GRRGross Revenue Retention
NRR's honest sibling: no expansion, just survival.
Put this metric to work: explore FlowMRR's MRR retention analytics to track revenue movements, retention and forecasts from your Stripe account, or see how FlowMRR compares as a Baremetrics alternative for Stripe.
Same cohort and window as NRR, but each customer counts for at most what they paid twelve months ago: an upgrade counts at the starting value, a downgrade at the lower amount, a departure at zero. Expansion therefore cannot lift it, and GRR never exceeds 100%. It answers 'of what we had a year ago, how much did we keep?' without any upsell flattering.
Formula
GRR = Σ over customers paying 12 months ago of min(MRR today, MRR then) ÷ What they paid 12 months agoWorked example
12 months ago the cohort paid $100K. Since then, customers who left took $4K and downgrades $2K; others expanded by $12K, which GRR ignores → GRR = (100 − 4 − 2) ÷ 100 = 94%.
Other definitions you will see
A common shortcut is (opening MRR − contraction − churn) ÷ opening MRR from a movement report, either for one month (a figure close to 100%, not comparable with an annual one) or summed over twelve months. If that report nets expansion and contraction across customers, one account's upgrade hides another's downgrade and GRR reads higher than with FlowMRR's per-customer cap.
Illustrative comparisons
B2B SaaS, annualisedThese unsourced ranges are illustrative scenarios, not measured industry benchmarks, percentiles or recommended targets. Their qualitative labels are editorial interpretations. Compare your own cohorts using the same segment, period, currency and measurement basis.
Why it matters
If NRR is your growth story, GRR is your defence. A company with 130% NRR and 75% GRR is using expansion to paper over heavy customer loss — which only works while a few big accounts keep growing. A high, stable GRR is the foundation of revenue that compounds.
Common mistakes
- Reporting only NRR because it looks better.
- Forgetting that GRR is by construction ≤ 100% — if you got >100%, you made a mistake.
Calculating it from Stripe data
GRR uses the same inputs as NRR: subscription MRR per customer, normalised to a month (annual plans ÷ 12), with trialing, incomplete, incomplete_expired and paused subscriptions excluded and unpaid subscriptions treated as churned at the end of their last paid period. The cohort is the customers who were paying twelve months ago. The difference lies in how each customer is counted at the end: take the lower of their current MRR and their starting MRR. A customer who doubled counts at their starting value, one who downgraded counts at the lower amount, and one who cancelled counts at zero.
The per-customer cap is the point. If you instead subtract contraction and churn from starting MRR using a monthly movement report, check that expansion and contraction were not netted across customers first; otherwise one account's upgrade silently cancels another's downgrade and GRR is overstated. Net within a customer (moving from one plan to another at the same price is not a loss), never between customers.
Watch for movements that look like losses but are not: a customer switching from a monthly to an annual plan of the same monthly value, a buyer moving to a new Stripe customer record, or a currency change. Normalise every interval to a monthly value, group customer records that belong to the same buyer, and convert both months at one exchange rate. A paused subscription bills nothing, so it counts as a loss only if it is still paused at the end of the window. FlowMRR derives GRR from the same Stripe cohort as its NRR, with each customer capped at their starting MRR.
Reading the result
GRR isolates the leak. Expansion cannot lift it, so the only ways to raise it are fewer cancellations, fewer downgrades and fewer failed payments that end in churn. That makes it the cleanest test of whether the product keeps the customers it already has.
Split the loss into its sources before acting. Involuntary churn, where cards failed until the subscription ended, is often the cheapest to cut through retries and dunning emails. Downgrades point to pricing or packaging, typically customers paying for capacity they do not use. Cancellations call for exit reasons and onboarding work. A GRR that falls while NRR holds steady means expansion is covering a growing leak, which lasts only as long as the expanding accounts keep growing.
Frequently asked questions
What is a good GRR for SaaS?
It depends mostly on customer size and contract length: annual contracts with larger businesses usually lose revenue more slowly than monthly plans sold to very small companies. Compare with businesses that have similar plans and customers, on the same 12-month cohort basis; the trend of your own GRR is often more informative than any outside comparison.
Can GRR be over 100%?
No. Each customer counts for at most their starting revenue, so GRR tops out at 100%, reached only when no customer cancelled or downgraded. A figure above 100% means expansion leaked into the calculation.
What is the difference between GRR and gross MRR churn?
Gross MRR churn is a monthly rate: in FlowMRR, the MRR of customers who stopped paying in one month ÷ MRR at the start of that month, with downgrades reported separately (many tools include them). GRR follows one cohort over twelve months and counts both departures and downgrades, so it shows what is left after a year of such losses. The two do not convert exactly, because the cohort ignores customers acquired during the year.
Is GRR more important than NRR?
Not more important, but harder to flatter: a few expanding accounts can push NRR up while many smaller ones leave. GRR shows how durable the revenue base is; NRR shows how much it grows on its own. Report both.