Churn benchmarks: median NRR of 101% in B2B SaaS

The most solid public churn benchmark for recurring-revenue B2B companies shows a more modest number than the market repeats. In SaaS Capital’s 2025 survey, a questionnaire answered in the first quarter by more than a thousand private B2B software companies, median net revenue retention (NRR) was 101% and median gross revenue retention (GRR) was 91%. In churn terms, the typical company in the sample lost close to 9% of its installed-base revenue in a year, between cancellations and downgrades, and won back a little more than that through expansion and price increases. The 120% figures that show up in investor decks exist, but they describe venture capital targets, selected portfolios and public companies at the top of the distribution. A Brazilian leadership team can use these numbers as a yardstick once it adjusts three things before comparing: annual deal size, the definition behind each metric and the inflation adjustment, which in Brazilian contracts enters NRR without any sales effort.

I bring these benchmarks to board meetings when customer success presents an NRR above 100% and the room concludes the base is healthy. The formulas are in B2B churn and RevOps metrics; here I cover where the circulating numbers come from, who is in each sample and how much each one is worth for a Brazilian operation.

What is the churn and NRR benchmark for B2B companies?

Surveys with a broad sample of private companies converge on NRR near 100% and GRR near 90%. The numbers above 110% come from fund portfolios, recommended targets and public companies. The table summarizes the sources, with the method behind each.

Source	Year	Sample and method	Main finding
SaaS Capital	2025	14th annual survey, more than 1,000 private B2B software companies, December 2024 revenue against December 2023	Median NRR of 101% and median GRR of 91%; GRR of 95% above US$250k in annual contract value
SaaS Capital	2023	12th annual survey, more than 1,500 private B2B software companies	Median NRR of 102% and GRR of 91%; GRR of 90% below US$25k in annual contract value and about 93% above
KeyBanc Capital Markets and Sapphire Ventures	2024	15th survey of more than 100 private SaaS companies, results released in a press release	Gross retention around 90% and net retention around 101%
KeyBanc Capital Markets and Sapphire Ventures	2025	16th edition of the same survey, results released in a press release	Gross retention fell to 86% in 2023 and is expected to approach 90%; net retention stayed above 100%
Bessemer Venture Partners	2021	Financial data from the fund’s cloud portfolio, 2010 to the first half of 2021, not a random sample	Median NRR between 115% and 125%, depending on revenue band; GRR between 85% and 90%
Bessemer Venture Partners	2023	Recommended targets for the typical Series B or C enterprise software company	NRR of 100% good, 110% better and 120% or more best
ICONIQ	2025	Quarterly data from more than 100 enterprise software companies	NRR settled in the 110% to 120% range in 2025

What do bank and investor surveys say?

SaaS Capital, the largest open sample

SaaS Capital lends to software companies against recurring revenue and has run the same questionnaire every first quarter for 14 years. Its NRR uses December 2024 monthly revenue from customers who were already customers in December 2023, divided by total December 2023 revenue, and includes upsell, cross-sell and price increases. GRR repeats the calculation without expansion and cannot exceed 100%. In the 2025 edition, companies without equity investors reported NRR of 104% and GRR of 92%, above equity-backed companies (101% and 90%), which reverses the survey’s historical pattern. Multi-year contracts came in at 103% NRR and 94% GRR, annual contracts at 101% and 90%, and month-to-month at 100% and 89%.

The authors treat GRR as table stakes: to have a shot at growing in line with peers, it needs to be at least 90%. Companies with NRR of 110% or more grew faster than the sample’s 24% median. For a reader in Brazil, the cut by contract type matters more than the cut by investor, because it shows how much term and commitment weigh on gross loss.

KeyBanc and Sapphire, the survey born at Pacific Crest

The KeyBanc Capital Markets survey descends from the old SaaS survey of Pacific Crest Securities, a technology investment bank that KeyCorp announced it would acquire in July 2014. Since 2023 it has been published with Sapphire Ventures, and the full report requires registration, so I used the official press releases. In October 2024, the 15th edition, with more than 100 private SaaS companies, reported gross and net retention steady at around 90% and 101%. In November 2025, the 16th edition stated that gross retention fell to 86% in 2023 and is expected to approach 90% again, with net retention above 100% over the same period.

