The NRR formula, with a worked example
Take every customer who was paying you at the start of the period. That is your cohort. NRR compares what that cohort pays at the end of the period with what it paid at the start. ChartMogul and SaaS Capital both define it this way, over 12 months by default (ChartMogul, SaaS Capital).
| Movement | MRR | Counts in NRR | Counts in GRR |
|---|---|---|---|
| Starting MRR of the cohort | $100,000 | Denominator | Denominator |
| Expansion: extra seats, upgrades, cross-sells, price increases | +$14,000 | Yes | No |
| Contraction: downgrades, fewer seats, discounts at renewal | −$3,000 | Yes | Yes |
| Churn: customers who left | −$6,000 | Yes | Yes |
| Revenue from customers who joined during the year | +$22,000 | No | No |
- NRR = ($100,000 + $14,000 − $3,000 − $6,000) ÷ $100,000 = 105%.
- GRR = ($100,000 − $3,000 − $6,000) ÷ $100,000 = 91%.
- The $22,000 from new customers is growth, but it belongs in new bookings, not in retention.
Net dollar retention (NDR) and gross dollar retention (GDR) are the same metrics under other names. Logo retention (the share of customers who stayed) is a different metric: you can keep 95% of logos and still post 85% GRR if the biggest accounts are the ones that left.
Six calculation mistakes that make NRR look better than it is
NRR is easy to inflate by accident. Before you compare yourself with any benchmark, check your method against these.
| Mistake | What it does to NRR | Fix |
|---|---|---|
| New customers leak into the numerator | Inflates it, sometimes by many points | Freeze the cohort on day one. Anyone who signed after the start date is excluded. |
| Averaging twelve monthly NRRs | Hides churn that lands at renewal months | Measure one 12-month cohort, start to end. |
| Counting one-off services or usage spikes as expansion | Inflates expansion that will not recur | Use recurring revenue only, and normalize usage-based revenue to a trailing average. |
| Netting churn against reactivations silently | Understates churn | Record churn and reactivation as separate movements, as ChartMogul does. |
| Mixing currencies at today's exchange rate | Moves NRR with exchange rates, not customers | Use constant-currency figures for both ends of the period. |
| Reporting only blended NRR | One large expansion hides a churning segment | Report NRR and GRR by segment and by contract size. |
NRR benchmarks: what good looks like in 2025
The most useful public benchmark for private B2B companies is SaaS Capital's annual survey. Its 2025 brief puts the median at 101% NRR and 91% GRR across all SaaS companies, and it recommends benchmarking by annual contract value (ACV) because companies at the same price point sell and support customers alike (SaaS Capital 2025 retention brief).
| Segment | Median NRR | Median GRR |
|---|---|---|
| All companies | 101% | 91% |
| ACV $25,000 to $50,000 | 102% (bottom quartile 97%, top quartile 111%) | Not published for this band |
| ACV above $250,000 | Higher than lower-ACV bands | 95% |
| ACV below $250,000 | Rises with ACV | About 91% |
| Multi-year contracts | 103% | 94% |
| Annual contracts | 101% | 90% |
| Month-to-month contracts | 100% | 89% |
| Bootstrapped companies | 104% | 92% |
| Equity-backed companies | 101% | 90% |
ChartMogul's December 2025 analysis of 3,500 software companies, built from billing data rather than a survey, found a much lower median of 82% NRR for B2B SaaS, with the top quartile at 97%, and around 48% to 49% for consumer and AI-native products (ChartMogul). The two sources measure different populations in different ways, so pick the one that resembles you and compare against it consistently.
Why the number matters: NRR and growth
- In SaaS Capital's 2025 survey, the median company grew 24% a year. Groups with NRR of at least 110% grew faster than that, and groups below 100% grew slower.
- Moving from the 100% to 110% NRR band to the 110% to 120% band came with 9 percentage points more growth, and the highest-NRR companies grew at double the median rate.
- SaaS Capital treats GRR as table stakes: "to have a shot at performance parity with your peers, GRR must be at least 90%."
Where NRR is actually decided
NRR has two halves, and they need different owners and different signals. The gross half is defense: keep the revenue. The expansion half is offense: grow it. Most teams staff the renewal call and miss the months before it.
| Lever | Leading indicator to watch | Usual owner | Play |
|---|---|---|---|
| Logo churn | Champion leaves or goes quiet; usage falls for 30 days; renewal in 120 days with no exec contact | CSM | Multi-thread the account, run a value review, escalate to an exec sponsor |
| Contraction | Seats unused for 60 days; a reorg or layoffs; budget owner changed | CSM or AM | Re-onboard the unused seats before renewal, or right-size them early in exchange for a longer term |
| Seat expansion | Active seats near the plan limit; new hires in the buying team | AM | Propose the next tier before they hit the wall |
| Cross-sell | A question on a call about a problem another product solves; a new team asking for access | AM or AE | Book a discovery call with the new team's owner |
| Price | Contract up for renewal on old pricing | AM and finance | Pair any increase with a visible value review, not a letter |
How to build a customer health score that predicts churn
A customer health score is a single number per account that tells a CSM where to look first. The common failure is a score built from opinions that nobody has tested against what actually churned. Build it in four steps.
