The published SaaS churn and retention benchmarks, each with the organisation that measured it, the sample size and the date. Why the two biggest samples disagree, and how to calculate your own number correctly.
Short answer
What is a good SaaS churn rate?
There is no single number, and the two biggest published samples disagree. SaaS Capital (2026, 1,000+ private B2B SaaS companies, $3M to $20M ARR) reports median gross revenue retention of 91%, roughly 9% of the existing book lost in a year, with the 90th percentile at 100%. ChartMogul (2025, 2,700 B2B SaaS companies above $250k ARR, measured from billing data) reports median net revenue retention of 82%, and an upper quartile of 97% that still sits below SaaS Capital's median of 103%. These are not two estimates of one number, they are two populations: compare yourself to the sample that charges what you charge.
What's a good SaaS churn rate? Most pages answering that question print a tidy table with no source on it, and founders calibrate to numbers nobody measured. This page does the opposite: it gives the two largest published samples with their sample sizes and dates, shows that they disagree by 21 points on the same metric, explains which one applies to you, and then walks through calculating your own number correctly, which is the only one that decides anything.
Three numbers matter, and they're often confused:
When someone quotes a churn rate, ask which one. Most SaaS benchmarking comparisons fail because the comparing party is using a different denominator.
Two organisations publish retention data for private B2B SaaS from samples large enough to mean anything, and they do not agree. Both are below, unaveraged, because the gap between them is the most useful thing on this page.
| Segment | GRR | NRR | Measured by |
|---|---|---|---|
| Private B2B SaaS, $3M to $20M ARR, median | 91% | 103% | SaaS Capital, 24 April 2026 |
| Private B2B SaaS, $3M to $20M ARR, 90th percentile | 100% | 117.9% | SaaS Capital, 24 April 2026 |
| B2B SaaS on ChartMogul, above $250k ARR, median | Not published | 82% | ChartMogul, 10 December 2025 |
| B2B SaaS on ChartMogul, above $250k ARR, upper quartile | Not published | 97% | ChartMogul, 10 December 2025 |
| B2C SaaS on ChartMogul, median | Not published | 49% | ChartMogul, 10 December 2025 |
| AI-native companies, median | 40% | 48% | ChartMogul, 10 December 2025 |
SaaS Capital puts median net revenue retention for private B2B SaaS at 103%. ChartMogul puts the median for B2B SaaS at 82%. Same metric, same year, same nominal category, twenty-one points apart. Neither is wrong, and averaging them produces a number that describes nobody.
It is worse than a disagreement about the middle. ChartMogul's upper quartileis 97%, still below SaaS Capital's median of 103%. Three quarters of one population sit below the halfway mark of the other, which is not two estimates of one number. It is two different populations.
The difference is who is in the sample and how the number was obtained. SaaS Capital surveys companies between $3M and $20M ARR, and those companies answer for themselves. ChartMogul reads billing data directly from businesses on its platform, and its population starts at $250k ARR, so it includes a great many small, self-serve products that a survey of eight-figure companies never sees. Smaller average contracts churn harder, and self-reported figures tend to flatter.
The practical rule: if your contracts are five figures and sold by a person, the SaaS Capital numbers are your comparison. If your product is self-serve at tens of dollars a month, the ChartMogul population looks like you and the SaaS Capital median will make you feel like a failure for no reason.
The clearest public demonstration is ChartMogul's split of its AI-native cohort by what a seat costs. This is AI-native companies only, not B2B SaaS at large, and it should not be read as a benchmark for a conventional SaaS product at those prices. What it shows is the size of the effect.
| Price per month | GRR | NRR |
|---|---|---|
| Above $250 a month | 70% | 85% |
| $50 to $249 a month | 45% | 61% |
| Under $50 a month | 23% | 32% |
Three times the gross retention between the cheapest tier and the dearest one, inside a single cohort measured the same way. A product under $50 a month loses roughly three quarters of its revenue base in a year and survives only if acquisition is cheap enough to outrun it. Before comparing yourself to any published median, check that its population charges what you charge.
Net churn is the more useful cross-segment comparison because it neutralises for expansion strength. A net figure above 100% means the existing book grows without a single new logo, which is what the 90th percentile in the SaaS Capital survey reports at 117.9%.
