Net revenue retention (NRR) answers a simple question: of the recurring revenue you had from a cohort of customers a year ago, how much do you still have from those same customers today after churn, contraction, expansion and price changes? It is the cleanest single number for whether customer success is creating growth or just slowing losses. This guide walks through the formula with a worked example, compares NRR to gross revenue retention (GRR), cites current private SaaS benchmarks from SaaS Capital's primary research, and maps the CS levers that move the number.
What Net Revenue Retention Measures
NRR is a cohort revenue metric. You start with the recurring revenue from customers who were active at the beginning of the period, then look at what that same group produces at the end of the period. New logos acquired during the period are excluded. Upsells, cross-sells, usage growth and price increases are included. Logo churn and downgrades reduce the result.
Because expansion can more than offset losses, NRR can exceed 100%. That is the difference between "we kept most of the dollars" and "the existing base is growing on its own." Boards and finance teams care about NRR because it compounds. A book that retains and expands needs far less new-logo growth to hit the same ARR plan.
How to Calculate NRR: Formula and Worked Example
SaaS Capital's annual survey defines net retention as:
NRR = (ending MRR from the starting cohort) ÷ (starting MRR of that cohort)
In their 2025 research brief, that means December 2024 MRR only from customers who were customers in December 2023, divided by total MRR in December 2023. You can use ARR instead of MRR as long as both sides of the fraction use the same unit and the same cohort.
Worked example. At the start of the year your cohort has $1,000,000 in ARR.
- $80,000 churns completely (logo loss).
- $40,000 contracts through seat or tier downgrades.
- $150,000 expands through upsells, cross-sells and usage.
- $20,000 comes from price increases on renewing accounts.
Ending ARR from that same cohort:
$1,000,000 − $80,000 − $40,000 + $150,000 + $20,000 = $1,050,000
NRR = $1,050,000 ÷ $1,000,000 = 105%
Two calculation habits keep the number honest:
- Hold the cohort fixed. Do not add customers who arrived mid-period into the denominator or numerator.
- Use recurring revenue, not bookings theater. One-time professional services and non-recurring fees usually stay out of NRR unless your finance team has a documented policy that includes them.
NRR vs GRR: What Each Number Tells You
Gross revenue retention uses the same cohort and the same starting revenue, but strips out expansion and price increases. For each customer, ending revenue for GRR is capped at their starting revenue. That is why GRR cannot exceed 100%.
In the example above, GRR ignores the $150,000 of expansion and the $20,000 of price increases:
Ending GRR ARR = $1,000,000 − $80,000 − $40,000 = $880,000
GRR = $880,000 ÷ $1,000,000 = 88%
Read the two metrics together:
| Metric | Includes | Ceiling | What a weak number usually means |
|---|---|---|---|
| GRR | Starting revenue minus churn and contraction | 100% | Retention and packaging problems; expansion cannot paper over the leak |
| NRR | GRR base plus expansion and price changes | None | Either the base is leaking, expansion is thin, or both |
A team with 95% GRR and 112% NRR has a strong expansion engine on a solid base. A team with 82% GRR and 101% NRR is running hard on upsell just to stand still. Report both. NRR alone can hide a churn problem.
What Is a Good NRR? SaaS Capital 2025 Benchmarks
Public-company NRR figures are a poor peer set for most private SaaS teams. SaaS Capital's Research Brief 32 (2025 B2B SaaS Retention Benchmarks) is a primary source based on its 14th annual survey of more than 1,000 private B2B SaaS companies. The brief excludes companies under $1M ARR for these cuts.
Headline numbers from that brief:
- Median NRR is 101% and median GRR is 91% across the sample.
- Companies with ACV above $250,000 show median GRR of 95%; companies below that band show roughly 91%.
- Median NRR rises with ACV: 98% under $12k, 103% at $12k–$25k, 102% at $25k–$50k, 104% at $50k–$100k, 102% at $100k–$250k, and 106% above $250k.
- For the $25k–$50k ACV band, SaaS Capital's accompanying analysis reports a median NRR of 102%, with a top quartile of 111% and a bottom quartile of 97%.
