Lester Leong
Net Revenue Retention: The SaaS Metric That Predicts Whether You Have a Business
The One Number That Tells You If You Have a Business
Most SaaS metrics describe a slice of the company. Net revenue retention describes the whole thing. It answers a single brutal question: if you stopped acquiring new customers tomorrow, would your revenue grow, hold, or bleed out? Every other growth metric assumes you keep filling the top of the funnel. NRR is the only one that turns the funnel off and shows you what is actually underneath.
Net revenue retention (NRR) measures how the revenue from a fixed cohort of existing customers changes over a period, accounting for expansion, contraction, and churn, while deliberately excluding every new logo you signed. If NRR is above 100%, your existing base grows on its own. If it is below 100%, you are running an acquisition treadmill: you have to sell new customers just to replace the revenue leaking out the back, and the moment acquisition slows, the business shrinks.
I have watched NRR sort the durable businesses from the fragile ones in three different environments. As a consultant working with 20+ SMBs and startups through Gradient Growth, where I have seen companies with impressive top-line growth discover that their NRR was 84% and their entire growth story was paid acquisition masking a leaking base. At a financial social media startup before its acquisition, where revenue retention was one of the first numbers any serious acquirer asked to see. And on a GenAI squad at a major finance technology company, where the data volume lets us decompose retention by cohort and segment and watch exactly which customers expand and which quietly contract. Across all three, NRR survived scrutiny when growth-rate vanity metrics did not.
This is not a metric you bolt onto a dashboard for completeness. It is the metric investors weight above almost everything else, because it is the hardest one to fake and the most predictive of long-term value.
What NRR Actually Measures
The cohort-based definition is the only one that means anything:
``` NRR = (Starting MRR + Expansion - Contraction - Churn) / Starting MRR * 100 ```
Every term in that formula is measured on the customers you already had at the start of the period. New logos do not appear anywhere. That exclusion is the entire point. NRR isolates the behavior of your existing base so that growth from new sales cannot paper over decay in your installed customers.
The four components, defined precisely:
1. Starting MRR is the monthly recurring revenue from the cohort of customers active at the beginning of the period. This is your denominator and your baseline. 2. Expansion is additional recurring revenue from that same cohort: upsells, seat expansion, tier upgrades, usage growth on consumption pricing. 3. Contraction is recurring revenue lost from customers who stayed but shrank: downgrades, seat reductions, dropping to a cheaper plan. 4. Churn is recurring revenue lost from customers who left entirely.
The number can exceed 100% because expansion has no ceiling. A cohort that expands more than it contracts and churns nets out above its starting point. That is the structural difference between NRR and the metric it is most often confused with.
NRR Versus GRR: The Distinction That Matters
Gross revenue retention (GRR) is the same cohort measured without the expansion term:
``` GRR = (Starting MRR - Contraction - Churn) / Starting MRR * 100 ```
GRR caps at 100% by definition. It measures pure leakage: how much of your existing revenue you keep before any upsell. It cannot be rescued by a few large expansions, which is exactly why it is the more honest measure of product stickiness.
The gap between NRR and GRR is one of the most diagnostic spreads in SaaS. If NRR is 118% and GRR is 88%, you have a powerful expansion engine sitting on top of a leaky base. The expansion is real, but it is masking a churn problem that will eventually outrun it as the base grows. If NRR is 112% and GRR is 104%, you have a healthy, sticky product with modest, durable expansion. Same NRR, completely different business. I always report both. A team that only watches NRR can run for a year convinced it is healthy while GRR quietly erodes underneath the expansion that hides it. For the mechanics of decomposing the leakage side, [how to calculate churn rate](/insights/how-to-calculate-churn-rate) walks through the churn term in detail.
A Fully Worked Example
Abstract formulas hide the intuition. Here is a concrete cohort.
You start the period with 100 customers paying a combined 120,000 dollars in MRR. That is your starting MRR. Over the next 12 months, holding to that original cohort and ignoring every new customer you signed:
- Expansion: 22 customers upgraded plans or added seats, adding 18,000 dollars in new recurring revenue. - Contraction: 14 customers downgraded or cut seats, losing 6,000 dollars in recurring revenue. - Churn: 9 customers canceled entirely, taking 12,000 dollars in recurring revenue with them.
