What share of revenue do we keep before counting upsell?
№ 005 · GRR cohort heatmap
Definition
MetricGross revenue retention
Unit% of year-ago cohort revenue, excluding expansion
Same structure as NRR but caps at 100% by excluding upsell, so only churn and downgrades pull retention down. Pull subscription revenue per cohort each month, subtract expansion, divide by month-zero. Healthy SaaS holds GRR above 90% for enterprise, 80% for SMB. Example: Jan cohort at 91% at M5; Apr cohort 96% at M2.
- GRR
- Gross revenue retention. Same as NRR but with expansion excluded, so it caps at 100 percent.
- NRR
- Net revenue retention. Cohort revenue this period divided by the same cohort revenue a year ago, including expansion, contraction and churn. New logos excluded.
- SaaS
- Software as a service. Software sold as a recurring subscription rather than a licence.
- SMB
- Small and medium business. Segment label, usually under 200 employees or a defined contract-value band.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| 82-85% or below | 88-90% | 93%+ |
Private B2B SaaS, $1M+ ARR, blended ACV, 2023-2025 data; GRR excludes upsell, cross-sell and price increases and cannot exceed 100% · Surveys disagree at the margin: Benchmarkit has the median sliding 90% to 88% over three years and calls it a canary in the coal mine; KeyBanc reported a dip to 86% in 2023 recovering toward 90%; SaaS Capital holds 90% as the table-stakes bar. The 88-90% median band reflects that spread.
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| ACV under $25k | under 85% | ~90% | 93%+ |
| ACV above $100k | under 90% | 92-95% | 96%+ |
When it looks bad
A vertical cold stripe: all cohorts, old and new, lose revenue in the same calendar months, which signals an external shock or a price change rather than cohort-quality decay.
Every cohort row drops 3-4 points in the same two quarters. That pattern usually means a competitor launch, a botched migration, or a price increase pushed to the whole base at once, not a bad vintage.
What to do about it
- Separate the two failure shapes before acting: diagonal decay (new cohorts worse) means acquisition quality drifted, fix ICP and onboarding; vertical stripes (all cohorts hit together) mean a base-wide event, fix the event.
- Recover the involuntary component first: failed payments run 20-40% of subscription churn (ProfitWell data), and dunning optimization lifts recovery toward 70% of failed invoices (Recurly); this is the cheapest GRR point available at low ACV.
- Hold GRR at 90%+ before funding expansion plays: SaaS Capital frames 90% as table stakes, and NRR built on sub-85% GRR is a few large expanding accounts masking a leaking base.
- For high-ACV books, invest in implementation depth: SaaS Capital attributes the higher GRR of high-ACV companies to scoping, implementation and dedicated account management; a paid onboarding package pays for itself in retained revenue.
Sources
- Benchmarkit GRR decreased from 90% to 88% median over the past three years; GRR rises with ACV, consistent across 4 years benchmarkit.ai ↗
- SaaS Capital GRR of at least 90% is table stakes for performance parity; 90% is the norm below $25k ACV, higher above saas-capital.com ↗
- KeyBanc Capital Markets and Sapphire Ventures Gross retention declined to 86% in 2023 and is expected to approach 90% in the near term sapphireventures.com ↗
- KeyBanc Capital Markets and Sapphire Ventures Gross retention ~90% and expected to remain consistent sapphireventures.com ↗
- ChurnZero (reporting the SaaS Capital B2B SaaS benchmarking survey) GRR robust at 90%, NRR around 100% in the surveyed base churnzero.com ↗
- SaaS Capital Companies with the very highest ACVs show the highest gross retention; implementation depth and account management make the product stickier saas-capital.com ↗