What is GRR (Gross Revenue Retention) and how does it differ from NRR?
Direct Answer
Gross Revenue Retention (GRR) is the percentage of recurring revenue you keep from your existing customer base after churn and downgrades, capped at 100% because it ignores expansion. Net Revenue Retention (NRR) is the same calculation plus upsell, cross-sell, and price-increase expansion — which is why it routinely runs above 100% in healthy SaaS.
GRR tells you if your product is sticky; NRR tells you if your account base is a growth engine. In 2027, the median private B2B SaaS company runs GRR around 88% and NRR around 101% per SaaS Capital, while best-in-class enterprise SaaS clears GRR 95%+ and NRR 120%+ per Bessemer Venture Partners.
1. The Two Formulas — And Why Boards Care About Both
Every operator should be able to write both formulas on a whiteboard in under thirty seconds. They share a numerator base but differ on what counts as "kept."
1.1 GRR formula
GRR = (Starting ARR − Churned ARR − Downgrade ARR) ÷ Starting ARR
Notice what is not in that formula: expansion, upsell, cross-sell, price increases. GRR is a defensive metric. It answers one question: of the dollars I had on January 1, how many do I still have on December 31? Maximum value is 100%. If yours is higher than 100%, you are calculating wrong.
1.2 NRR formula
NRR = (Starting ARR − Churned ARR − Downgrade ARR + Expansion ARR) ÷ Starting ARR
NRR adds expansion back in. It is an offensive metric. A 130% NRR means the existing customer cohort grew 30% without acquiring a single new logo.
Snowflake hit a peak 158% NRR in FY2022 on the back of consumption pricing. Twilio peaked at 155%. Datadog still posts ~120% NRR on $3.43B of 2025 revenue per their earnings disclosures.
1.3 Why investors weigh them differently
Series B and later boards now ask for both on every monthly deck. GRR sets the floor — it is the strongest single predictor of churn-driven valuation discounts. NRR sets the ceiling — it drives forward revenue multiples.
Bessemer's "good / better / best" framework (100% / 110% / 120% NRR) is now the de facto board scorecard at growth-stage SaaS.
2. What Gets Included — The Definitional Landmines
Most retention disasters in board meetings come from one team computing the metric one way and another team computing it differently. Lock the definitions before the next QBR.
2.1 What counts as churn
Churn includes logos that left entirely and subscriptions that did not renew. It does not include customers who downgraded — that's a separate bucket called contraction or downgrade. Both GRR and NRR subtract churn and contraction, but they must be reported separately so leadership can tell whether you have a logo problem or a price problem.
2.2 What counts as expansion
Expansion = upsell (more seats, more usage), cross-sell (new product line into the same logo), and price increase on renewal. NRR includes all three. GRR includes none. A common mistake: counting a renewal that simply rolls forward at the same ARR as "expansion." It is not. Flat = retained, not expanded.
2.3 What gets excluded entirely
New logo revenue is excluded from both metrics. So is one-time services revenue, professional services, and implementation fees. Retention is recurring-revenue only. If you sell a $50K implementation alongside a $200K ARR contract, only the $200K hits the retention calculation.
3. The 2027 Benchmark Map
3.1 By company stage
Per SaaS Capital 2026 and Pavilion 2026 B2B benchmarks:
- Pre-Series A / Seed: GRR median 75-85%, NRR median 95-105%. Volatile because cohorts are tiny.
- Series A ($1-5M ARR): GRR 85-90%, NRR 100-110%.
- Series B ($5-20M ARR): GRR 88-92%, NRR 108-115%.
- Series C+ ($20M+ ARR): GRR 90-95%, NRR 115-125%.
3.2 By ACV band
Per OpenView 2026 SaaS Benchmarks and Benchmarkit 2025:
- SMB SaaS (<$25K ACV): GRR median 85%, NRR median 97%. High churn baked in.
- Mid-Market ($25K-$100K ACV): GRR median 90%, NRR median 108%.
- Enterprise (>$100K ACV): GRR median 93%, NRR median 118%.
3.3 By pricing model
Consumption-based SaaS (Snowflake, Datadog, Cloudflare, MongoDB) systematically posts higher NRR because usage grows naturally with customer adoption — Datadog at ~120%, Snowflake at 125% in Q4 FY2026. Pure subscription SaaS sees lower expansion ceilings and clusters around 108-115% NRR at the median.
4. How to Move Each Number — They Need Different Plays
GRR and NRR look similar on a dashboard but require completely different operating motions to improve. Mixing them up is the most common mistake new RevOps leaders make.
4.1 Moving GRR up — the churn playbook
GRR is a Customer Success and product problem. The levers:
- Health scoring that actually predicts churn 60-90 days out — Gainsight and ChurnZero both publish reference models.
- Executive Business Reviews for the top 20% of ARR every quarter, run by named CSMs.
- Multi-year contracts with 10-15% discount — locks ARR for 24-36 months.