Run the numbers:
``` NRR = (120,000 + 18,000 - 6,000 - 12,000) / 120,000 * 100 = 120,000 / 120,000 * 100 = 100.0%
GRR = (120,000 - 6,000 - 12,000) / 120,000 * 100 = 102,000 / 120,000 * 100 = 85.0% ```
This cohort sits at exactly 100% NRR and 85% GRR. Read those two numbers together and the story is clear. The base is not growing on its own, it is treading water, and the only reason it is not shrinking is that 18,000 dollars of expansion exactly offset 18,000 dollars of contraction plus churn. Strip out the expansion engine and the underlying product is losing 15% of its revenue a year. That is a churn problem dressed up as stability. The fix is not to celebrate the 100% NRR. It is to attack the 15-point GRR leak before the expansion engine can no longer keep pace with it.
Now change one input. Suppose expansion had been 33,000 dollars instead of 18,000, because the product had a genuine usage-based growth loop. NRR jumps to 112.5% while GRR stays at 85%. The headline improves, the leak does not. This is precisely why a single number deceives and the pair informs.
Benchmarks, Honestly Stated
NRR benchmarks vary enormously by segment, and blending them produces a meaningless company-wide average. Stated by segment:
1. SMB and self-serve PLG: Around 100% is a reasonable target, and many healthy businesses sit slightly below it. Small customers churn at structurally higher rates and expand less, because there is less room to upsell a 5-person team. An SMB product at 100% NRR is doing well. Holding it there is hard work. 2. Mid-market: 105% to 115% is the healthy band. There is enough seat and usage headroom for expansion to outrun churn, but accounts are still small enough to leave. 3. Enterprise: 110% is good, 120%+ is elite. The best enterprise SaaS businesses post NRR of 120% to 130%, which means an enterprise cohort that signs zero new customers still grows 20% to 30% a year on expansion alone. That is the closest thing to a compounding machine in software.
The benchmark you should care about is your own segment, measured against your own trend. A mid-market company comparing itself to enterprise NRR will torture itself over a healthy number. An SMB company benchmarking against 120% will conclude its business is broken when it is normal for its segment. Segment first, benchmark second.
Why Investors Weight NRR Above Almost Everything
Investors have learned, often expensively, that top-line growth rate is the most gameable number in SaaS. You can buy growth with paid acquisition for as long as the funding lasts. NRR cannot be bought. It is a property of the product and the customers, revealed only after the sale, and it compounds.
The logic is straightforward. A business at 120% NRR has a base that grows 20% a year with no new sales. Layer any new-logo acquisition on top and the growth is explosive and capital-efficient, because you are not spending sales and marketing dollars to replace churned revenue. A business at 85% NRR has to acquire 15% of its revenue in brand-new customers every year just to stand still, and every dollar of that acquisition carries its own [customer acquisition cost and payback period](/insights/cac-payback-period). One business is a compounding asset. The other is a bucket with a hole that you keep refilling.
NRR also feeds directly into the customer economics that determine whether the unit model works at all. A cohort with NRR above 100% has a [lifetime value that expands over time](/insights/customer-ltv-calculation-startups) rather than decaying, which is the single biggest lever on LTV and therefore on how much you can rationally spend to acquire a customer. When an investor sees 120%+ NRR, they are seeing a business where LTV is structurally underwritten by the customers themselves. That is why NRR routinely outranks growth rate, gross margin, and CAC in diligence. It is the metric that is hardest to fake and most predictive of what the business is worth in five years.
The Five Mistakes That Corrupt the Number
Most reported NRR numbers are wrong, and almost always wrong in the optimistic direction. The five failures I see repeatedly:
Mistake 1: Blending Segments
A company-wide NRR averages enterprise expansion with SMB churn into a single number that describes no actual customer. I have seen a blended 108% that decomposed into 128% enterprise and 91% SMB. The blended figure hid both the crown jewel and the leak. Always compute NRR per segment, then look at the blend last, if at all.
