AI-native SaaS products kept a median of just 40% of gross revenue over the 12 months ending September 2025, according to ChartMogul’s analysis of 3,500 software companies (n=200 AI-native). That means the typical AI app loses 60 cents of every recurring dollar within a year, roughly three times the churn rate of a traditional B2B SaaS business. Churn is not one number: it varies sharply by company stage, price point, and whether the product is B2B, B2C, or AI-native, and the 2026 data below breaks it down by each of those segments.
If your product is a SaaS or software subscription, the segment your company sits in matters more than any single industry-wide “average churn rate” figure, because the gap between the best and worst segments below is wider than most of those averages themselves.
How much revenue do AI-native SaaS companies actually keep?
AI-native software companies kept a median of just 40% of gross revenue and 48% of net revenue over the year ending September 2025, per ChartMogul analyst Kyle Poyar’s study of 3,500 software companies sorted into B2B SaaS, B2C SaaS, and AI-native segments. Gross revenue retention (GRR) excludes upsell; net revenue retention (NRR) includes it. A GRR of 40% means 60% of last year’s revenue base is gone, before counting any new sales.
Figure 1: Share of AI-native SaaS gross revenue retained versus churned over 12 months. Source: ChartMogul, “The SaaS Retention Report: The AI Churn Wave,” analysis of 200 AI-native companies with at least $250k ARR, data through September 2025.
There is a silver lining in the trend line. AI-native GRR jumped from a median of 27% in January 2025 to 40% by September 2025, which Poyar attributes to early “AI tourists” moving on while more committed, production-stage customers stay. Even with that recovery, AI-native retention still looks nothing like mature B2B SaaS, and a company built on ARR-style valuation multiples needs to know which side of that gap it sits on. If your product bundles AI features into a subscription, a terms and conditions generator that spells out what happens to usage-based or trial pricing at renewal is one of the cheaper fixes available before a renewal cycle turns into a cancellation.
How does churn differ between B2B, B2C, and AI-native SaaS?
Business model is one of the biggest churn predictors ChartMogul’s dataset captured. Median net revenue retention was 82% for B2B SaaS (upper quartile 97%), compared with 49% for B2C SaaS and 48% for AI-native software, drawn from roughly 2,700 B2B SaaS, 600 B2C SaaS, and 200 AI-native companies in the same 3,500-company dataset.
Figure 2: Median net revenue retention by business model. Source: ChartMogul, “The SaaS Retention Report: The AI Churn Wave,” 2025 data.
SaaS Retention in Four Numbers (2025 Data)
Price point compounds the business-model effect within AI-native products specifically. AI-native products selling for more than $250 a month see 70% GRR and 85% NRR, essentially matching B2B SaaS. Products priced between $50 and $249 a month see 45% GRR and 61% NRR, about 15 points worse than typical B2B SaaS. Products under $50 a month see just 23% GRR and 32% NRR, roughly 20 points worse than either B2B or B2C SaaS. Low-priced, self-serve AI products are easy to buy and just as easy to cancel.
The segment differences hold up because they track how deliberately a customer bought in the first place: cheap, self-serve, AI-native products attract more “tourists” who never move past experimentation, while higher-priced B2B contracts filter for buyers who planned to keep paying. How a company structures its pricing tiers and contract length in the first place, a topic worth its own dedicated look at SaaS pricing models and averages, shapes which side of this gap a product lands on before churn ever becomes a retention-team problem.
Churn on the vendor side also has a buyer-side mirror. The average company now runs 106 SaaS applications, down from a peak of 130 in 2022, which means procurement teams are actively trimming their own subscription stacks. A vendor with weak retention numbers is exactly the kind of line item that gets cut first in that trimming process.
How does churn shrink as a SaaS company scales?
