Last updated: September 17, 2026
Net revenue retention is the one number that tells you whether your business would still grow if you stopped signing new customers tomorrow. It bundles churn, downgrades and expansion into a single percentage, which makes it the favorite metric of boards and investors, and also the easiest one to misread. This guide covers the formula, what a good NRR looks like by company type, why NRR and GRR must be read together, and the operational levers that move the number.
Table of contents
Net revenue retention (NRR), also called net dollar retention (NDR), is the percentage of recurring revenue you keep from a group of existing customers over a period, usually twelve months, after adding expansion and subtracting contraction and churn. Take every customer you had a year ago, add up what they pay you today, and divide by what they paid you back then. New customers acquired during the year are excluded on purpose.
That exclusion is what makes NRR useful. Top-line ARR can grow for years while the installed base quietly shrinks, because new logos paper over the leak. NRR strips out acquisition and asks a narrower question: are the customers you already won becoming more or less valuable?
Three forces determine the answer. Expansion revenue (upsells, cross-sells, added seats, higher usage) pushes NRR up. Contraction (downgrades, seat reductions, discounts at renewal) pulls it down. Churn (full cancellations) pulls it down hardest. An NRR above 100% means expansion outweighs both losses combined, so the base compounds on its own. Below 100%, growth depends entirely on acquisition filling the gap.
There are two equivalent ways to write the formula. The cohort version is the mental model; the movements version is what billing and analytics tools compute.
Cohort version: NRR = (MRR today from customers who were paying a year ago) / (MRR from those same customers a year ago) x 100
Movements version: NRR = (Starting MRR + Expansion MRR + Reactivation MRR - Contraction MRR - Churned MRR) / Starting MRR x 100
Suppose you start a year with $400,000 in MRR from existing customers. Over the year those customers add $60,000 in upgrades, downgrade by $15,000, and $35,000 of MRR cancels entirely.
($400,000 + $60,000 - $15,000 - $35,000) / $400,000 = $410,000 / $400,000 = 102.5% NRR
The same inputs give a gross revenue retention (GRR) of ($400,000 - $15,000 - $35,000) / $400,000 = 87.5%. The 15-point spread between the two numbers is your expansion engine, and as the next sections show, the size of that spread is as informative as either number alone.
Key takeaway: Keep every input in the same time window. Mixing monthly expansion figures with annual churn figures is the most common reason two teams in the same company report different NRRs for the same quarter. Pick MRR or ARR, twelve months or a quarter, and hold it constant.
There is no single benchmark, because the population you compare against changes the median dramatically. Three recent datasets illustrate the spread:
| Dataset | Population | Median NRR | Top performers |
|---|---|---|---|
| SaaS Capital (2026) | Bootstrapped B2B SaaS, $3M-$20M ARR | 103% | 117.9% (90th percentile) |
| KBCM survey, via Gainsight | Private B2B SaaS | ~101% | Public SaaS median ~111% |
| ChartMogul (Dec 2025) | 3,500 software companies, B2B SaaS segment | 82% | 97% (top quartile) |
Why does ChartMogul's median sit twenty points below the others? Its dataset skews toward smaller, self-serve and lower-ACV businesses, where account churn is structurally higher and per-account expansion is small. SaaS Capital and KBCM survey companies that have already reached meaningful scale. Neither is wrong; they describe different populations, which is exactly why you should benchmark against companies with a similar contract size and sales motion, not against a headline number.
Gainsight's benchmark guidance offers useful bands: below 90% is urgent, 90-100% means the base is contracting and acquisition is doing all the work, 100-110% is solid, and above 120% is exceptional and mostly seen in enterprise SaaS where seat and module expansion is built into how customers grow. For SMB-focused products, an NRR just under 100% is not automatically a crisis; GRR or logo retention may be the more honest KPI.
The reason investors care so much is captured in one ChartMogul finding: low-retention companies are three times as likely to be shrinking as to be growing quickly. NRR is a leading indicator of whether growth is durable or borrowed.
Gross revenue retention is NRR with expansion removed. It can never exceed 100%, and it isolates how much of your base survives on stickiness alone. The two metrics answer different questions, and reading NRR without GRR is how healthy-looking companies get surprised.
Consider a company with $1M in starting MRR. Five large accounts expand by a combined $300K while fifty small accounts churn out $200K. NRR lands at 110%, which looks excellent. GRR is 80%, which is alarming. The headline is being carried by a handful of accounts; if two of them pause expansion, NRR collapses. Gainsight flags an NRR-GRR gap above 30 points as concentration risk rather than retention strength, and notes that seat growth across SaaS has slowed to roughly 2.2%, so the expansion tailwind that inflated NRR for years is weaker than it was.
Practical rule: report both in every board deck, and track what share of expansion comes from your top 10% of accounts. If it is more than half, your retention story is really a concentration story.
