churn | 2.0% |
| 2Monthly retention rate | 98.0% |
| 3Retention after 3 months (0.98 ^ 3) | 94.1% |
| 4Retention after 6 months | 88.6% |
| 5Retention after 12 months | 78.5% |
| 6Retention after 24 months | 61.6% |
| 7Implied average lifetime (1 ÷ 0.02) | 50 months |
Result
98% monthly retention keeps roughly four in five customers through a year and three in five through two. The implied 50-month average lifetime comes from a constant-hazard assumption real cohorts rarely satisfy — measured cohort curves flatten, so late retention usually beats this projection while early retention falls short of it.
The mirror of churn, with the same caveats
Retention rate carries no information that the matching churn rate does not. It is the same measurement expressed from the other side, and every definitional choice underneath churn — customer or revenue, gross or net, which denominator, whether trials count — applies identically here. A retention rate quoted without its basis is exactly as ambiguous as a churn rate quoted without one.
The reason to have both words is presentational and it is not trivial: retention is the framing that supports comparison across horizons. Nobody intuitively compounds churn, but everyone reads a retention curve. Which is why the useful form of this metric is almost never a single percentage — it is a curve over elapsed months, which is cohort analysis.
Customer retention and revenue retention are different animals
The single most common confusion in this family. Customer retention rate is bounded at 100%: you cannot end a period with more of the original customers than you started with. Net revenue retention is not bounded, because expansion within surviving accounts can push the cohort's revenue above where it began.
So "our retention is 115%" is only coherent as a revenue statement. If someone says retention without qualification and the number exceeds 100, they mean NRR. If it is below 100 it could be any of four metrics, and it is worth asking which.
Retention compounds, so short windows mislead
A 98% monthly retention rate is not 98% over a year — it is 78.5%, because retention multiplies rather than averaging. This is the same compounding that makes annualising churn by multiplying by twelve wrong, seen from the other direction, and it is why monthly retention percentages sound so much healthier than they are. Two points of monthly retention, the difference between 98% and 96%, is the difference between keeping 78% and 61% of a cohort through the year.
Constant hazard versus a real curve
The compounding above assumes every customer carries the same probability of leaving in every month. Real cohorts do not behave that way: cancellation is front-loaded as poorly-fitting customers discover the mismatch, and the survivors then churn far more slowly than the early average implies. The curve flattens; the geometric model never does.
The practical consequence is that a retention rate computed from a young base overstates early loss and understates the durability of the survivors — and any LTV derived from it inherits both errors. Once you have twelve months of history, read the measured cohort curve instead of projecting a single rate forward.
Where Retention Rate goes wrong
- Quoting a retention rate without saying whether it is customers or revenue, gross or net. The word alone is compatible with at least four different metrics that can differ by twenty points.
- Reporting a customer retention rate above 100%. It is impossible by construction — a closed cohort cannot gain members — and it always means reactivations or new customers have entered the numerator.
- Extending a monthly retention rate to a year by subtracting twelve times the churn. Retention compounds: 98% monthly is 78.5% annually, not 76%, and the gap widens fast at higher churn.
- Projecting a single retention rate forward when you already have cohort data. The constant-hazard assumption misprices both ends of a real curve, and cohort tables show the actual shape for free.
- Deriving a retention rate from one churn definition and comparing it to a benchmark computed from another. This is the same trap as churn benchmarking, one arithmetic step further from the source.
Related
Metrics that move with this one
No metric explains a business on its own. These are the figures that qualify, offset or explain Retention Rate.
Churn Rate
Churn rate is the share of customers or recurring revenue lost over a period, most often calculated as the number of customers who cancelled during a month divided by the number active at the start of it. There is no single correct churn rate: customer churn and revenue churn, gross and net, and start-of-period and average denominators all produce different figures from identical data, so a churn rate is only interpretable alongside the definition that produced it.
Learn moreNet Revenue Retention (NRR)
Net revenue retention (NRR, also called net dollar retention) is the recurring revenue a fixed group of existing customers generates at the end of a period, expressed as a percentage of what the same group generated at the start, including expansion and after churn and contraction. New customers are excluded entirely. Above 100% means the existing base grew on its own; it does not mean customers are staying, because heavy expansion from a few accounts can cover substantial churn among the rest.
Learn moreGross Revenue Retention (GRR)
Gross revenue retention (GRR, also called gross dollar retention) is the share of a cohort's starting recurring revenue still present at the end of a period, counting cancellations and downgrades but excluding all expansion. Because expansion is excluded, GRR can never exceed 100%, which makes it the only retention figure a strong upsell quarter cannot flatter.
Learn moreLogo Churn
Logo churn is the share of customer accounts lost over a period, counting each account once regardless of what it paid. It is the customer-count view of churn, and comparing it to revenue churn reveals whether the accounts leaving are larger or smaller than average — the two rates diverging is usually more informative than either level on its own.
Learn moreCohort Analysis
Cohort analysis groups customers by when they started and tracks each group separately over elapsed time, producing a triangular table where rows are signup periods and columns are months since signup. It exposes what an aggregate churn rate cannot: whether retention is improving for newer customers, where in the lifecycle customers leave, and whether a flat headline number is hiding a deteriorating base propped up by durable older cohorts.
Learn moreCustomer Lifetime Value (LTV)
Customer lifetime value (LTV, also written CLV or CLTV) is the total gross profit a business expects to earn from one customer across the whole of their relationship. The standard subscription estimate divides average revenue per account by the customer churn rate and multiplies by gross margin, but that formula assumes every customer has the same constant probability of cancelling every month — an assumption real cohorts violate — so LTV is a directional planning input rather than a measured figure.
Learn moreRetention Rate: frequently asked questions
How do I calculate retention rate?
Can retention rate be over 100%?
What is the difference between retention rate and churn rate?
Why does my annual retention look so much worse than my monthly?
Stop recalculating Retention Rate by hand.
Connect Stripe and Bastle computes this metric — and the rest of the glossary — against your own billing history, with the definitions under your control. Free while in beta, no card required.