Mistake 2: Counting New Logos
The most common and most damaging error. Teams include revenue from customers acquired during the period in the numerator, which mechanically inflates NRR by mixing acquisition into a metric whose entire purpose is to exclude it. If a new customer can raise your NRR, you are not measuring NRR. The cohort is fixed at the start of the period, full stop.
Mistake 3: Measuring Monthly Instead of Annually
Month-over-month NRR is noisy and structurally flattering, because most expansion and most churn play out over quarters, not weeks. A monthly NRR of 101% annualizes to a very different place than a true 12-month NRR of 101%, and teams that report the monthly figure systematically overstate retention. Measure NRR on a trailing 12-month cohort. It is the period over which expansion and churn actually resolve.
Mistake 4: Ignoring the Cohort View
A single point-in-time NRR tells you where you are, not where you are heading. The teams that act on NRR track it by signup cohort, the same discipline that makes [retention curve analysis](/insights/retention-curve-analysis-guide) so much more powerful than a blended retention number. Cohorting NRR reveals whether newer customers are retaining and expanding better or worse than older ones, which is the leading indicator of where the aggregate number is going next.
Mistake 5: Confusing Logo Retention with Revenue Retention
Logo retention counts customers. Revenue retention counts dollars. They diverge violently when your churn is concentrated in your largest or smallest accounts. You can retain 95% of your logos and 80% of your revenue if the 5% who left were your biggest contracts. NRR is the dollar-weighted number, and it is the one that pays the bills. Never substitute a logo count for it.
How to Act on a Weak Number
A weak NRR is not a single problem. It is two independent problems wearing one number, and the first move is always to pull them apart. NRR below your segment benchmark is some combination of a broken expansion engine and a churn leak, and the interventions for each are completely different. Diagnosing one while fixing the other wastes a quarter.
Separate the engine from the leak. Compute GRR alongside NRR. GRR isolates the leak, the contraction and churn, with no expansion to hide it. The spread between NRR and GRR is your expansion engine. Now you know which one is failing.
1. If GRR is the problem (say, below 90% for mid-market), you have a stickiness problem, and no amount of upsell will durably fix it. Root-cause the churn: segment churned revenue by customer size, plan, tenure, and acquisition channel, then trace it upstream to onboarding and early engagement, because most revenue churn is set in motion in the first 30 days. The leak almost always starts before the customer ever expands. 2. If GRR is healthy but NRR is barely above it, your base is sticky but not growing, which means the expansion engine is the failure. Look at whether your pricing even allows expansion. A flat per-company price with no seat, usage, or tier dimension structurally caps NRR at 100% no matter how much customers love the product. Examine your [pricing and ARPU structure](/insights/arpu-analysis-pricing) before you blame the customer success team. You cannot expand revenue the pricing model has no mechanism to capture.
Root-cause each independently, in numbers, before touching either. The teams that move NRR resist the urge to treat it as one lever. It is two, and they often pull in opposite directions: a company can have its best expansion quarter and its worst churn quarter at once and report a flat NRR, learning nothing unless it decomposed the number first.
The Bottom Line
Net revenue retention is the metric that survives when every growth-rate story is stripped away. Above 100%, your existing customers compound and new acquisition is pure acceleration. Below 100%, you are running to stand still, and the day acquisition slows is the day the business starts shrinking. That is why investors weight it above almost everything, and why it belongs at the top of your board deck rather than buried three slides into the appendix.
Measure it on a fixed cohort. Exclude every new logo. Report GRR beside it so the expansion engine and the churn leak are never confused for each other. Segment before you benchmark, cohort before you trend, and decompose before you act. Done with that discipline, NRR will tell you, more reliably than any other single number, whether you have a business that grows on its own or one that only grows as long as the acquisition budget holds.
---
I help SaaS teams measure net revenue retention rigorously and separate the expansion engine from the churn leak so they fix the right problem. [lester@gradientgrowth.com](mailto:lester@gradientgrowth.com)