SaaS churn rate benchmarks
SaaS churn rate is the share of customers or revenue a subscription business loses in a period. Here is every published figure we can show, side by side, with what each one counts.
3.22% to 7%
"Churn rate" means at least two different things, and this page keeps them separate rather than blending them. Logo churn — customers who cancel or fail to renew — runs a median 3.22% a year for SaaS specifically on Recurly's subscription-billing network as of July 2026, the lowest of any vertical Recurly tracks. Revenue churn is a different metric: ChartMogul's December 2025 report on 3,500 software companies found a median Net Revenue Retention of 82% for B2B SaaS (upper quartile 97%), meaning the typical company in that sample lost about 18% of revenue net of expansion — in a report specifically framed around what it calls "the AI churn wave" of 2025, not necessarily a stable baseline. 14 figures from 3 sources, grouped by what they count (annualized customer/logo churn (lighter capital), annualized revenue churn (lighter capital) is charted too, but not part of the headline range). We never average them.
Published SaaS churn rate figures by way of counting, from 3.22% to 27.25%. Every figure is listed in the source list below.
Tap or hover a point to see its source.
Key findings
Each one cites its source figures.
Recurly's median annual logo churn for SaaS specifically (July 2026 data) is the lowest of the verticals it tracks — below business/professional services (3.44%), travel & hospitality (3.91%), digital media (4.14%), ecommerce (4.25%) and education (4.99%).
RecurlyRecurly splits SaaS logo churn into voluntary (active cancellation, 2.16%) and involuntary (failed payments, 1.06%) — voluntary churn is roughly twice involuntary churn, and only voluntary churn is really about product/value fit.
RecurlyLogo churn and revenue retention move together: ChartMogul's 2024 report found companies with Net Revenue Retention under 60% had a median 7% logo churn rate, "double the rate of those with ≥100% NRR." Weak revenue retention and weak customer retention tend to show up in the same accounts.
ChartMogulSame data, different answers
How the way of counting changes the number, on one scale.
SaaS logo churn vs other subscription verticals (Recurly, July 2026)
SaaS churns less than every other vertical Recurly tracks on its network.
Voluntary vs involuntary churn, SaaS (Recurly, July 2026)
Failed payments (involuntary) are a real but smaller share of SaaS churn than active cancellations (voluntary).
Logo churn by revenue-retention tier (ChartMogul, 2024 data)
Companies with weak net revenue retention also churn more customers — ChartMogul states the low-NRR figure directly and describes the high-NRR figure as half of it.
Source by source
Open any source for its definition, sample and method.
RecurlyChurn Rate Benchmarks: SaaS, Media, Retail & More Industries · Live benchmark, updated with July 2026 data3.22%logo / customer churn · SaaS3.6%logo / customer churn · All industries on the Recurly networkGrade A
- What counts
- Logo (customer) churn: subscribers who cancel or fail to renew within a given period, split into voluntary (active cancellation) and involuntary (payment failure). Explicitly distinguished from MRR/revenue churn on the page.
- Statistic
- Median annual churn rate, with voluntary/involuntary split
- Sample
- Not stated (Recurly's network of subscription businesses using its billing platform)
- Data window
- Trailing period ending July 2026 (exact window not stated)
- Method
- Live benchmark drawn from Recurly's own subscription-billing customer network, reported as median annual churn by vertical; the page notes that annual churn is not simply 12× monthly churn because of compounding.
- Published cuts
- vertical (SaaS, business/professional services, travel/hospitality, digital media, ecommerce, education) · voluntary vs involuntary
- Limitations
- Number of businesses/subscribers behind the medians not disclosed
- Vendor-reported: Recurly's own billing-platform customers only
- Live/rolling benchmark: the exact figure will have moved by the time this is read
- Licence
- Recurly's published terms found in this pass govern use of its billing SaaS product; silent on citing its published research/benchmark pages. Headline plus several vertical cuts, attributed and linked.
- Checked
- Sep 30, 2026 · Read the source
ChartMogulThe SaaS Retention Report: The New Normal For SaaS · 2024 report · older data7%logo / customer churn · Companies with Net Revenue Retention below 60%3.5%logo / customer churn · Companies with Net Revenue Retention ≥ 100%Grade B
- What counts
- Logo (customer) churn rate, segmented by the same company's Net Revenue Retention tier, not by ARR band.
- Statistic
- Median
- Sample
- Over 2,500 SaaS businesses
- Data window
- H1 2021 through H1 2024
- Method
- ChartMogul's own analytics-platform customer data, H1 2021–H1 2024, correlating logo churn with NRR tier.
- Published cuts
- NRR tier (below 60%, up to ≥100%)
- Limitations
- Stale relative to the 2025 report; multi-year window makes the exact vintage of any single figure unclear
- Segments by NRR tier, not by ARR band or company stage
- The ≥100%-NRR figure (3.5%) is ChartMogul's own stated ratio ("double the rate") applied to its directly published 7% figure, not an independently published number
- Licence
- Same assessment as ChartMogul's 2025 report. A single comparative figure quoted with attribution.
