NRR vs GRR: the two retention numbers, and why you need both

Revenue retention comes in two flavors, and the difference between them is not pedantry. NRR tells you whether the customers you already have are growing you. GRR tells you how fast the bucket leaks before anything refills it. A business can post a beautiful NRR while its GRR quietly falls apart, and if you only track one of the two, that is exactly the failure mode you will not see coming.

This post is the whole comparison: what each number measures, the same month computed both ways, what good looks like, and which one to quote to whom. The one-screen references live in the glossary: net revenue retention and gross revenue retention.

The two questions

NRR

Did the revenue we already had grow or shrink, all in?

(starting MRR + expansion + reactivation − contraction − churn) ÷ starting MRR × 100

What moves it: Expansion and reactivation push it up; downgrades and churn pull it down. New customers never touch it.
Who asks for it: Investors, first. Above 100% means the business grows with zero new sales.

GRR

How much of the revenue we already had did we keep?

(starting MRR − contraction − churn) ÷ starting MRR × 100

What moves it: Only losses: downgrades and churn. Expansion cannot repair it, so it never exceeds 100%.
Who asks for it: Anyone who wants retention without the makeup. It is the floor NRR stands on.

The structural difference is one term: expansion (plus reactivation, if you count returning customers). NRR includes it, GRR forbids it. Everything else about the two formulas is identical, which is why the gap between them is such a useful number: it is exactly how much work your expansion motion is doing.

Both metrics share one hard rule: new customers never count. Revenue retention is a statement about the cohort you started the period with. New business lands in MRR growth, not here. The single most common NRR inflation mistake is letting new logos leak into the numerator, and it turns the metric into a slower-moving copy of MRR growth that measures nothing.

The same month, computed twice

Take a book that starts the month at $50,000 MRR. During the month, existing customers upgrade by $3,200, one churned customer comes back for $400, downgrades remove $1,100, and cancellations remove $2,400.

Movement	NRR counts it	GRR counts it
Expansion, +$3,200	Yes	No
Reactivation, +$400	Yes	No
Contraction, −$1,100	Yes	Yes
Churn, −$2,400	Yes	Yes
New business (any amount)	No	No

NRR: (50,000 + 3,200 + 400 − 1,100 − 2,400) ÷ 50,000 = 100.2% GRR: (50,000 − 1,100 − 2,400) ÷ 50,000 = 93.0%

Same month, same customers, and the two numbers tell different stories. NRR above 100% says the installed base grew without a single new sale. GRR at 93% says you lost 7% of existing revenue in one month, which annualizes catastrophically, and expansion is currently papering over it.

Try your own numbers:

Starting MRRExpansionReactivationContractionChurned MRR

NRR

100.2%

GRR

93%

Full calculator

Same inputs, two answers: the gap between them is exactly your expansion and reactivation.

The gap is the diagnosis

Reading the two together is where the insight lives, because each combination points at a different problem:

NRR	GRR	What it means
High	High	The healthy quadrant: you keep revenue and grow it
High	Low	Expansion is masking a leak. Works until expansion slows, then both collapse at once
Low	High	You keep customers but cannot grow them: pricing or packaging problem, not a churn problem
Low	Low	The product is not retaining. Fix this before spending anything on acquisition

The high-NRR-low-GRR case is the dangerous one, because the headline number looks great. A few large accounts expanding hard can carry NRR past 100% while the long tail churns out underneath them. That is concentration risk wearing a growth costume, and GRR is the only one of the two numbers that refuses to hide it.

What good looks like

Benchmarks move around by segment more than by year, so ranges beat single numbers:

Metric	Median	Strong	Segment effect
NRR	103% to 105%	110%+	Enterprise ~118%, mid-market ~108%, SMB ~97%
GRR	around 90%	95%+	Enterprise sits high, SMB structurally lower

Two readings worth internalizing. First, if you sell to SMB and your NRR is below 100%, that is largely your market, not your product: small businesses die and downgrade at rates no onboarding flow fixes. Second, GRR has a ceiling of 100% by construction, so the distance between your GRR and 100 is your gross leak, and 95% GRR means you replace a twentieth of your book every period just to stay level.

Which one to quote, when

Fundraising: NRR, and investors will ask for it by name. Above 100% is the "the base grows itself" story, and it compounds: at 110% NRR, revenue doubles from existing customers alone in about seven years.
Diagnosing retention: GRR. It is the honest one, because nothing can dress it up.
Board reporting: both, side by side. The pair costs one extra line and removes the single most common way a retention slide misleads.
Targets for the team: GRR for customer success (they own the leak), NRR for account management (they own expansion on top of it).

Computing them from movements, not from memory

Both formulas are trivial arithmetic. The work is in the inputs: expansion, contraction, churn and reactivation have to be classified correctly, per customer, every period, with the edge cases (proration, currency drift, cancel-and-resubscribe) handled the same way every time. That classification is a movement ledger, and once it exists, both retention numbers fall out of it.

That is also what makes them answerable in a conversation instead of a spreadsheet:

What were our NRR and GRR last quarter?

From the movement ledger, Q2 on the April 1 base of $48,200: expansion $4,110, reactivation $350, contraction $2,050, churn $3,180. NRR: 97.5%. GRR: 89.2%. The 8.3-point gap says expansion is doing real work, but the base still shrank: you need $1,220 of net new business a quarter just to stand still. GRR below 90% is the number to fix first; the churn concentrates in 11 monthly Starter accounts.

The Kometrics connector in Claude, computing both from the same movements. Demo workspace data.

Kometrics builds that ledger from your billing data (free under $1,000 MRR), and the NRR calculator is free if you just need the arithmetic once.

FAQ

Can NRR be over 100%?

Yes, and that is the point of it. NRR above 100% means expansion and reactivation outgrew contraction and churn, so existing customers alone grew the business. Best-in-class B2B SaaS runs 110% to 130%.

Can GRR be over 100%?

No. GRR only subtracts losses from the starting base, so 100% is a hard ceiling. A GRR above 100% means new business or expansion leaked into the calculation.

Does NRR include reactivation?

Conventions differ. We include it, and say so: a churned customer returning is base revenue recovered, not a new sale. What matters more than the choice is writing it down; two teams with different reactivation conventions will report different NRRs from identical data.

Is NRR the same as net dollar retention (NDR)?

Yes. NRR, NDR and net revenue retention are the same metric under different names. GRR likewise appears as gross dollar retention.

What period should retention be measured over?

Annual is the convention investors expect, monthly is the one you operate on. Monthly retention compounds: 99% monthly GRR is roughly 89% annual, and 98% monthly is 78% annual, which is why small monthly slippage reads as fine and ends the year not fine.