- List the signals you can actually measureProduct usage trend, active seats against purchased seats, number of engaged contacts, time since the last executive conversation, open support issues and their age, sentiment on recent calls, payment problems, and days to renewal.
- Back-test them on last yearTake every account that renewed or churned in the last 12 months. For each signal, compare its value 90 days before the renewal date for the two groups. Keep the signals that separate them. Drop the ones that do not, however intuitive they feel.
- Weight and bandGive the surviving signals weights, score accounts from 0 to 100 and set three bands. Start simple: a score your CSMs understand beats a precise one they ignore.
- Attach a play to every bandA score without an action is a dashboard. Green gets the expansion conversation, yellow gets a value review within 30 days, red gets an exec sponsor and a save plan this week.
| Signal | Example weight | Red flag |
|---|---|---|
| Usage trend, last 30 days vs prior 30 | 25 | Down more than 20% |
| Active seats vs purchased | 20 | Below 60% |
| Engaged contacts in the last 60 days | 20 | One, or zero |
| Time since the last exec-level conversation | 15 | More than 6 months |
| Open support issues older than 14 days | 10 | Two or more |
| Sentiment on the last three calls | 10 | Negative or no calls |
A renewal and expansion calendar for every account
Put this on every account with an annual or multi-year contract. The dates are a starting point: lengthen them if your renewals need procurement or legal review.
| When | What happens | Output |
|---|---|---|
| All year | Watch expansion signals: seat ceilings, new teams, hiring, product questions on calls | Expansion opportunities opened as they appear, not saved up for renewal |
| 120 days before renewal | Check health score, engaged contacts and exec coverage | At-risk list with a reason per account |
| 90 days | Value review with the champion: outcomes against the goals they bought for | Written value summary; expansion or right-size proposal |
| 60 days | Exec-to-exec conversation on red and yellow accounts | Save plan or confirmed renewal path |
| 30 days | Commercials and paperwork | Signed renewal, or a churn reason recorded in the CRM |
| After renewal or churn | Log the outcome and reason; feed it back into the health score | A better score next year |
Where AI agents help, and where a person still has to
The work around a renewal is mostly watching and chasing: reading usage, noticing that nobody has spoken to the champion in two months, finding the new team that might need seats, logging what happened. That is the part agents are good at. The value review, the exec conversation and the negotiation are not. Use agents to make sure no account goes unwatched, and keep people on the conversations.
Frequently asked questions
What is a good net revenue retention rate?
Above 100% means existing customers grow on their own. In SaaS Capital's 2025 survey of private B2B SaaS companies, the median was 101% NRR and 91% GRR, and medians rose with contract size. ChartMogul's billing-data analysis of 3,500 software companies found a median of 82% for B2B SaaS. Compare yourself with companies at a similar contract value.
What is the difference between NRR and GRR?
Both start from the revenue a customer cohort paid at the beginning of the period. NRR adds expansion and subtracts contraction and churn, so it can exceed 100%. GRR subtracts contraction and churn but ignores expansion, so it can never exceed 100%. GRR shows how well you keep revenue; NRR shows whether the base grows.
Does net revenue retention include new customers?
No. NRR only measures customers who were already paying at the start of the period. Revenue from customers who signed during the period is new business and is excluded from both the numerator and the denominator.
Do price increases count toward NRR?
Yes. SaaS Capital's definition includes upsells, cross-sells and price increases. Report how much of your expansion came from price so that a list-price rise is not mistaken for customers buying more.
Is NRR the same as net dollar retention?
Yes. Net dollar retention (NDR) and net revenue retention (NRR) are the same metric, and gross dollar retention (GDR) is the same as gross revenue retention (GRR).
How do you improve net revenue retention?
Protect the gross half and grow the expansion half separately. Build a health score back-tested against last year's churn, start risk checks 120 days before renewal, multi-thread every account, and open expansion opportunities when signals appear rather than at renewal. Agents can do the watching and logging; people should hold the value reviews and negotiations.
- 1.SaaS Capital, 2025 SaaS Retention Benchmarks for Private B2B Companies (research brief) (2025)
- 2.SaaS Capital, What is a Good Retention Rate for a Private SaaS Company in 2025? (2025)
- 3.ChartMogul, Net Revenue Retention (NRR), with its December 2025 benchmark analysis (checked 2026-11-18) (2026)
Spend your reps' day on conversations, not ringing phones.
TwinsAI parallel-dials up to 10 lines at once, validates every number with the carrier before it rings, and bridges a live prospect to your rep in under 400ms. It works inside HubSpot, Salesforce, Close, Attio, Outreach, Zoho, HighLevel, Pipedrive, Apollo and SmartMoving, with no seat minimum.