- Renewals owned by a dedicated team, not the original AE, per Andy Whyte's MEDDPICC renewal motion.
The 2025 Customer Revenue Leadership Study by ChurnZero found teams using a Customer Success Platform average 100% NRR versus 94% without — measurable infrastructure ROI.
4.2 Moving NRR up — the expansion playbook
NRR above 100% is an account management and product-led growth problem.
- Land-and-expand pricing — start with a single seat or team, expand by usage tier.
- Named expansion AEs (sometimes called Account Managers) carrying their own quota separate from new logo AEs.
- Consumption or hybrid pricing for technical products — every Snowflake-class NRR (>140%) is consumption-driven.
- Annual price increases built into the contract — 5-7% bumps compound into 5-7 NRR points.
4.3 The expansion-vs-retention tradeoff
A team that pushes expansion aggressively can mask underlying churn. A company can post 115% NRR while running 80% GRR if a few huge accounts expand fast. That is why boards now demand both numbers — NRR alone is gameable.
5. The Operating Cadence
5.1 What RevOps owns
The RevOps team owns the source of truth — the monthly ARR waterfall in Salesforce or HubSpot, tagged with retained / new / expansion / contraction / churn. Tools commonly stitching this together in 2027: Mosaic, Maxio (formerly SaaSOptics), ChartMogul, Drivetrain, Equals.
5.2 What Customer Success owns
CS owns GRR. They get measured on it. Nick Mehta at Gainsight has been beating this drum since 2018 — if CS is not on the hook for a retention number, the role drifts into a hospitality function.
5.3 What Account Management owns
AM owns NRR minus GRR — the expansion delta. Bridge Group 2026 reports Account Manager OTE in 2027 SaaS at $160-220K, with 40-50% variable tied to expansion ARR.
6. The Common Mistakes That Wreck the Number
6.1 Counting auto-renew price escalators as "expansion"
A 5% CPI bump on renewal is price increase expansion — it counts toward NRR but boards will discount it. Report it as a separate line.
6.2 Mixing cohorts
Reporting NRR across all customers at once hides churn in the newest cohort. Always report NRR by signing-year cohort. Bessemer publishes this view in every state-of-the-cloud report.
6.3 Confusing logo retention with revenue retention
Logo retention counts how many customers stayed. Revenue retention counts how many dollars stayed. They diverge dramatically: a company can lose 30% of its small logos but only 5% of its dollars if the big accounts stay.
FAQ
Q: Can GRR ever exceed 100%? No. GRR by definition excludes expansion. If your dashboard shows GRR over 100%, you are double-counting expansion or your ARR snapshot logic is broken. Audit immediately.
Q: What is the single most important retention metric for a Series B board deck? NRR by signing-year cohort, with GRR as a floor check. Bessemer, Insight Partners, and Iconiq all open with this view.
Q: Should renewals sit in Customer Success or Sales? Renewals belong to a dedicated renewal team at $20M+ ARR. Below that, the most common 2027 pattern is CS owns renewal motion, AM owns expansion. Force Management's MEDDPICC playbook treats renewal as its own qualification cycle.
Q: How does consumption pricing change the math? Consumption inflates NRR because usage growth shows up as expansion automatically. It can also depress GRR because customers can dial down without churning entirely. Report both gross and net carefully — Snowflake's 10-K breaks them out explicitly.
Q: Where do downsells go — GRR or NRR? Both subtract them. A customer who drops from $100K to $60K contributes $40K of contraction, which hits the numerator of both metrics. They are not new churn but they are not retained ARR either.
Bottom Line
GRR measures stickiness; NRR measures account-level growth. Run both, by cohort, every month. In 2027 the median SaaS company posts GRR ~88% / NRR ~101%, the top quartile clears GRR 95% / NRR 120%, and consumption-pricing leaders like Snowflake and Datadog clear NRR 125%+.
Build the operating cadence first (monthly ARR waterfall + tagged source of truth in Salesforce or HubSpot), then put Customer Success on the GRR hook and Account Management on the expansion delta. Get those two roles owning two different numbers and the entire retention motion compounds.
Sources
- SaaS Capital — 2026 Private B2B SaaS Retention Benchmarks Report
- Bessemer Venture Partners — State of the Cloud 2026 (GRR top-quartile thresholds, NRR good/better/best framework)
- OpenView Partners — 2026 SaaS Benchmarks (by ACV band)
- Pavilion — 2026 B2B SaaS Performance Benchmarks Report
- ChurnZero — 2025 Customer Revenue Leadership Study (CSP impact on NRR)
- Benchmarkit — 2025 SaaS Performance Metrics
- Bridge Group — 2026 SaaS Sales & AM Compensation Report
- Snowflake Inc. — FY2026 Q4 Earnings (125% NRR disclosure)
- Datadog — 2025 Annual Report (NRR ~120%, $3.43B revenue)
- Force Management — MEDDPICC Renewal Playbook (Andy Whyte / John Kaplan)