What's the difference between NRR and GRR — and which one does your board actually care about?

Direct Answer

GRR measures pure retention — what survives churn and downgrades, capped at 100%. NRR adds expansion on top, so it can exceed 100%. Boards at growth-stage SaaS companies care more about NRR because expansion is the cheapest growth dollar you'll ever earn, but sharp CFOs watch GRR closer because NRR hides churn behind upsells.

The honest answer: both belong in the deck, side by side. If a board only asks about NRR, that board is being lied to — or doing the lying themselves. Track both, alert on both, defend both.

TL;DR

• GRR = (start MRR minus churn minus downgrades) divided by start MRR. Caps at 100%. World-class is 95%+.
• NRR = (start MRR minus churn minus downgrades plus expansion) divided by start MRR. Public SaaS median is roughly 108%; top quartile 120%; world-class 135%.
• Worked example: $10M start MRR, $700K of churn and downgrades, $2M of expansion gives a GRR of 93% and an NRR of 113%.
• Boards favor NRR because expansion drives compounding growth at low CAC; CFOs favor GRR because it cannot be inflated by upsell motion.
• The three games to watch: counting acquisition roll-ups as expansion, counting raw price hikes as expansion, and counting FX tailwinds as expansion.

The Math (with a worked example)

The formulas look almost identical, and that visual similarity is exactly why the two get confused in board meetings. GRR — gross revenue retention — takes the MRR a cohort had at the start of a period, subtracts everything that left (full cancellations, called gross churn) and everything that shrank (downgrades, seat reductions, plan compressions).

The denominator is the starting MRR. The result is bounded between zero and one hundred percent because the numerator can only shrink. NRR — net revenue retention, sometimes called net dollar retention or NDR — uses the same starting cohort but adds the dollars that expanded inside it: seat increases, tier upgrades, cross-sell of new SKUs, usage overages on consumption pricing.

Because expansion can be unlimited, NRR has no ceiling.

Walk through a concrete cohort. A vertical SaaS company starts a quarter with $10.0M MRR locked in from the customers it had on day one. During the quarter, customers worth $500K cancel outright — that's gross churn.

Another set worth $200K downgraded from Enterprise to Pro, or dropped twenty seats apiece — that's downgrade MRR. Those two together are $700K of contraction. So retained MRR is $9.3M, and GRR is $9.3M divided by $10.0M, or 93%.

Now layer in expansion: the same cohort grew by $2.0M through more seats, premium add-ons, and a new analytics module. Ending cohort MRR is $11.3M. NRR is $11.3M divided by $10.0M, or 113%.

Note that NRR uses the same denominator as GRR — the starting cohort, not the new logos added during the period. New-logo ACV never enters either calculation. That is the single most common error in homemade NRR dashboards.

Benchmarks by Segment

Benchmarks vary widely by ACV band and motion. The numbers below blend the Bessemer 2024 State of the Cloud Report, ICONIQ's 2024 Growth and Efficiency benchmarks, and the Meritech/Battery public-SaaS comp sets. Read them as directional, not absolute — and always pair them with your own cohort age, because a five-year-old cohort behaves nothing like a one-year-old cohort.

For context on the absolute ceiling: Snowflake reported NRR peaks above 170 percent in 2021 and still runs in the 125 to 135 percent zone. Datadog has held NRR above 130 percent for the better part of a decade. MongoDB, Cloudflare, and HashiCorp all run consumption motions that push NRR structurally higher than seat-based peers.

Public SaaS median NRR sits around 108 percent in current Bessemer cuts, down from a 120 percent peak in 2021 — the post-ZIRP compression. Treat 110 percent NRR as "good" for a growth-stage seat-based business, 120 percent as "great," and 130-plus as the rarefied air reserved for consumption pricing and best-in-class land-and-expand motions.

How Teams Game NRR + How to Detect It

There are three classic ways NRR gets inflated to flatter the board deck, and a good audit committee or due-diligence analyst can spot all three in under twenty minutes if they ask for the right cuts.

The first is roll-up expansion. When Customer A acquires Customer B and the combined entity signs one larger contract, weak accounting policies count the entire delta as expansion inside the original A cohort. The honest treatment splits it: A's pre-acquisition spend stays in the cohort, B's spend moves to "new logo" or to its own historical cohort.

Detection: ask for a cohort cut excluding any customer with an M&A event in the period. If NRR drops more than three points, you're being shown a roll-up story dressed as organic expansion.

The second is price-increase expansion. When a vendor raises list prices ten percent across the renewal base and customers absorb it, that ten percent flows straight into expansion MRR. Technically true, economically misleading — there is no new value delivered, just a price lever pulled.

Detection: ask for NRR computed at constant pricing (last year's price card applied to this year's units). A gap of more than four points means price, not product, is doing the lifting.

The third is FX-tailwind expansion. International ARR reported in USD inflates whenever the dollar weakens. A Euro customer paying 1,000 EUR/month went from $1,080 to $1,150 in MRR without changing a thing. Detection: ask for constant-currency NRR. The gap between reported and constant-currency NRR is the FX game.

A fourth, related sleight-of-hand worth naming: cohort-definition drift. Some finance teams quietly change which customers count as the "starting cohort" — for example, excluding customers who churned in the first thirty days, or backfilling new logos signed mid-quarter into the opening base.

Both inflate NRR by shrinking the denominator or smuggling expansion into a cohort that did not exist on day one. The fix is policy discipline: lock the cohort definition in writing, audit it quarterly, and force any change to require CFO and audit-committee sign-off with restated history.

If the methodology changes without restatement, treat the new number as non-comparable and refuse to put it next to prior quarters on the same chart.

Frequently Asked Questions

Is NRR the same as DBNR or NDR? Yes. Net Revenue Retention, Net Dollar Retention, and Dollar-Based Net Retention are three names for the same metric. Some companies use slightly different period conventions (TTM vs quarterly cohort) but the formula is identical.

Why does Snowflake have 130-percent-plus NRR? Consumption pricing. Customers pay per query and per terabyte stored. As they migrate more workloads, usage grows without any new sales motion required. Seat-based businesses cannot mechanically reach those numbers without aggressive cross-sell.

What if our NRR is 95 percent? It means your cohort is shrinking. Expansion is not covering churn and downgrades. Diagnose immediately: is it logo churn, downsell, or expansion drought?

Each has a different fix. NRR below 100 percent at scale is an existential signal — investors will model your business as a leaky bucket and discount the multiple accordingly.

Sources

• Bessemer Venture Partners, *2024 State of the Cloud Report*
• ICONIQ Capital, *2024 SaaS Operating Metrics Benchmark*
• Meritech Capital, *Public SaaS Comps and NRR Tracker*
• Battery Ventures, *State of the OpenCloud 2024*
• OpenView Partners, *2024 SaaS Benchmarks Report*
• Pavilion, *2024 RevOps Benchmarks Survey*
• Snowflake Investor Relations, FY24 10-K Net Revenue Retention disclosures
• Datadog Investor Relations, FY24 10-K Dollar-Based Net Retention disclosures