- SaaS Capital treats GRR of at least 90% as table stakes for peer parity.
- By contract length: multi-year contracts show median NRR of 103% and GRR of 94%; annual contracts 101% / 90%; month-to-month 100% / 89%.
- Bootstrapped companies in the brief report 104% NRR and 92% GRR; equity-backed companies report 101% NRR and 90% GRR.
NRR also correlates strongly with growth in the same survey. Population median growth was 24%. Companies below 90% NRR grew at a 15% median; the 100–110% NRR band grew at 21%; the 110–120% band grew at 30%. Moving from the 100–110% band to the 110–120% band added nine growth points in that dataset.
Use ACV-matched peers, not a single "good NRR is 120%" slogan. A mid-market seat product at 102% NRR can be top-quartile for its band while a high-ACV platform at 102% is lagging.
The Three CS Levers That Move NRR
Every point of NRR comes from one of three places: less churn, less contraction, or more expansion. Customer success owns signals and plays on all three.
1. Churn (logo and full revenue loss)
Churn is the largest downward force on both GRR and NRR. Watchable signals before the renewal conversation:
- Activation or adoption stalls after go-live.
- Champion or executive sponsor changes.
- Support volume rising while product usage falls.
- QBR attendance thinning or success criteria going unmet.
- Engagement gaps in the 90/60/30-day renewal window.
The CS response is a named save motion: confirm the risk reason, attach evidence, assign an owner and run a time-boxed plan. Guessing from an exit survey alone is too late.
2. Contraction (downgrades and seat loss)
Contraction often looks like "they renewed" in a logo report while dollars quietly leave. Signals:
- Seat utilization falling for several months.
- A business unit stopping use while the parent contract continues.
- Feature adoption concentrated in one team with the rest idle.
- Customers asking to drop modules or environments at renewal.
Treat contraction as a product-fit and packaging problem, not only a discount problem. If a segment routinely buys more than it can adopt, fix onboarding and packaging before the next cohort renews.
3. Expansion (upsell, cross-sell, usage, price)
Expansion is how NRR clears 100%. The accounts that expand usually show the opposite of churn signals: rising usage, seat pressure, new workspaces or environments, feature ceilings, API thresholds and buying activity from champions. Quivly's expand-NRR framing tracks product usage milestones, seat utilization, feature adoption gaps and engagement trends so those accounts surface before a CSM has to hunt for them. For a deeper signal checklist, see expansion signals. For why healthy accounts still fail to expand, see why existing customers stall expansion.
How to Improve NRR Without Guessing
Improving NRR is an operating system, not a quarterly scramble.
- Split the metric. Report GRR and NRR monthly. If GRR is below ~90% for your ACV band, fix retention before pouring energy into expansion theater.
- Segment by ACV and motion. Self-serve, mid-market and enterprise will not share the same churn or expansion patterns. Benchmark each against its ACV peer set.
- Attach a play to each signal class. Rescue plays for churn risk, packaging or adoption plays for contraction, and expansion plays for usage and buying signals. A play is a named motion with a trigger, steps and a channel, so the same motion runs when the same signal hits.
- Time the work to the renewal calendar. Most preventable churn and missed expansion show up inside the last 90 days. Put 90/60/30 reviews on the calendar for strategic accounts and automate nudges for the rest.
- Close the loop with evidence. Every save or expansion attempt should leave a record of the signal, the action and the outcome so the next quarter's NRR review is diagnostic, not anecdotal.
In Quivly, expansion and rescue work is designed around that loop. Expand NRR surfaces accounts crossing usage and buying thresholds, and Plays let you write the expansion or rescue motion once and run it when the signal hits, with drafts held for review unless you allow an agent to send. Use that only as the operating layer after the math and the peer benchmarks are clear.
NRR improves when the cohort you already won keeps paying and buys more. Calculate it the same way every month, pair it with GRR, benchmark against ACV peers from primary data, and put a watchable signal and a named play on churn, contraction and expansion. That is the whole job.