- Checked
- Sep 30, 2026 · Read the source
Lighter Capital2025 B2B SaaS Startup Benchmarks · 20259.35%annualized customer/logo churn (lighter capital) · 25th percentile16.25%annualized customer/logo churn (lighter capital) · median20.46%annualized customer/logo churn (lighter capital) · meanGrade A
- What counts
- Reports both customer/logo churn and revenue churn, separately, both annualized from a trailing-12-month average monthly churn rate using 1 − (1 − monthly churn)^12 — not the same calculation as a directly observed one-month churn rate.
- Statistic
- 25th/median/mean/75th percentile distribution
- Sample
- 155 private B2B SaaS startups across 11 verticals in the United States, Canada and Australia — connected business data, not a survey
- Data window
- Connected business data recorded CY 2020–2024, aggregated over five years into the 2025 benchmark output
- Method
- Connected business data from 155 private B2B SaaS startups, CY2020–2024, aggregated over five years for the 2025 output.
- Published cuts
- customer vs. revenue churn · percentile (25th, median, mean, 75th) · ARR band · vertical · company age
- Limitations
- Private B2B SaaS startup population; five-year aggregation
- Annualized from monthly churn, not a directly observed annual figure — do not blend with a source that measures annual churn directly
- No numeric breakdown independently confirmed beyond what is stated on the public blog page
- Licence
- Reported here as factual findings in our own wording and presentation, with attribution and a link to the source. This is not a claim that Lighter Capital has granted a reproduction licence, and we do not copy its prose, charts, tables or page design. Its Terms of Use contain generic sitewide IP boilerplate (no copying or reproducing site content); we read it as not specifically restricting citation of published figures. Flagged for review: the clause is broadly worded.
- Checked
- Oct 1, 2026 · Read the source
ChartMogulThe State of SaaS Retention: The AI Churn Wave · December 2025 reportContext sourceGrade B
- What counts
- Net Revenue Retention (NRR) and Gross Revenue Retention (GRR) — revenue-based, not logo/customer-based. "The median NRR was 82% (the upper quartile was 97%)" for B2B SaaS.
- Statistic
- Median, with upper-quartile figures for B2B SaaS
- Sample
- Roughly 2,700 B2B SaaS, 600 B2C SaaS and 200 AI-native software companies, all at least $250k ARR
- Data window
- Annualized retention throughout 2025
- Method
- ChartMogul's own analytics-platform customer data, businesses with at least $250k ARR (chosen because "historical retention data is less reliable for very early-stage startups"), split into B2B SaaS, B2C SaaS and AI-native cohorts.
- Published cuts
- business model (B2B SaaS, B2C SaaS, AI-native) · pricing tier (for AI-native cohort)
- Limitations
- Revenue-based (NRR/GRR), not logo churn — do not blend with Recurly's customer-churn figures
- Excludes businesses under $250k ARR
- Specifically framed around a 2025 "AI churn wave"; may not represent a typical/stable year
- Vendor-reported: ChartMogul's own analytics customers
- Licence
- PENDING: ChartMogul's Terms of Service bar copying/reproducing "any portion of ChartMogul or its Services," which in context appears to describe the analytics SaaS product rather than this public marketing report. Treated as quote-eligible for a couple of headline figures with attribution; owner should re-check before wider reuse.
- Checked
- Sep 30, 2026 · Read the source
By industry
Selected figures, each with its source. Full tables live in each publisher's report.
Why sources disagree
Each factor moves the number on its own.
is ChartMogul's December 2025 median Net Revenue Retention for B2B SaaS companies (upper quartile 97%) in its report titled "The State of SaaS Retention: The AI Churn Wave." This is a revenue-churn measure — it moves with expansion/contraction/upsells, not customer count — and should not be blended with the logo-churn figures on this page.
ChartMogul, The State of SaaS Retention: The AI Churn WavePaddle's SaaS churn-rate guide, last substantially updated in 2022, compiles older third-party studies (a 2018 Recurly study of 1,500+ companies giving 4.8% monthly SaaS churn; a private 2018 KeyBanc Capital Markets survey of 162 companies giving a 21% median annual logo churn; a 2016 Totango report; a 300-company dataset from Nathan Latka). These are old, secondary citations rather than current primary data and are not used for the headline figures on this page.
Paddle, "SaaS churn rate: What is a good average churn rate for SaaS?"Logo churn vs revenue churn
Logo churn counts customers; revenue churn counts dollars, and its "net" version (NRR) also credits upsell revenue. Recurly's 3.22% SaaS logo churn and ChartMogul's 82% B2B SaaS NRR (≈18% net revenue churn) are not comparable numbers, even though both get called "churn."
Voluntary vs involuntary
Recurly splits SaaS churn into voluntary (customer actively cancels, 2.16%) and involuntary (payment fails, 1.06%). Only the voluntary share is really about whether customers want to stay; involuntary churn is often fixable with better payment retry logic.
