Net Revenue Retention Benchmarks for SaaS 2026: Is Your NRR Actually Normal?

The most-cited NRR benchmark for public SaaS is roughly 114%, and that number is close to useless for a sub-$1M ARR PLG team. Public medians are weighted toward enterprise, sales-led, high-ACV cohorts whose expansion mechanics a self-serve startup does not share. You need your row in a segmented grid, not the aggregate.

There Is No Single “Good” NRR (Here's the Segmented Answer Up Front)

The published numbers conflict because they segment differently, not because one source is wrong.

As reported by ICONIQ's State of GTM 2026 (a January 2026 survey of 150+ B2B GTM executives, via SaaStr), median NRR sits in the 108-110% range, with the top quartile above 123%. That figure covers AI-forward GTM teams and is the best available 2026 cross-stage anchor. As reported by SaaS Capital (independent study, September 2025), private B2B companies at $25K-$50K ACV run a 102% median. As reported by Optifai's Pipeline Study (vendor-reported, N=939, Q2 2025-Q1 2026), enterprise sits at 118%, mid-market at 108%, and SMB at 97%. Genuinely different “normals” for genuinely different companies.

The public SaaS aggregate of roughly 114% is an industry estimate across public filings, not a primary source with a methodology you can inspect. Public SaaS companies that report NRR are survivors with high ACVs and mature expansion motions. A $400K ARR PLG product shares almost none of those structural conditions.

Two things the floating benchmark line never explains: why NRR can exceed 100% while GRR is hard-capped, and how usage-based and AI-native pricing change what your NRR number means for cross-company comparison. Section 3 has your row.

NRR vs GRR: Why One Can Beat 100% and the Other Can't

NRR and GRR measure the same customer base but answer opposite questions, and the gap between them is your expansion engine made visible.

The formulas:

• NRR = (starting MRR + expansion - contraction - churn) / starting MRR
• GRR = (starting MRR - contraction - churn) / starting MRR

Expansion revenue is in the NRR numerator. It is absent from GRR by definition. That one difference is why GRR is hard-capped at 100%: you cannot retain more than 100% of what you started with once you strip out upsells, seat adds, and tier upgrades. NRR exceeds 100% precisely when expansion revenue outruns contraction plus churn. It is also why NRR does not include new customers: the metric tracks what your existing base did, not what your sales team closed last month.

A worked example that keeps the math internally consistent: a SaaS losing 8% to churn and contraction but gaining 18% from expansion in the same period reports approximately 110% NRR and approximately 92% GRR. Both numbers are correct. They describe different phenomena.

The spread between them is what a VP actually wants to read. A wide NRR-GRR spread (say, 110% NRR and 85% GRR) signals a healthy expansion motion: customers who stay are expanding meaningfully. A narrow spread with high NRR (say, 115% NRR and 112% GRR) can signal fragility. You are masking moderate churn with a small number of outsized enterprise upsells. If that expansion disappears in one renewal cycle, your NRR collapses. Catch the narrow-spread pattern before a board presentation, not after. Watching for early signals a customer is about to contract or churn makes that catch possible before the spread moves.

For GRR benchmarks by churn input, you can benchmark your churn rate the same way this article segments NRR, with gross logo and revenue churn broken out by stage and ACV.

The “Is Your NRR Normal?” Benchmark Grid (by Stage, ACV, and Motion)

Your expected NRR depends on three things: company stage (ARR band), ACV tier, and GTM motion. Here is the grid built from confirmed, named sources only. Every cell traces to a primary or secondary source named in the footnote column.

The pattern in that table is not accidental. NRR rises with ACV and contract size because bigger accounts carry more seats, modules, and usage to expand into. A $200K ACV enterprise account has natural room to grow to $280K; a $1,200 ARR self-serve account rarely does. Low-ACV, self-serve products structurally produce lower NRR, and an NRR near 100% in the SMB band is not a retention failure. It reflects the expansion ceiling that comes with the pricing architecture.

SaaS Capital's 2025 study also notes that across companies with more than $1M in ARR, those with NRR at or above 110% posted growth rates above the survey median of approximately 24%, while companies with NRR below 100% fell below it. NRR is not just a retention health signal; it is a growth-rate predictor. That correlation holds across ACV tiers, but the absolute NRR threshold that matters shifts by segment, which is exactly what the grid above captures.

For how NRR feeds the full efficiency picture, how NRR feeds your LTV/CAC math is worth tracing before you take the benchmark table into a board conversation.

Where Does a $400K ARR PLG Team Actually Sit?

If you run a self-serve, low-ACV product at sub-$1M ARR, the 120%+ figures built for enterprise sales-led SaaS will make you feel behind for structural reasons that have nothing to do with your performance.

Your row in the grid is the SMB band: approximately 97% median up toward 100-105%, per Optifai's vendor-reported segmentation. That is your structural peer group. An NRR around 100-105% for a PLG product at this stage is solid. Self-serve expansion (seat adds, plan upgrades, tier migrations) is smaller-ticket and slower than enterprise land-and-expand. A self-serve customer moves up a tier when they hit a usage wall; an enterprise account expands because a salesperson identified a second department. Different mechanics, different timelines, different revenue yields.