The two releases do not match in the detail: one describes gross retention as steady near 90% and the next describes a drop to 86% in 2023. Neither explains the difference, which may come from a change in sample or calculation. For leadership, the useful reading is the range: the private software company that answers this survey loses between 10% and 14% of its base revenue per year before expansion.

Bessemer and ICONIQ, the growth investor’s view

Bessemer Venture Partners’ “Scaling to $100 Million”, from 2021, uses data from the fund’s cloud portfolio from 2010 to the first half of 2021, and its footnote warns that the sample is not random. Median NRR goes from 125% at companies with US$1 to 10 million in annual recurring revenue to 115% at those above US$100 million, and GRR sits between 85% and 90% across bands. The sample gathers companies a fund chose and helped grow, in a period of low rates and abundant capital for software.

In the 2023 State of the Cloud, Bessemer published the scale that circulates in presentations in Brazil: NRR of 100% is good, 110% is better and 120% or more is best. The text itself presents these values as targets for the typical Series B or C enterprise software company and warns that they shift with stage and segment. ICONIQ, which tracks more than 100 enterprise software companies, recorded NRR settling between 110% and 120% in 2025 and warns that free-trial and pilot revenue, common among AI-native companies, can inflate the metric for a while.

How does retention vary by deal size and segment?

SaaS Capital recommends benchmarking retention by annual contract value first, before age, revenue or industry, because companies selling at the same price organize and support customers in similar ways. The report attributes the higher GRR of large contracts to longer sales cycles, implementation and dedicated support.

Cut	Finding	Source
Annual contract value below US$25k	Median GRR of 90%	SaaS Capital, 2023
Annual contract value above US$25k	GRR near 93% and median NRR of at least 103%	SaaS Capital, 2023
Annual contract value above US$100k	Top quartile with NRR of 118% to 120%	SaaS Capital, 2023
Annual contract value of US$25k to 50k	Median NRR of 102%, with 97% in the bottom quartile and 111% in the top	SaaS Capital, 2025
Annual contract value above US$250k	Median GRR of 95%, against about 91% in the bands below	SaaS Capital, 2025
SMB, annual contract value near US$2k	Mindbody’s 109% NRR at its IPO	Bessemer, 2021
Enterprise, annual contract value above US$50k	Okta’s 123% NRR at its IPO	Bessemer, 2021
SMB, three years before the IPO	HubSpot’s 70% to 80% GRR	Bessemer, 2021

In SaaS Capital’s 2025 edition, the bottom quartile shows contraction in every contract-value band except the one above US$250k: at least a quarter of companies shrank their own base revenue. The widest spread appears between US$12k and 25k per year. In the operations I follow, SMB bases lose customers for reasons customer success does not control, such as the customer’s business closing, and enterprise bases lose more through scope reductions than through cancellations.

What public companies disclose

Public companies publish retention in their results, each with its own definition. The table shows five examples, as data only, with the number and definition taken from regulatory filings or the official release.

Company	Customer base	Metric and number	How the company calculates it
HubSpot	288,706 customers in 2025, with average subscription revenue of US$11,414 per customer for the year	Net Revenue Retention of 103.5% in 2025 and 101.8% in 2024	Monthly rates weighted by revenue and annualized; in 2025 the company changed the calculation, and under the previous rule 2024 was 102.2%
Freshworks	Customer service and IT service software	Net dollar retention of 104% as of June 30, 2026, against 106% a year earlier	Current ARR from customers of 12 months ago divided by their ARR on that date
Snowflake	Large companies, consumption pricing	Net revenue retention of 126% in the quarter ended July 31, 2026	A cohort’s product revenue in the last 12 months divided by its revenue in the prior 12 months
Salesforce	CRM software with a global customer base	Attrition of about 8% as of January 31, 2026	Reduction or loss of the annualized value of contracts over 12 months, excluding Slack self-service and current-year acquisitions
TOTVS	Management software in Brazil	Retention rate of 97.3% in Q2 2026, against 97.9% in Q2 2025	Share of recurring revenue at the start of the period kept at the end, with price adjustment reported separately

The numbers in the table do not compare with each other. Snowflake measures consumption revenue, which grows when the customer uses more, and Salesforce measures gross loss, close to revenue churn. HubSpot restated its own series when it changed a commercial rule, and TOTVS separates price from volume, a split that the surveys cited here do not publish. The Brazilian software sector is covered in software market in Brazil.

