"100% NRR is the threshold" is one of those benchmarks that sounds right until you apply it. A $3K ACV SMB tool and a $200K ACV enterprise platform operate under completely different retention physics — comparing them to the same number is like grading a sprinter and a marathoner on the same split time. The real diagnostic isn't NRR alone. It's the decomposition into GRR and expansion rate that nobody publishes — and that tells you whether your retention number is healthy or masking a crisis.
What is a good NRR for SaaS?
Median net revenue retention across all B2B SaaS sits between 100% and 105%. That number is almost useless on its own. NRR ranges from 85% at the low end (SMB, low-touch, monthly billing) to 140%+ at the high end (enterprise, usage-based, multi-product). A company reporting 100% NRR could be perfectly healthy or actively dying, depending on its ACV tier and go-to-market motion.
The variation is structural, not accidental. Low-ACV products face higher churn rates because switching costs are low and buyers make emotional purchase decisions. High-ACV products retain better because procurement cycles are long, integrations run deep, and expansion happens through seat growth, usage increases, and module upsells. A single NRR benchmark that ignores ACV is measuring the average temperature of a hospital — technically correct, clinically useless.
The right question is never "is my NRR good?" It is "is my NRR good for my ACV tier, and is the underlying retention healthy or is expansion papering over churn?" That second question requires decomposing NRR into its components — which most benchmark reports skip.
NRR formula — and why it's deceptively simple
The standard formula
Net revenue retention measures how much revenue you retain from your existing customer base over a given period, including expansion, contraction, and churn. The formula:
NRR = (Starting MRR − Churned MRR − Contraction MRR + Expansion MRR) / Starting MRR
A company that starts the month with $100K MRR, loses $3K to churn, $2K to downgrades, and gains $8K from upsells has an NRR of 103%. The formula is one fraction. The complexity is in the inputs.
Net Revenue Retention
Revenue retained from existing customers including expansion, contraction, and churn.
"Starting MRR" means the cohort's MRR at the beginning of the period — only customers who were paying at that point. New customers acquired during the period are excluded entirely. This is where most spreadsheet implementations break. If new MRR leaks into the numerator, NRR inflates. If a reactivated customer is counted as "existing" instead of "new," the same inflation happens. The formula is simple. The classification logic is not.
NRR vs NDR — same metric, different names
Net revenue retention (NRR) and net dollar retention (NDR) are the same calculation. NDR is the term used in most public company filings and investor decks. NRR is more common in operator communities and billing tools. Some companies use "net MRR retention" when computing monthly and "NDR" when annualizing. The distinction is cosmetic — if someone quotes you an NDR of 115%, they mean the same thing as NRR 115%.
NRR benchmarks by ACV tier
ACV tier is the single strongest predictor of NRR. It dominates stage, vertical, and GTM motion. A $5K ACV product at $20M ARR and a $5K ACV product at $2M ARR will have more similar NRR profiles than two $20M ARR companies at different ACV tiers. The table below reflects aggregated benchmarks from billing data, public filings, and investor surveys from 2024–2026.
| ACV Tier | Median NRR | Top Quartile | Bottom Quartile |
|---|---|---|---|
| < $5K (SMB) | 85–100% | 105%+ | < 80% |
| $5K–$50K (Mid-market) | 95–110% | 115%+ | < 90% |
| $50K+ (Enterprise) | 105–130% | 140%+ | < 100% |
< $5K ACV — 85–100%
SMB SaaS products live in a high-churn, low-expansion environment. Monthly logo churn rates of 3–5% are common. Expansion is limited because seat counts are small and usage-based pricing is rare at this tier. Median NRR of 85–100% means most SMB products lose revenue from their existing base every month — they grow by acquiring new customers faster than they lose existing ones.
An NRR above 100% at this ACV tier is genuinely exceptional. It typically requires either a usage-based pricing model (where customers naturally grow into higher tiers) or a product-led expansion motion (where teams adopt the tool organically and seat count grows without sales involvement). If your SMB product is at 95% NRR with GRR above 85%, you are outperforming most of your peer set.
$5K–$50K ACV — 95–110%
Mid-market is where NRR starts to exceed 100% for median performers. Churn rates drop to 1–2% monthly because switching costs increase — integrations are deeper, procurement cycles are longer, and the product is more embedded in workflows. Expansion comes from seat growth, tier upgrades, and add-on modules.
The mid-market tier is also where the GRR + expansion decomposition matters most. A 105% NRR could mean GRR of 92% with 13% expansion (healthy) or GRR of 82% with 23% expansion (concerning). Both produce the same headline number. The first company has a sustainable retention engine. The second is running an expansion treadmill to offset a leaky bucket — and treadmills break.
$50K+ ACV — 105–130%+
Enterprise NRR above 100% is the norm, not the exception. Long contracts, high switching costs, and multi-threaded relationships keep GRR above 90% for most enterprise products. Expansion comes from department-level rollouts, usage growth, and platform add-ons. The companies posting 130%+ NRR — Snowflake, Datadog, Twilio in their growth phases — combine consumption-based pricing with deep platform lock-in.