NRR is a lagging metric. A cohort needs to mature, usually two to three quarters, before NRR becomes readable. If you are being asked to prove channel ROI before that window closes, you need leading indicators. Pair GRR and expansion rate signals with your NRR tracking: upgrade rate (share of accounts that moved to a higher plan this month), seat-expansion rate (net seat growth within existing accounts), and feature-adoption-to-upgrade (whether accounts that hit Feature X upgrade within 30 days). These are the signals you are already reading in Mixpanel or Amplitude; they surface expansion momentum before the NRR line moves.

At a small customer base, a single large expansion or one churned mid-size account swings NRR hard. Small-N noise is real. Read the three-month trend rather than a single month. For the full metrics context behind what to track and when, the SaaS metrics that matter in the AI era covers the prioritization layer.

How AI-Native and Usage-Based Pricing Distort NRR

Usage-based and AI-native pricing change what an NRR number even means. Comparing your usage-priced NRR to a seat-priced competitor's is often comparing two different things.

Under consumption or usage-based pricing, NRR moves with customer usage volume, not retention decisions. A heavy-usage quarter can push NRR well above 120%; a usage dip (a team cutting API calls during a budget review) can drop it below 100% with no churn event. AI-native SaaS with per-token or per-agent pricing amplifies this swing further: a customer adding a new agent workflow and a customer pausing one both show up in NRR, even though neither is a retention outcome.

The honest picture on AI tooling: as reported by SaaStr in June 2026, 95% of enterprise AI pilots generate no financial return. That applies to enterprise pilots specifically. It does anchor a targeting-over-tooling principle that matters for NRR work. AI-assisted usage-health scoring can forecast expansion and contraction earlier by reading product-usage signals, and in-product upgrade prompts triggered by usage thresholds can lift expansion revenue at scale. These are concrete, bounded applications. ICONIQ's 2026 data shows AI-forward GTM teams posting 108-110% median NRR with a top quartile above 123%, which is useful evidence that the motion can support healthy retention, not a promise that AI tooling will move your specific number.

For save plays that lift retention before NRR can recover, the tactical retention playbook covers the workflows a small team can ship without a dedicated CS function.

Frequently Asked Questions

What is the NRR benchmark for SaaS?

There is no single benchmark. As reported by ICONIQ State of GTM 2026 (via SaaStr), the 2026 cross-stage median sits in the 108-110% range with the top quartile above 123%, covering AI-forward GTM teams. As reported by SaaS Capital (September 2025), private B2B companies at $25K-$50K ACV show a 102% median. The right benchmark depends on your stage, ACV tier, and GTM motion.

What is a good net revenue retention?

Above 100% is the common threshold, but “good” is segment-relative. For SMB or PLG self-serve products (ACV below $25K), 97-105% is structurally normal per Optifai's vendor-reported segmentation. For enterprise sales-led SaaS (ACV above $100K), 115%+ is a more realistic bar. Measuring against the wrong segment's benchmark turns a solid number into a false alarm.

What is a good NRR for SaaS?

Good NRR is your row in the benchmark grid, not the public median. Use the table in Section 3 to find the cell matching your ACV tier and GTM motion, then compare against the median and top-quartile for that specific cell.

Does net revenue retention include new customers?

No. NRR includes expansion from existing customers (upsells, cross-sells, seat growth within accounts already on your books) and excludes net-new logos entirely. A company can post 120% NRR with zero new customer acquisition if its existing base expands fast enough. That is the line between NRR and total revenue growth.

What does 120% net revenue retention mean?

It means the existing customer base grew revenue 20% net of all churn and contraction, with zero contribution from new logos. As illustrated by SaaStr's historical named-company data (Snowflake 169%, Datadog 146%, CrowdStrike 124% at IPO, all IPO-era figures, not current GAAP), 120%+ is typical of mature enterprise SaaS with a land-and-expand motion, large ACVs, and multiple expansion levers. HubSpot and Squarespace both reported below 100% NRR in the same SaaStr historical dataset; the number reflects pricing model and customer segment, not just retention execution.

What is a good GRR and NRR?

A wide GRR-NRR spread signals a healthy expansion engine. A narrow spread with high NRR should prompt scrutiny: a few large upsells may be covering moderate churn. For SMB/PLG, GRR above 85-90% paired with NRR at 100-105% is a reasonable healthy range. For enterprise SLG, GRR above 90% paired with NRR at 115%+ is more typical.

For early warning signals that a customer is about to contract or churn before GRR deteriorates, the churn warning signs playbook covers the leading indicators worth watching.

The Bottom Line: Find Your Row, Then Watch the Trend

Ignore the floating public-SaaS median. Find your stage, ACV, and motion row in the grid in Section 3. For a $400K ARR PLG product, that row is the SMB band at roughly 97% median, where 100-105% is solid performance, not a benchmark miss. Pair your NRR with the leading expansion indicators (upgrade rate, seat-expansion rate, feature-adoption-to-upgrade) so you are not waiting two-plus quarters for the lagging metric to move. If your pricing is usage-based, state clearly which NRR view you are presenting to your board.

Two diagnostics built into this article: the segmented grid with per-cell sourcing answers “is my NRR normal,” and the NRR-GRR spread reading answers “is my NRR fragile.” Both together give a VP a picture they can actually work from.

If you want the benchmark refreshes and SaaS metrics playbooks as they publish, subscribe to the SaasFlywheel newsletter for the metrics cluster as new data lands.