Why do US benchmarks mislead Brazilian leadership?

Who is in the sample

Bank and fund surveys hear from software companies that have or seek an investor or lender, and the summaries I consulted do not break the sample down by country. SaaS Capital states that it finances companies registered in the United States, Canada and the United Kingdom. Whoever answers is alive: the company that went under the year before does not fill in the questionnaire, and that survivorship bias pushes medians up. Among public companies the selection is stronger, because only those that already convinced investors reach the stock exchange.

Inflation adjustments enter NRR

SaaS Capital’s formula includes price increases in NRR. In Brazil, B2B recurring contracts usually carry an annual adjustment by the IPCA or IGP-M inflation indexes, and the TOTVS release states that about 80% of the recurring revenue base in its management segment renews on its own every year, adjusted by 12-month inflation. In an illustrative example, a company with 95% GRR and an average 5% adjustment reaches NRR close to 100% without selling a single real of expansion and lands on the US median through the index alone. Separate volume NRR, which comes from expansion sold, from the price effect.

Deal size in dollars and business model

The surveys’ contract-value bands are in dollars and in annual revenue per customer, and you need to convert your deal size at the exchange rate before picking the row in the table. Many recurring B2B operations in Brazil sell services, such as accounting, security or outsourcing, with longer contracts, termination penalties and little usage-based expansion. For them, the SaaS benchmark works as a reference distance and not as a target.

Interest rates and the cycle

TOTVS attributed the 0.6-point drop in its Q2 2026 retention rate to interest rates held high for a prolonged period, which worsened the financial capacity of part of its customers, and its provision for expected losses rose 70.5% in the quarter. Bessemer’s 2021 figures come from a decade of cheap credit. Compare your retention with a survey from the same period and remember that your customer pays Brazilian interest rates.

What these data do not say

None of these sources measures the typical Brazilian B2B company, and I found no public Brazilian survey with a described methodology on NRR and GRR; the closest local references are releases from public companies such as TOTVS. Private surveys depend on what each company reports, without audit. Public companies have audited statements, but the retention metric is each company’s own indicator.

The surveys measure retention in money and almost never report customer loss by count, a metric many Brazilian companies follow. The median hides the distribution: in SaaS Capital’s US$25k to 50k band, the bottom and top quartiles sit 14 points apart. The relationship between NRR and growth that SaaS Capital shows is correlation, and a good product with well-chosen customers can drive both at once.

Ranges such as “5% to 7% annual churn is healthy” or “3% to 5% monthly churn is normal for SMBs” circulate in blogs and presentations, and I found no primary source for them. When someone brings one of these numbers to the meeting, ask for the original report, the sample and the definition.

How do you compare your retention without fooling yourself?

The first adjustment is the period. The surveys work with annual retention, and many Brazilian companies track monthly churn. The table converts a constant monthly revenue churn into approximate annual GRR, in an illustrative calculation that ignores seasonality and renewals concentrated in certain months.

Gross revenue churn per month	Approximate annual GRR	Reading against the surveys
0.5%	94%	Above SaaS Capital’s 91% median
0.8%	91%	At SaaS Capital’s median
1%	89%	Below the 90% floor SaaS Capital treats as table stakes
2%	78%	Range Bessemer found at HubSpot before its IPO, with an SMB base
3%	69%	Loss of almost a third of base revenue per year

A 2% monthly churn, which sounds low in the meeting, takes more than a fifth of base revenue in twelve months. Once the period is right, separate what goes into NRR and pick the right contract-value band; with the three adjustments, the comparison shows whether your base sits above or below the group that looks like you.

