Net Revenue Retention: The Growth Number That Ignores New Customers
Net revenue retention, or NRR, is the recurring revenue an existing cohort of customers produces at the end of a period as a percentage of what the same cohort produced at the start. It counts upgrades, downgrades and departures, and excludes anything from customers won during the period.
Key takeaways
- Above 100 percent means the existing base grew on its own, which changes what acquisition is worth and what the sales target needs to be.
- Expansion can hide heavy logo loss, so NRR should always be read beside gross revenue retention and logo retention.
- Usage-based pricing inflates it mechanically when customers grow, and deflates it in a downturn without anyone being unhappy.
- Cut by segment it is the strongest evidence you own about who was worth acquiring, which win rates cannot tell you.
Net Revenue Retention: The Growth Number That Ignores New Customers
Net revenue retention, usually written NRR, is the recurring revenue a cohort of existing customers produces at the end of a period, expressed as a percentage of what the same cohort produced at the start, counting upgrades, downgrades and departures but excluding anything from customers won during the period. An NRR of 100 percent means the customers you already had are worth exactly what they were worth a year ago. Above 100 means the base grew on its own. Below means it shrank.
The exclusion is the whole design. Take out new business and what remains is a direct measurement of whether the product keeps and grows the customers it has.
How it is calculated
Fix a cohort: every customer active on the first day of the period. Take their recurring revenue on that day. Take the recurring revenue from those same customers, and only those customers, on the last day. Divide the second by the first.
Suppose the cohort started the year at 1,000,000 in annual recurring revenue. Over twelve months, departures removed 90,000, downgrades removed another 40,000, and upgrades and seat growth among survivors added 210,000. The cohort ends at 1,080,000, and NRR is 108 percent. Any revenue from customers signed during those twelve months is outside the calculation entirely, however large it was.
Recurring revenue from every customer active on day one
Customers from the cohort who left
Survivors who reduced spend
Survivors who added seats, usage or products
NRR of 108 percent. New customers are not in either figure.
Bar widths are equal here because these stage values are not a single comparable measure.
Two variants show up beside it. Gross revenue retention applies the same method without the expansion term, so it caps at 100 percent and measures pure leakage. Logo retention counts customers rather than revenue. NRR is the only one of the three that can exceed 100, which is precisely why it is the one quoted.
Why it carries so much weight
A business with NRR above 100 percent grows without selling anything new. That is a genuinely unusual property, and it changes what every other number means.
It changes the value of acquisition. A customer worth 12,000 in year one and 15,000 in year three justifies a much larger acquisition cost than one worth 12,000 declining to 9,000, which flows directly into what a defensible target cost per acquisition is.
It changes the sales target. If the existing base grows 8 percent on its own, an overall growth target of 30 percent requires new business worth 22 percent, not 30. Teams that plan the sales number without netting off expansion consistently over-hire.
And it separates two businesses that look identical from the outside. Two companies both growing 40 percent, one at 120 percent NRR and one at 85 percent, are running at completely different levels of difficulty: the second has to replace what it loses before it grows at all, which is the arithmetic set out under churn rate.
Where the textbook definition breaks
Expansion can hide a retention problem entirely. A handful of large accounts doubling their usage can carry an NRR above 100 while half the customer base leaves. The headline number reports health, the logo retention figure reports the truth, and the two are frequently quoted a hundred slides apart. Always read NRR beside gross retention and logo retention.
Usage-based pricing inflates it mechanically. In a consumption model, a customer whose own business grew increases spend without any decision to expand. That is real revenue and it is not evidence that the product is retaining better. In a downturn the same mechanism runs in reverse, and NRR falls without a single customer being unhappy.
Price rises count as expansion. An annual uplift applied across the base raises NRR without any change in what customers use or value. It is legitimate to count and worth disclosing, because it is one of the few expansion levers with a hard ceiling.
The cohort definition is negotiable and rarely stated. Are customers who left during the period included in the starting base? They must be, or departures never appear at all. Are customers who signed in month two and expanded in month eight excluded? They should be. Both mistakes flatter the number, and neither is visible in the percentage.
Above 100 is not a target for every business. NRR above 100 requires a product whose usage grows with the customer's own work. A tool with a fixed footprint per company can be excellent and sit at 95 percent forever, and chasing expansion revenue there usually means adding products nobody asked for. The metric describes a model as much as a performance.