Which businesses are in the sample
Recurly's figures come from businesses using Recurly's own billing platform; ChartMogul's from businesses on ChartMogul's analytics platform with at least $250k ARR. Neither is a random sample of all SaaS companies, and the two samples do not overlap.
Report vintage
ChartMogul's 82% NRR figure is from a report specifically framed around 2025's "AI churn wave" and may run lower than a typical year; its own 2024 report ("The New Normal") used a different multi-year window (H1 2021–2024) and a different lens (churn by NRR tier, not by ARR band).
Segmentation that could not be confirmed
Several widely circulated claims about SaaS churn by company stage or ARR band (for example, "early-stage companies under $300K ARR churn around 6.5% a month") could not be traced to a specific, checkable, dated page in this research pass — the ChartMogul reports we could read directly segment by revenue-retention tier and by B2B/B2C/AI-native category, not by ARR band for churn specifically. That segmentation is left off this page rather than repeated from an unverified source.
What is a good SaaS churn rate?
It depends on how your platform counts and who you send to.
If you track logo/customer churn (the usual SaaS default)
Recurly's July 2026 median for SaaS specifically is 3.22% a year; its stated top-quartile performers churn 1.78% or below. Compare against SaaS (3.22%), not the 3.60% all-industries figure on Recurly's network.
If you're separating voluntary from involuntary churn
Recurly's SaaS split is 2.16% voluntary (active cancellation) and 1.06% involuntary (failed payment) — involuntary churn is usually the cheaper problem to fix.
If you track revenue churn including expansion
This is a different metric (NRR/GRR), not logo churn. ChartMogul's December 2025 data puts B2B SaaS median NRR at 82% (≈18% net revenue churn) — see the SaaS Churn Rate page's sibling, Customer Retention Rate, for the fuller GRR/NRR picture.
Before you compare
- Never blend logo churn with revenue churn (GRR/NRR) into one number — they use different denominators and measure different things.
- Recurly does not disclose how many businesses sit behind its published medians.
- ChartMogul's 82% B2B SaaS NRR figure is from a report specifically framed around 2025's "AI churn wave" and may run lower than a typical year; its 2024 report used a different, older data window.
Show 4 more cautions
- Segmentation of logo churn by company size or ARR band could not be verified against a primary, checkable source in this pass and is deliberately left off this page rather than repeated from secondary blog compilations.
- Paddle's often-cited churn-rate roundup compiles studies from 2016–2018 relayed secondhand; treated here as historical context only, not as a current benchmark.
- Every figure on this page is vendor-reported (each platform's own customers), not a random sample of all SaaS companies.
- Some SaaS churn benchmarks report churn as a directly-measured annual figure; others (named in Sources below) annualize a trailing monthly average using 1 − (1 − monthly churn)^12 — these two approaches can diverge meaningfully even for the same underlying business, so check which method a source uses before comparing.
How to calculate SaaS churn rate
- State whether you mean logo (customer) churn or revenue churn — they are different metrics and rarely match.
- Split voluntary (customer cancels) from involuntary (payment fails) churn where you can; only voluntary churn tells you about product/value fit.
- Annual churn is not simply 12× monthly churn, because of compounding — Recurly's own page notes this explicitly.
- Revenue churn's "net" version (contributing to NRR) credits expansion revenue from upsells and cross-sells, so it can be negative (net revenue growth) even while some customers churn.
Questions about SaaS churn rates
Tools to measure and improve it
Chosen for what they do, not ranked.
Sources and licence notes4 sources · 3 quoted · 0 linked only · checked Oct 1, 2026
- Recurly, Churn Rate Benchmarks: SaaS, Media, Retail & More Industries · Live benchmark, updated with July 2026 data · last checked Sep 30, 2026Recurly's published terms found in this pass govern use of its billing SaaS product; silent on citing its published research/benchmark pages. Headline plus several vertical cuts, attributed and linked.
- ChartMogul, The State of SaaS Retention: The AI Churn Wave · December 2025 report · last checked Sep 30, 2026PENDING: ChartMogul's Terms of Service bar copying/reproducing "any portion of ChartMogul or its Services," which in context appears to describe the analytics SaaS product rather than this public marketing report. Treated as quote-eligible for a couple of headline figures with attribution; owner should re-check before wider reuse.
- ChartMogul, The SaaS Retention Report: The New Normal For SaaS · 2024 report · last checked Sep 30, 2026Same assessment as ChartMogul's 2025 report. A single comparative figure quoted with attribution.
- Lighter Capital, 2025 B2B SaaS Startup Benchmarks · 2025 · last checked Oct 1, 2026Reported here as factual findings in our own wording and presentation, with attribution and a link to the source. This is not a claim that Lighter Capital has granted a reproduction licence, and we do not copy its prose, charts, tables or page design. Its Terms of Use contain generic sitewide IP boilerplate (no copying or reproducing site content); we read it as not specifically restricting citation of published figures. Flagged for review: the clause is broadly worded.
These are figures reported by the platforms themselves about their own customers; we do not measure them ourselves. Found an error or a newer edition? Tell us