An enterprise product with NRR below 100% has a serious problem. At this ACV tier, the structural forces favor retention. Falling below 100% means either the product is losing competitive bake-offs at renewal, contracts are being renegotiated downward, or the customer base is consolidating (M&A reducing seat counts). Any of those signals warrants immediate investigation.
Decomposing NRR into GRR + expansion rate
NRR is the sum of two forces: how much revenue you keep (GRR) and how much you grow from that kept base (expansion rate). The decomposition:
NRR = GRR + Expansion Rate
GRR measures revenue retained from existing customers excluding any expansion — it can never exceed 100%. Expansion rate measures additional revenue from those same customers as a percentage of starting MRR. Together they reconstruct NRR completely.
Gross Revenue Retention
Revenue retained from existing customers excluding expansion — the floor of your retention.
This decomposition is the diagnostic that most NRR discussions miss. An NRR of 110% can arise from two completely different businesses:
Company A: GRR 90% + 20% expansion = 110% NRR. The base is sticky. Expansion adds a healthy layer on top. This is a durable retention profile.
Company B: GRR 70% + 40% expansion = 110% NRR. The base is hemorrhaging. Aggressive upselling masks a retention crisis. The moment expansion slows — and it always does during macro contractions — NRR collapses to 70%.
The GRR floor — why below 80% is a red flag
GRR below 80% means the company loses more than 20% of its existing revenue annually before any expansion. At that rate, the customer base turns over almost entirely every four years. No expansion engine can sustainably compensate for that level of attrition — it requires selling increasingly large expansions to a shrinking base.
The math is unforgiving. If GRR is 75%, the company needs 25%+ expansion just to hold NRR at 100%. That level of expansion requires either aggressive price increases (which accelerate churn) or rapid usage growth (which depends on product-market fit that the GRR number already questions). Median GRR across all SaaS sits between 85% and 90%. Below 80% is bottom-quartile regardless of ACV tier.
Investors who see GRR below 80% paired with high NRR know exactly what it means: the expansion engine is compensating for a retention problem. The question shifts from "is this company growing?" to "how long can it keep expanding fast enough to mask the churn?" The answer is usually two to four quarters.
The post-2022 NRR compression
NRR across B2B SaaS compressed roughly 5–10 percentage points from 2021 peaks. The median company that reported 110% NRR in 2021 was reporting 100–105% by 2024. Top-quartile companies that hit 130%+ during the expansion boom settled into the 115–120% range. The compression was broad-based — it affected SMB, mid-market, and enterprise tiers, though enterprise held up better.
Three forces drove it. First, budget scrutiny increased across virtually every buyer segment. CFOs audited software spend, eliminated redundant tools, and renegotiated contracts at renewal. Second, seat-based expansion stalled as hiring slowed — fewer new employees meant fewer new seats. Third, usage-based products saw consumption decelerate as companies optimized their infrastructure spend.
The compression is important context for benchmarking. A company reporting 105% NRR in 2026 is performing as well as a company that reported 110–115% in 2021 — the macro environment has shifted the distribution. Comparing current NRR to 2021-era benchmarks systematically understates performance. Use 2024–2026 data as the reference frame.
Is 120% NRR good?
Yes — 120% NRR is top-quartile at any ACV tier. It means the existing customer base generates 20% more revenue each period without any contribution from new logos. At $10M ARR, that is $2M of organic revenue growth per year from the installed base alone.
But "good" still requires the decomposition. A 120% NRR with GRR of 92% and 28% expansion is excellent — the base is sticky and customers are expanding meaningfully. A 120% NRR with GRR of 78% and 42% expansion is a ticking clock. The company is losing more than a fifth of its base annually and compensating with aggressive upsells that cannot compound forever.
Public companies with sustained NRR above 120% — Snowflake (127%), Datadog (125%), CrowdStrike (120%) — share three characteristics: consumption-based or usage-linked pricing, deep platform integrations that raise switching costs, and multi-product portfolios that enable cross-sell. If your pricing model lacks these expansion vectors, NRR above 120% is difficult to sustain and the right target is probably 105–115%.
Tracking NRR from billing data
Computing NRR from raw billing events requires solving the classification problem: which subscription changes are expansion, which are contraction, and which are churn. Trial-to-paid conversions, mid-cycle plan changes, annual-to-monthly switches, and pause/resume cycles all need consistent handling. Most spreadsheet implementations get at least one of these wrong, which is how a company ends up with three different NRR numbers from three different people.
North Metric computes NRR, GRR, and expansion rate directly from Stripe subscription events with the classification logic built in. The decomposition is surfaced by default — not as an advanced view but as the primary display — so the GRR + expansion split that this article argues is essential is visible the first time you open the dashboard.
The goal is to make the diagnostic question easy to answer: is my retention profile durable, or is expansion masking a churn problem? One number cannot answer that. Two can.