The percentage of recurring revenue you retain from existing customers before counting any expansion.
GRR = ((starting MRR − contraction − churn) ÷ starting MRR) × 100
A cohort starts at £100,000 MRR and loses £4,000 to downgrades and £6,000 to cancellations (expansion ignored).
((£100,000 − £4,000 − £6,000) ÷ £100,000) × 100 = 90% GRR
GRR measures how much of your existing revenue you keep, counting only losses from downgrades and cancellations. Crucially, it ignores expansion revenue — so unlike NRR, GRR can never exceed 100%.
Because expansion can mask underlying churn, GRR is the more conservative retention measure. A business can post 110% NRR while a few large accounts are quietly churning; GRR strips out the expansion and shows the raw leakage.
Reading GRR and NRR together is powerful. A high GRR with an even higher NRR means a sticky base that also expands. A low GRR propped up by expansion is a warning: the bucket is leaking and you are relying on upsells to hide it.
GRR shows the true durability of your revenue with no help from upsells. It exposes churn that expansion might otherwise paper over, which is why investors look at GRR alongside NRR to judge how solid the foundation really is.
GRR of 90%+ is strong for SMB SaaS; enterprise products often exceed 95%. GRR is always at or below NRR.
No. GRR ignores expansion revenue, so it can only ever be at or below 100%. Only net revenue retention can exceed 100%.
NRR can hide churn behind expansion. GRR strips expansion out to show raw retention, so reading both together reveals whether a healthy NRR rests on a solid base or on upsells masking leakage.
Connect your Stripe account and see your real MRR, churn, and LTV in real time — on desktop and mobile.
No credit card required · Connect Stripe in 1 click
No credit card required. Connect Stripe in 1 click.
We're building iOS and Android apps that'll bring your metrics everywhere.
Our mobile apps are currently in active development. Follow us on social media for updates and be the first to know when they launch.