Company stage is the other major churn predictor. A median early-stage SaaS company under $300k ARR has a monthly customer churn rate of 6.5%, falling to 4.1% at $300k-1M ARR, 3.7% at $1-3M ARR, 3.8% at $3-8M ARR, and 3.1% at $8-15M ARR, according to ChartMogul’s benchmarks aggregated from over 2,500 SaaS businesses (figures current as of July 2026, per ChartMogul’s live SaaS metrics library).
Figure 3: Median monthly customer churn rate by ARR stage. Source: ChartMogul SaaS metrics benchmarks, aggregated from 2,500+ SaaS businesses, as of July 2026.
The pattern is not perfectly linear. Churn ticks back up to 4.1% in the $15-30M ARR band, which ChartMogul’s own commentary does not fully explain, so treat that single band as a real but unexplained wrinkle in the trend rather than dropping it from the picture. It is a reminder that “grow past $10M ARR and churn keeps falling forever” is not something even this dataset actually supports. That climb happens against a market still expanding fast overall: our SaaS market size and company count breakdown puts the global SaaS market at over $315 billion in 2025, so even a middling churn rate compounds into real dollars at that scale.
Later-stage companies tell a similar story from a different data source. The 16th Annual KeyBanc Capital Markets and Sapphire Ventures Private SaaS Company Survey, published November 13, 2025, reports gross revenue retention approaching 90% and net revenue retention above 100% among the surveyed private SaaS companies, alongside 20% projected 2025 ARR growth, up from 15% in 2024. That cohort skews toward larger, venture-backed companies than ChartMogul’s median, which is consistent with churn continuing to compress well past the $30M ARR ceiling of ChartMogul’s own bands.
Does net revenue retention predict future churn risk?
Yes, and the relationship is close to two-to-one. Companies with net revenue retention below 60% have a median customer churn rate of 7.3%, roughly double the 3.1% to 3.5% range seen in the 60-100% NRR bands and nearly quadruple the 1.7% median seen at 100%+ NRR, according to ChartMogul’s benchmark data.
ChartMogul’s own dataset shows a related pattern worth a specific caveat: rapid growth correlates with somewhat higher churn, especially for early-stage, high-growth B2C companies, because a wave of viral signups tends to include a large share of people who never intended to stick around past the first few months. A company growing revenue faster than 100% a year had a median churn rate of 5.3% in ChartMogul’s data, above every slower-growth band except the shrinking (below 0% growth) cohort. NRR still functions as the more reliable early-warning signal of the two, because it is measuring what current customers actually do rather than how fast new ones are arriving.
What is the difference between voluntary and involuntary SaaS churn?
Voluntary (proactive) churn is a customer deliberately canceling, often because they finished the job the product was hired to do. Involuntary (passive) churn happens when a payment simply fails, most often an expired or declined card, and ChartMogul reports observing cases where passive churn accounts for as much as a third of all churn at a given company, though it cautions that share varies a great deal by billing setup and is not a fixed industry constant.
Figure 4: The two paths revenue takes out the door: involuntary (failed payment) versus voluntary (deliberate cancellation). Source: synthesized from ChartMogul’s churn-type definitions.
Recovery tooling is not a footnote here. Software companies using Recurly’s payment-recovery tools reclaimed more than $155 million in revenue from failed payments in 2025 alone, according to Recurly’s 2026 State of Subscriptions report, drawn from a dataset of over 76 million subscribers across software, digital media, healthcare, and education. Since involuntary churn is a billing-operations problem rather than a product or pricing problem, it is usually the cheapest segment of churn to fix.
Where is SaaS retention headed into 2026?
The 2025 data points in one direction: retention is bifurcating by business model even as it slowly repairs within the AI-native segment specifically.
Figure 5: Retention and churn-recovery data points across 2025 and into 2026. Sources: ChartMogul, KeyBanc Capital Markets / Sapphire Ventures, Recurly.