Three calculation pitfalls trip teams up:
The denominator drift problem.If you measure monthly churn as “canceled MRR ÷ end-of-month MRR,” the number looks better than reality (because end-of-month includes new sales). The correct denominator is start-of-period MRR.
The cohort smearing problem. Aggregate churn rates hide cohort-specific patterns. A vintage cohort that churns hard at month 14 will be averaged out by newer cohorts that haven't aged into the danger zone yet. Always benchmark by cohort, not by aggregate.
The annualization problem. Compounding 5% monthly churn doesn't equal 60% annual churn - it's 46%. The formula is 1 - (1 - monthly)^12. Most public benchmarks state annual figures; if you're using monthly internally, convert properly.
The patterns that consistently separate top-quartile retention from average aren't mysterious. They're three behaviors:
If your gross churn is above the median for your tier, the highest- ROI move is almost never product (assuming the product works). The highest-ROI move is closing the gap between “customer decided to leave” and “CS team finds out.” AI churn prediction is the structural fix; everything else (better onboarding, better QBRs, better save offers) follows from it. Read the breakdown of the six signals that predict churn for the inputs that matter most.
A published median is only useful if its population looks like yours. SaaS Capital's 2026 survey of more than 1,000 private B2B SaaS companies between $3M and $20M ARR puts median gross revenue retention at 91%, which is about 9% of the existing book lost in a year, and the 90th percentile at 100%. ChartMogul, reading billing data across 2,700 B2B SaaS companies above $250k ARR, puts median net revenue retention at 82%, with an upper quartile of 97% that is still below SaaS Capital's median of 103%. If you sell five-figure contracts, compare yourself to the first number. If you are self-serve at tens of dollars a month, the second population is the one that looks like you.
There is no single average, and the two largest published samples disagree by 21 points on net revenue retention for nominally the same category: 103% from SaaS Capital's 2026 survey and 82% from ChartMogul's 2025 platform data. The gap is the sample, not the metric. Price point is the strongest predictor: within ChartMogul's AI-native cohort, gross retention runs 70% above $250 a month, 45% between $50 and $249, and 23% under $50.
Gross revenue churn counts only the MRR lost to cancellations and downgrades. Net revenue churn subtracts expansion MRR from existing customers, so upgrades and seat additions offset the losses. Gross churn measures how leaky the bucket is; net churn measures whether the bucket is filling anyway. Net churn can be negative, gross churn cannot.
Divide the MRR lost in the period by the MRR at the START of the period, never the end. Using end-of-period MRR includes new sales in the denominator and flatters the number. Calculate it per cohort rather than in aggregate, because an old cohort churning hard is hidden by newer cohorts that have not aged into the danger zone yet.
Annual churn is 1 minus (1 minus monthly churn) raised to the twelfth power, not monthly churn multiplied by twelve. 5% monthly churn is 46% annually, not 60%. The difference matters because most published benchmarks are annual figures and most internal dashboards track monthly, so teams routinely compare themselves against a number 14 points harsher than the one they computed.
Negative net revenue churn means expansion revenue from existing customers exceeds everything lost to cancellations and downgrades, so the customer base grows in revenue terms even with zero new logos. Top-quartile SaaS companies run between -5% and -15% net revenue churn annually, which is the same thing as net revenue retention of 105% to 115%.
Sources, read 25 September 2026
Exeechain did not measure any of the figures above and does not aggregate churn across its customers' books. Every number on this page belongs to the organisation named beside it, and the reasoning between the tables is ours. Your own number is the only one that decides anything: run the churn calculator on your own start-of-period MRR, then put a dollar figure on it with the leak benchmark. Churn is the visible part; SaaS revenue leakage covers the money that leaves without anyone cancelling.
Keep reading
Customer success
9 min read · Sep 20, 2026
Retention metrics
8 min read · Sep 20, 2026
Customer success
8 min read · Sep 20, 2026
First scores in 15 minutes. Every connected signal read within 24 hours. From $299/mo. AI churn scoring and drafted save emails; daily scoring and automated QBRs from Multiplier.
Not ready to switch? Size your leak from your MRR and churn rate, with no billing access at all.