- Depends: The cohort is customers active at period start, including those who later left
- Depends: Revenue from customers won during the period is excluded
- Depends: Gross revenue retention and logo retention are reported alongside
- Depends: Price increases are disclosed separately from usage-driven expansion
- Depends: In a usage-based model, the effect of customers' own growth is called out
- Depends: The period is stated and is the same one used in prior reporting
What actually moves it
NRR has exactly four inputs, and each one belongs to a different part of the business, which is why a single company-wide target for it rarely produces action.
Product breadth. A customer can only expand into things you sell. A single-product company with a fixed footprint per customer has a structural ceiling on expansion, and no amount of account management moves it. Adding a second product that the same buyer needs is the largest single lever available, and it is a product decision rather than a commercial one.
Pricing structure. Whether growth in the customer's own usage is captured automatically or has to be renegotiated. Per-seat and consumption pricing convert customer growth into revenue growth without a conversation. Flat pricing does not, and every expansion becomes a sale.
Onboarding depth. How embedded the product becomes in the first ninety days, which predicts both departures and expansion better than almost anything measured later. Customers who reached a real outcome early expand; customers who never quite finished setting it up leave at renewal, and the two are visible in the same cohort table.
Who you sold to. The input nobody controls after the fact and everybody controls before it. A customer whose own business is shrinking cannot expand, however good your account management, and a customer in a growing segment expands without being asked.
That last input is the one worth pulling on, because it is the only one that is decided upstream. By the time a customer is in the base, three of the four levers are the ones you already built. The fourth was chosen when somebody decided who to contact.
Two practical readings follow. A weak NRR in a business with good onboarding and sensible pricing is usually a targeting result, not a customer success failure, and the standard response of adding account management effort will not fix it. And NRR is the slowest metric to respond to a change in targeting, because a decision to sell to a different segment today only shows up in the retention figure a year later, which is a reason to watch the cohort view rather than the blended number.
What it tells you about who to sell to
The connection to targeting is the most useful thing an outbound team can take from this metric, and it is almost never used.
NRR is computable by segment, and the segment view is a direct statement about which customers were worth acquiring. A segment retaining at 130 percent and one retaining at 80 are not variations on a theme, they are different businesses inside your business. The first justifies a far higher acquisition cost, a longer sales cycle and more expensive channels. The second may not be worth acquiring at all, whatever its win rate looks like.
That comparison corrects an ideal customer profile using evidence no sales-side data contains. Win rates tell you who says yes. Pipeline tells you who engages. Only the retention view tells you who was still worth having two years later, and when the two disagree the retention view is the one to trust, because it observed more of the relationship.
The practical version is a single query: NRR by industry, by size band, and by acquisition channel, for cohorts old enough to have renewed. Most teams have never run it, and it is usually the most decision-changing hour available to a go-to-market team that already has customers. Then the arithmetic feeds back into what a customer is worth, which is where cost per customer acquisition becomes a budget rather than a report.
One caution about running it that way. Cohorts have to be old enough to have been through a renewal, which means the segment view is always describing decisions taken a year or more ago. That lag is the price of the metric being about the whole relationship rather than about the sale, and it is a reason to read the direction of travel across several cohorts rather than to act decisively on a single one.
Related terms
Churn rate is the loss term without the expansion. Annual recurring revenue is the base it measures. And product-led growth is the model most dependent on getting it above 100.
The short version
Net revenue retention measures what your existing customers are worth now against what they were worth then, with new business excluded. Read it beside gross and logo retention, watch for price rises and usage effects, and then do the thing almost nobody does with it: cut it by segment and let it tell you who to go after next.
Once the segment worth more has a name, building the list of companies that match it is the next step: see what one campaign against it produces.
Frequently asked questions.
Frequently asked questions- How is net revenue retention calculated?
- Fix a cohort of every customer active on the first day of the period. Take their recurring revenue on that day, then take the recurring revenue from those same customers on the last day, and divide the second by the first. Revenue from customers signed during the period is excluded from both figures.
- What is the difference between net and gross revenue retention?
- Gross revenue retention applies the same method without counting expansion, so it caps at 100 percent and measures pure leakage from departures and downgrades. Net includes expansion and can exceed 100. Reading them together is what stops a few large expansions concealing a wide base of departures.
- Should every company target NRR above 100 percent?
- No. Exceeding 100 requires a product whose usage grows with the customer's own work. A tool with a fixed footprint per company can be excellent and sit at 95 percent indefinitely, and chasing expansion revenue there usually means adding products nobody asked for. The metric describes a model as much as a performance.
- What does NRR tell you about who to sell to?
- Cut by segment it says which customers were worth acquiring. A segment retaining at 130 percent justifies a far higher acquisition cost, a longer sales cycle and more expensive channels than one retaining at 80. Win rates say who buys; only retention says who was still worth having two years later.