How do the segments compare side by side?
| Segment | Median retention | Effective annual churn | Source |
|---|---|---|---|
| AI-native SaaS, all price tiers | 40% GRR / 48% NRR | 60% gross | ChartMogul, n=200, Sept 2025 |
| AI-native SaaS, priced under $50/mo | 23% GRR / 32% NRR | 77% gross | ChartMogul, Sept 2025 |
| B2C SaaS | 49% NRR | 51% net | ChartMogul, n=~600, Sept 2025 |
| B2B SaaS | 82% NRR | 18% net | ChartMogul, n=~2,700, Sept 2025 |
| Later-stage private SaaS (venture-backed) | ~90% gross retention, 100%+ NRR | ~10% gross | KeyBanc/Sapphire, Nov 2025 |
“Effective annual churn” above is calculated as 100% minus the retention figure in the same row; it mixes gross and net churn concepts across rows on purpose, to show what each stated retention number implies rather than to claim a single standardized metric.
The Bottom Line
There is no single “average SaaS churn rate” worth quoting on its own in 2026. A self-serve, sub-$50-a-month AI tool loses the majority of its revenue in a year, a median early-stage SaaS company churns 6.5% of customers a month, and a mature B2B SaaS company with strong net revenue retention keeps around 82% of revenue without lifting a finger toward new sales. The fastest, cheapest wins sit at the edges: recovering involuntary churn through better dunning, and matching pricing and contract length to how deliberately customers actually buy. Get renewal terms and auto-billing disclosures right in your terms and conditions, and the involuntary side of that gap gets easier to close.
Frequently Asked Questions
What is a good churn rate for a SaaS company? For monthly customer churn, top-decile companies sit near 1.3 to 1.5 percent regardless of ARR stage, while a median company runs 3.1 to 6.5 percent depending on size, according to ChartMogul’s benchmarks drawn from over 2,500 SaaS businesses. Below 60% net revenue retention roughly doubles a company’s median churn rate.
How much revenue do AI-native SaaS companies retain? AI-native software companies had a median gross revenue retention of just 40% and net revenue retention of 48% for the 12 months ending September 2025, according to ChartMogul’s analysis of 200 AI-native companies within a 3,500-company dataset, meaning the typical AI app loses 60 cents of every recurring dollar within a year.
Does SaaS churn improve as a company grows? Yes. Median monthly customer churn falls from 6.5% for companies under $300k ARR to 3.1% for companies between $8M and $15M ARR, per ChartMogul. Later-stage, venture-backed private SaaS companies report gross revenue retention near 90%, according to the 16th Annual KeyBanc Capital Markets and Sapphire Ventures Private SaaS Company Survey published in November 2025.
What is the difference between voluntary and involuntary SaaS churn? Voluntary churn is a customer deliberately canceling; involuntary (passive) churn happens when a payment fails, such as an expired card, and ChartMogul has observed cases where it accounts for as much as a third of all churn. Recurly’s software-industry customers recovered more than $155 million in revenue through failed-payment recovery tools in 2025 alone.
Sources and References
- ChartMogul. Poyar, K. (2025). “The SaaS Retention Report: The AI Churn Wave.” Analysis of 3,500 software companies (2,700 B2B SaaS, 600 B2C SaaS, 200 AI-native), retention data through September 2025.
- ChartMogul. “Customer Churn Rate” benchmarks, SaaS Metrics Library. Aggregated and anonymized revenue data from over 2,500 SaaS businesses; live benchmarks page, data as of July 2026.
- KeyBanc Capital Markets / Sapphire Ventures. (2025). “16th Annual Private SaaS Company Survey.” Published November 13, 2025.
- Recurly. (2026). “2026 State of Subscriptions Report.” Dataset of over 76 million subscribers across software, digital media, healthcare, and education; software-industry failed-payment recovery figure covers 2025.
Note: All figures verified as of September 2026. ChartMogul’s SaaS Metrics Library is a live benchmarks page rather than a fixed report, so its churn figures are hedged as “as of July 2026” throughout this article and will be re-checked at the next refresh. Retention data shifts quickly for AI-native software specifically, so headline figures here are scheduled for review at least twice a year.
