Net revenue retention (NRR) is the percentage of recurring revenue you keep from an existing group of customers over a period, usually twelve months, after adding what they expanded and subtracting what they downgraded or cancelled. Revenue from customers you won during the period is left out. The formula is (starting ARR + expansion − contraction − churn) ÷ starting ARR.
Above 100% means the customers you already had are paying you more than they were a year ago, even after the ones who left. Below 100% means the base is shrinking and new sales are refilling it. Net dollar retention (NDR) is the same metric under another name.
The NRR formula
Pick a cohort: every customer paying you on the first day of the period. Freeze that list. Then, at the end of the period, measure what those customers alone are paying.
NRR = (starting ARR + expansion − contraction − churn) ÷ starting ARR × 100
| Component | In NRR | In GRR | What it is |
|---|---|---|---|
| Starting ARR | Yes | Yes | Recurring revenue from the cohort on day one of the period |
| Expansion | Added | Excluded | Upgrades, added seats, cross-sells and price increases to those same customers |
| Contraction | Subtracted | Subtracted | Downgrades, removed seats and discounts that lower what an existing customer pays |
| Churn | Subtracted | Subtracted | Recurring revenue from customers who cancelled or did not renew |
| New customers | Excluded | Excluded | Anyone not in the cohort on day one, however large |
Leaving out new customers is what makes the metric useful. NRR isolates the revenue your existing base generates on its own, so it cannot be flattered by a strong sales quarter. A customer who signs in month three is not in this year's cohort, even if they expand twice before the year is out. They will be in next year's.
A worked example
A company starts the year with $2,000,000 of ARR across its customers. Over twelve months those same customers add $300,000 through upgrades and extra seats, cut $80,000 through downgrades, and cancel contracts worth $140,000.
NRR = ($2,000,000 + $300,000 − $80,000 − $140,000) ÷ $2,000,000 = $2,080,000 ÷ $2,000,000 = 104%.
The base grew by 4% with no new customers at all. The same cohort's gross revenue retention, which ignores the $300,000 of expansion, is ($2,000,000 − $80,000 − $140,000) ÷ $2,000,000 = 89%. Both are true at once, and the 15-point gap between them is how much of this year's result depended on expansion.
NRR and GRR calculator
Enter one cohort's numbers for the same twelve months, in any currency and in the same unit throughout. Leave out revenue from customers won during the period. Nothing leaves your browser.
NRR, NDR and net retention: one metric, several names
Net dollar retention (NDR), net retention rate and net revenue retention all describe the calculation above. Public SaaS filings tend to say "dollar-based net retention"; finance teams and investors tend to say NRR. If two documents give different numbers for what looks like the same metric, the difference is almost always in the inputs, not the name.
Three input choices change the number, so state them next to it:
ARR or MRR
ARR is the usual base. An MRR-based figure gives the same answer if you measure the same twelve months, but a company that mixes monthly and annual contracts should convert everything to one before calculating.
Annual or monthly
Monthly NRR is not annual NRR divided by twelve. Retention compounds, so 99% a month is about 88.6% a year (0.99 raised to the twelfth power), not 88%. Report the annual figure, or say clearly that yours is monthly.
One cohort or a trailing average
The clean version measures one cohort across twelve months. Many teams report a trailing-twelve-month figure that blends several cohorts. Both are legitimate. Comparing one to the other is not.
Mistakes that inflate NRR
Letting new customers into the cohort
The most common error, and the hardest to spot in a dashboard. If the end-of-period figure includes customers who signed during the period, NRR stops measuring retention and starts measuring sales. Freeze the customer list on day one.
Counting a returning customer as expansion
A customer who churned and then came back is a new customer in the period they returned. Booking their revenue as expansion on the original account makes churn look smaller and expansion look larger in the same move.
Treating a price rise as product growth
A list-price increase applied across the base is real revenue and belongs in expansion. It is also a one-off. Report how much of the year's expansion came from price, so next year's comparison is honest.
Netting churn against expansion inside one account
If a customer drops one product and adds another, record both movements. Netting them hides the contraction, and that hidden contraction is exactly what gross revenue retention exists to catch.
NRR benchmarks
Across private B2B SaaS, the typical company keeps its base roughly flat. SaaS Capital's 2025 survey put median NRR at 101% across all respondents. High Alpha's 2025 report found medians between 100% and 104% in every ARR band, with the middle half of companies spread far wider than the medians suggest.
| ARR band | Median NRR | Middle 50% of companies |
|---|---|---|
| Under $1M | 100% | 78% to 116% |
| $1M to $5M | 104% | 91% to 110% |
| $5M to $20M | 103% | 95% to 115% |
| $20M to $50M | 103% | 98% to 110% |
| Over $50M | 101% | 97% to 108% |
Contract size moves the number more than company size does. Cut by average contract value instead of ARR, SaaS Capital's medians run from 98% for contracts under $12K a year to 106% above $250K (companies under $1M ARR excluded). If you sell small self-serve plans, compare yourself with the first figure, not the second.
Sources: SaaS Capital, 2025 B2B SaaS Retention Benchmarks, more than 1,000 private B2B SaaS companies surveyed in Q1 2025, pp.1–2; High Alpha, 2025 SaaS Benchmarks Report, 800+ respondents, metrics as of Q2 2025, p.11. Survey medians describe the companies that answered, so treat them as a reference point rather than a target.
What NRR does and does not tell you
NRR is a measure of the existing base, and it is a blunt one. A cohort can reach 110% because most customers expanded a little, or because two large accounts expanded a lot while a third of the rest cancelled. The headline is identical. Gross revenue retention, logo retention and the spread of expansion across accounts tell those two businesses apart.
Why the number matters commercially, from pricing to how investors value it, is a separate question. It is covered in The NRR Advantage. The economics of keeping a customer versus replacing one are in retention economics.
Frequently asked questions
What does NRR mean?
NRR stands for net revenue retention: the percentage of recurring revenue a company keeps from its existing customers over a period, usually a year, after adding expansion and subtracting downgrades and cancellations. Revenue from customers won during the period is excluded, so the metric shows what the existing base does on its own.
What is the NRR formula?
NRR = (starting ARR + expansion − contraction − churn) ÷ starting ARR × 100. Starting ARR is the recurring revenue from customers active on the first day of the period. Expansion, contraction and churn are measured for those same customers only.
Is net dollar retention the same as net revenue retention?
Yes. Net dollar retention (NDR), dollar-based net retention, net retention rate and net revenue retention are names for the same calculation. When two reported figures differ, check the inputs: ARR or MRR, monthly or annual, a single cohort or a trailing-twelve-month blend.
Can NRR be over 100%?
Yes. NRR above 100% means the expansion from existing customers was larger than everything lost to downgrades and cancellations, so the base grew without any new customers. Gross revenue retention, which leaves expansion out, can never exceed 100%.
Does NRR include new customers?
No. Customers acquired during the measurement period are excluded, even if they expand before it ends. Including them turns NRR into a blend of retention and new sales, which defeats the purpose of the metric.
How do I convert monthly NRR to annual NRR?
Raise the monthly rate to the twelfth power rather than multiplying the monthly loss by twelve. A monthly NRR of 99% is 0.9912, about 88.6% a year. Monthly figures are useful for spotting changes early, but annual NRR is the convention most people compare against.