What does leadership do with these numbers?

The roadmap starts from the data the CRM and finance already have.

1. Calculate your annual baseline. Run GRR and NRR for the last four quarters with SaaS Capital’s formula: current revenue from the customers of twelve months ago against their revenue on that date.
2. Separate price from volume. Break NRR into index adjustment, expansion sold, downgrades and cancellations, and present NRR without the adjustment next to total NRR.
3. Segment by deal size and contract type. Use bands of annual revenue per customer converted to dollars and compare each band with the matching row in the surveys.
4. Pick one or two references and state the bias. Prefer surveys with a broad sample and a published method and record in the glossary why the reference fits your model.
5. Set the target through GRR. An NRR target without a GRR floor encourages offsetting loss with expansion; set the gross retention floor by segment first.
6. Take the numbers to the monthly revenue meeting. Customer success explains GRR, sales explains expansion and finance validates the price effect, with the same table for everyone.
7. Review the reference once a year. Surveys change sample and method between editions.

If the lower contract-value bands pull your GRR down, the conversation about ICP, the ideal customer profile with sales and marketing pays off more than a retention campaign. The practices the research associates with retention are in customer retention.

Retention benchmarks and L&D: what HR needs to know

HR usually receives the topic as a churn target for customer success, sometimes copied from a deck without a source. The training need shows up when leaders discuss the target without knowing whether it is monthly or annual, whether it includes price adjustments and what kind of company it came from. The group brings together the sales, marketing and CS leaders, finance and the revenue operations person, because GRR depends on the customer sales brought in and NRR on the expansion CS drives. To measure, record before the program whether the teams present the same GRR and NRR, whether NRR appears without the price effect and whether the target has a stated source, and repeat the check one quarter later.

If your board compares base retention with numbers nobody can trace, or holds CS to an NRR that mixes price adjustments with expansion, that is the work of the in-company Revenue Operations training. I run the program with the leaders, finance and the operations person, on your operation’s data, and choose with them the external reference that fits your deal size and your model. The group leaves with the RevOps metrics matrix, with formula, owner, CRM source and stated benchmark, and with the SLA between marketing, sales and CS that feeds those numbers.

Frequently asked questions

What is the average NRR for B2B SaaS companies?

In SaaS Capital’s 2025 survey of more than a thousand private B2B software companies, median NRR was 101%. The KeyBanc and Sapphire Ventures survey reports net retention above 100% in its 2024 and 2025 editions. The 110% to 125% figures come from fund portfolios, recommended targets and public companies.

What is a good GRR in B2B?

SaaS Capital found a median GRR of 91% in 2025 and treats 90% as the floor for a software company to keep pace with peers. GRR rises with deal size: 95% above US$250k a year and about 90% below US$25k. Compare your GRR with the contract-value band closest to yours.

What is the average churn in B2B SaaS?

The surveys publish retention, and annual gross revenue churn equals 100% minus GRR. With SaaS Capital’s 91% median GRR, typical revenue churn in the sample is close to 9% a year, roughly 0.8% a month. I found no primary source for the monthly churn ranges that circulate in blogs.

Can I compare a services company with SaaS benchmarks?

Yes, as a reference distance and with adjustments. Recurring services companies tend to have longer contracts, termination penalties and less usage-based expansion, which changes GRR and NRR. Compare GRR by contract-value band first and state in the glossary that the reference comes from software.

Does an inflation-indexed price adjustment count in NRR?

Under SaaS Capital’s convention, yes, because NRR includes price increases. In Brazil, the annual index adjustment can add a few points to NRR without any expansion sale. To compare with US benchmarks and to evaluate CS, also present NRR without the adjustment effect.

How do I prepare the team to work with retention benchmarks?

Start with the leaders: each needs to calculate GRR and NRR, separate price from volume and read a benchmark’s source before accepting it as a target. Then take the shared table to the monthly revenue meeting and to the CS portfolio review, with GRR by segment as the first number. The team learns more by working on the real portfolio than from a market number on the wall.