What Is a Good Net Revenue Retention Rate?
100% is the structural line, because above it your existing base grows with zero new logos. That threshold is not a benchmark someone chose. It is arithmetic. At exactly 100%, expansion from current customers cancels contraction from current customers, and the base is self-sustaining. Above it, the business compounds without acquisition. Below it, some portion of every new logo you sign is refilling a leak instead of adding growth.How far above 100% counts as good depends on what your product allows. A usage-based product with a broad upsell path can clear the line on consumption growth alone. A flat-fee single-product contract has to earn every point through cross-sell or price. Judging both against the same target rewards the pricing model rather than the retention work.
How Is Net Revenue Retention Calculated?
Beginning ARR, minus contraction from the existing base, plus expansion from that same base, divided by beginning ARR. New logos never appear anywhere in the formula. The discipline that makes the result trustworthy is a reconciling waterfall, where every dollar of movement lands on a named line rather than being backed into.| Waterfall line | Direction | What it captures |
|---|---|---|
| Beginning ARR | Starting point | Prior month's ending ARR, no exceptions |
| Churned customer ARR | Contraction | Accounts that left entirely |
| Churned product ARR | Contraction | A product dropped while the account stays |
| Product decrease ARR | Contraction | Downgrades and seat reductions |
| New product ARR | Expansion | Existing customer adds a product |
| Increased product ARR | Expansion | Existing customer grows an existing product |
| Ending ARR | Result | Feeds next month's beginning ARR |
Why Do Two Companies Report Different NRR From Similar Data?
Because at least four method choices sit behind the number, and each one moves it by points. The first is measurement period. A monthly rate annualized and an annual cohort rate answer different questions and rarely agree. The second is cohort versus snapshot. Cohort NRR follows a fixed group of customers forward. Snapshot NRR takes whoever was there at the start of the period, which quietly changes the population every time you measure.The third is netting inside accounts. Some teams net a downgrade against an upsell within the same customer before rolling up, which suppresses both contraction and expansion and flatters the stability of the base. The fourth is treatment of churned accounts in the denominator. Drop them and the rate rises for a reason that has nothing to do with performance.
None of these choices is wrong. Undocumented choices are. Write the method down, apply it every period, and treat a change to it as a restatement rather than an improvement.
What Should You Expect at Your ACV?
Expect the shape of your NRR to follow your expansion surface, not your customer satisfaction. Products with per-seat or consumption pricing expand as the customer's own business grows, which supplies a baseline of expansion that arrives without a sales motion. Products sold as a single flat-fee license expand only when someone sells something, so their NRR depends entirely on an account team's capacity.That is why the same NRR figure carries different information in different businesses. In the first model, a rate near 100% suggests the customer's usage is flat, which is an early product signal. In the second, the same rate may simply mean nobody worked the base this quarter, which is a coverage problem with a straightforward fix. Segment your NRR by pricing model before comparing it to anything.
What Breaks an NRR Number You Thought Was Stable?
Concentration. One large contraction can undo a quarter of expansion work spread across hundreds of accounts. NRR is a dollar-weighted metric, so a base with a few very large customers has a naturally volatile rate no matter how well retention is managed. A stable NRR across four quarters in that kind of book is often luck rather than control.Test for it directly. Ask what your NRR would be if your largest account contracted by 20%, then by 50%, then left. If any of those scenarios drops the number below 100%, your retention program is a concentration risk program and should be resourced accordingly. The earliest warning tends to show up in engagement rather than in commercial conversations: a customer with zero support cases in a year carries real churn risk, the same as one with seven or more, while three to five moderate tickets usually marks an engaged account.
How Do You Forecast NRR Instead of Reporting It?
Model expansion and contraction as forward lines with the same rigor you apply to new business, then hold the model to an accuracy standard. Most teams reach roughly 90% forecast accuracy on new and expansion revenue, and getting there takes heavy manual effort that goes stale the moment market conditions shift. ORM targets 95% accuracy on that scope, holds it from day 1 to day 90 of the quarter, and updates as the quarter progresses without manual adjustments.The difference matters because NRR reported after the fact cannot be managed. Knowing in week two that expansion is tracking behind and contraction is tracking ahead leaves time to redirect account teams. Knowing it in week thirteen leaves time to write commentary. Pair the forecast with a forecast accuracy measure on the retention lines specifically, since accuracy on new business tells you nothing about whether your base is behaving as modeled.
Frequently Asked Questions
What is a good net revenue retention rate?
100% is the structural line, because above it the existing base grows on its own with zero new logos. Below it, every new customer you sign is partly refilling a leak. How far above 100% counts as good depends on your ACV and expansion model: a usage-based product with a broad upsell path can clear that line comfortably, while a flat-fee single-product contract has to work much harder for every point.
How is net revenue retention calculated?
Take beginning ARR for the period, subtract contraction from the existing base, add expansion from that same base, and divide by beginning ARR. New logos never enter the calculation. ORM reads it off a monthly waterfall where beginning ARR equals the prior month's ending ARR and every dollar of movement lands on a named line, so the components reconcile rather than being estimated.
Why do two companies report different NRR from similar data?
Because the choices behind the number differ: monthly versus annual measurement, cohort-based versus snapshot, whether downgrades are netted against upgrades within an account, and whether churned accounts stay in the denominator. Each choice shifts the result by points. Ask for the method before comparing any two NRR figures.
Can NRR be above 100% while customers are leaving?
Yes, and it happens often. NRR nets expansion against contraction, so a base where large accounts are growing can post strong NRR while a long tail of small accounts leaves. That is why logo churn belongs next to NRR in any reporting. One number says the dollars are growing, the other says whether the customer base is.
How do you forecast NRR instead of just reporting it?
Model the expansion and contraction lines separately with the same rigor applied to new business. Most teams reach roughly 90% forecast accuracy on new and expansion revenue and only through heavy manual work that goes stale as conditions change. ORM targets 95% on that same scope, holds it from day 1 to day 90 of the quarter, and updates as the quarter progresses without manual adjustment.
See how ORM turns these insights into action
ORM builds custom revenue forecast models for B2B SaaS companies. Not dashboards. Prescriptive analytics that tell you what to do next.
Schedule a Demo