NRR is a lagging indicator. By the time it drops, the decisions that caused the drop were made months earlier: an onboarding that never reached first value, a champion who left, a renewal that came as a surprise. Improving it means acting on the leading indicators upstream. The levers below are ordered roughly by how early in the customer lifecycle they act.
Much of a cohort's churn is decided early in the relationship, when a customer either reaches the outcome they bought or drifts. Structured onboarding with milestone tracking, and a clean sales-to-customer-success handoff that preserves the promises made during the sale, do more for twelve-month NRR than any late-stage save play. We covered the mechanics in our guide to cutting time-to-value in B2B onboarding.
A health score that only reflects the last CSAT survey tells you about the past. One that combines login frequency, feature depth, support ticket sentiment and stakeholder changes can flag churn risk weeks before renewal. The same signals, inverted, tell you which accounts are ready for an expansion conversation: license utilization above a threshold, adoption of an adjacent module, a new team requesting access.
NRR above 100% requires expansion, and expansion that depends on individual account managers noticing opportunities does not scale. Usage-triggered plays, structured quarterly business reviews that surface growth, and a documented account expansion motion turn upsell from a lucky quarter into a repeatable line item.
This is where conversational AI is changing the economics of post-sales. An AI customer success agent like Darwin AI's Sophia runs onboarding check-ins, adoption nudges and renewal outreach over WhatsApp, email and voice for the long tail of accounts that a human CSM team cannot touch proactively, and escalates only the conversations that need a person. For mid-market portfolios where each CSM carries hundreds of accounts, that coverage gap is usually where GRR leaks.
Surprise churn is usually a renewal nobody worked. A renewal automation playbook that triggers value recaps, stakeholder mapping and pricing conversations well before the date converts a percentage of would-be cancellations into flat renewals, and flat renewals into expansions.
Expired cards, failed payments and unanswered invoices cause contraction that has nothing to do with product value. Automated dunning sequences that reach the right billing contact on the right channel recover revenue that would otherwise register as churn in your NRR.
Losing a $200K account with expansion potential and losing a $2K account both count as one logo, but they are not the same NRR event. Tier your book by revenue and potential, give high-value tiers human coverage, and give the long tail automated, digital-first coverage that still feels personal. Community, in-app guidance and adoption programs carry the low-touch tier without proportional headcount.
Example: A mid-market SaaS company with 600 accounts and 4 CSMs discovers that its 110% NRR hides an 84% GRR, with most expansion coming from 30 enterprise accounts. It moves the 450 smallest accounts to an automated success motion with usage-triggered outreach, redeploys CSM time to the top 150, and starts renewals at 120 days. Over the following year the NRR-GRR gap narrows from 26 points to 14, not because expansion fell, but because small-account churn did.
Comparing to the wrong benchmark. An SMB self-serve product at 95% NRR may be outperforming its peer group even though it trails the enterprise medians in the table above. Match ACV and motion before judging.
Reporting NRR without GRR. Covered above: the spread between them is where concentration risk hides.
Letting definitions drift. Whether reactivations count, whether multi-year prepaid contracts are annualized, and whether price increases count as expansion all change the number. Write the definition down and hold it constant across quarters.
Treating NRR as a CS-only metric. Sales sets the expectations that drive early churn; product determines whether there is anything to expand into; finance controls the billing failures that cause involuntary loss. NRR moves when all of them see their contribution to it.
Gaming the denominator. Excluding "strategic" churn or small accounts from the cohort improves the number and destroys its usefulness. Report the whole base.
It depends on your segment. Among private B2B SaaS companies, medians cluster around 101-103% according to SaaS Capital and the KBCM survey, while broader datasets including smaller self-serve companies show a median of 82%. Above 100% means your base grows without new customers; above 120% is exceptional.
Gross revenue retention excludes expansion and is capped at 100%; it measures how much revenue survives churn and downgrades alone. Net revenue retention adds expansion back in and can exceed 100%. NRR is always greater than or equal to GRR, and a wide gap between them signals that a few expanding accounts may be masking broader churn.
Yes. Net dollar retention (NDR), net revenue retention (NRR) and net MRR retention are different names for the same calculation.
Most companies report it on a trailing twelve-month basis, which smooths seasonality and captures a full adoption-and-expansion cycle. Monthly or quarterly views are useful for trend-spotting as long as the same interval is used consistently.
An NRR far above peers is worth celebrating, but check GRR and expansion concentration. If most of the expansion comes from a small share of accounts, the number is fragile and should be presented alongside the underlying distribution.
Cover every account proactively, not just the top tier
Sophia, Darwin AI's customer success agent, runs onboarding, adoption and renewal conversations across your entire base so churn gets caught before it reaches your NRR.
Meet Sophia