Customer Retention Stats 2026: SaaS Benchmarks and Methods
Customer retention statistics are useful only when the metric, period and customer population match the question. For SaaS, distinguish customer or logo retention from gross revenue retention (GRR) and net revenue retention (NRR). A company can retain fewer accounts while increasing revenue from the customers that remain.
The tables below separate financial benchmarks from survey findings about customer education. They link to the underlying publications and identify the sample for each metric where the source provides it. Publication year and measurement period are stated separately.
Current SaaS retention benchmarks
Benchmarkit’s June 2026 report covers calendar-year 2025 results from 342 B2B SaaS and AI-native software companies overall. The retention charts have smaller samples: 226 companies for GRR and 230 for NRR. Use those metric-specific counts when citing them.
| Metric | Reported median | Measurement period | Metric sample | Primary source |
|---|---|---|---|---|
| Gross revenue retention | 84% | Calendar year 2025 | 226 companies | Benchmarkit 2026 report, page 36 |
| Net revenue retention | 102% | Calendar year 2025 | 230 companies | Benchmarkit 2026 report, page 42 |
The report’s GRR trend compares 88% for 2024 with 84% for 2025: a decline of four percentage points, not four percent. These are report-level comparisons, not proof that each participating company experienced the same change. See the trend chart, page 37.
These figures describe the contributing sample, not a required target for every SaaS company. A median is the middle company result; it is not the aggregate retention rate of all customers in the sample. Do not subtract the two medians to estimate a typical company’s expansion: they come from different metric samples and need not describe the same company.
A benchmark for a defined contract-value segment
SaaS Capital’s September 2025 analysis reports the following annual NRR distribution for private B2B SaaS companies with average contract value of $25,000-$50,000. The underlying calculation compares December 2024 recurring revenue with December 2023 revenue from the same starting customer group.
| Position in the segment | Annual NRR |
|---|---|
| Lower quartile | 97% |
| Median | 102% |
| Upper quartile | 111% |
The public article does not state a separate respondent count for this ACV band. Cite the segment and period rather than attaching the overall survey size to these three values. Source: SaaS Capital, What Is a Good Retention Rate for a Private SaaS Company?, 18 September 2025.
Contract size is a useful comparison dimension, but it does not control for product category, company age, pricing changes, geography or customer mix. The fact that two reports have a 102% median does not mean their populations or methods are interchangeable.
Billing-data benchmarks: B2B, B2C and AI companies
ChartMogul’s The AI Churn Wave analysis reports annualized retention during 2025. It classified roughly 3,500 company websites into about 2,700 B2B, 600 B2C and 200 AI companies, then restricted retention analysis to companies with at least $250,000 ARR. Those classification counts must not be presented as the final filtered sample size for every metric. The AI category spans B2B and B2C products.
| Segment | Metric | Reported median |
|---|---|---|
| B2B | NRR | 82% |
| B2C | NRR | 49% |
| AI | NRR | 48% |
| AI | GRR | 40% |
These are billing-data observations from a particular platform population. Keep the annualization method and company classification attached to the figures when quoting them. Source: ChartMogul, The SaaS Retention Report: The AI Churn Wave.
Why these figures differ from a company survey
The 82% B2B NRR above and the 102% Benchmarkit NRR are not competing measurements of one identical population. The period, participating businesses, selection method and calculation approach differ. A billing platform’s customers are not a random sample of every SaaS business; a voluntary company survey is also selective.
Those differences justify caution. They do not, by themselves, establish how much of the gap comes from company size, financing, pricing or selection bias. Do not claim that struggling companies never answer surveys or that billing data eliminates every bias.
Choose the source whose documented population and method most closely resemble your own reporting. When neither is a close match, present both as context and make the limitation visible.
What customer-education research actually says
Forrester Consulting’s 2024 study, commissioned by Intellum, surveyed 300 US customer-education decision-makers in February-March 2024. Respondents reported a 22.3% increase in customer retention associated with their education programs. This is a reported relative improvement, not a 22.3-percentage-point change or a controlled causal estimate. Source: Forrester, page 15.
Do not substitute that result for your own renewal or cohort data. A survey of program decision-makers cannot tell an individual company how many accounts a new academy will save. It can support a hypothesis worth testing alongside product quality, commercial fit and service delivery.
An older TSIA survey, published in September 2017 with 2,800 respondents, found that 68% felt able to use the product more, 56% felt able to use more features and 87% felt able to work more independently after training. These describe respondents’ perceptions, not measured retention or observed product-use increases. Source: TSIA, 3 Ways Education Services Can Advance the Customer Journey.
For a practical measurement plan, connect customer education KPIs to a defined learner group, task outcome and account cohort. Record education exposure before the outcome being evaluated; otherwise the sequence of events may be impossible to interpret.
Definitions to keep beside every retention statistic
| Metric | What it measures | What it excludes or can hide |
|---|---|---|
| Customer or logo retention | Starting customers still active at the end, divided by starting customers | New customers; account size differences |
| Gross revenue retention | Starting recurring revenue retained after churn and contraction | Expansion; account-level losses can still be concentrated |
| Net revenue retention | Ending recurring revenue from the same starting customer group, divided by its starting revenue | New customers; expansion can offset losses elsewhere |
| Renewal rate | Eligible contracts or accounts renewed during a defined renewal window | Customers not due for renewal, unless the definition explicitly includes them |
| Product-use retention | Members of a defined user cohort who return or perform an event in a stated window | Accounts or revenue unless those are explicitly the unit |
Specify whether percentages are monthly, annual, trailing averages or annualized estimates. A chart labeled “retention” without a unit or period is not ready to reuse in a forecast.
Use the NRR calculation guide for account-level examples and consistent cohort handling. If a source’s definition differs from yours, show the difference instead of silently relabeling the number.
How to calculate a comparable internal result
Consider this hypothetical annual customer cohort:
- 200 paying accounts are active at the start.
- 20 of those accounts leave during the year.
- 40 new accounts arrive, of which five leave before year-end.
- The ending customer count is 215: 180 original accounts and 35 surviving new accounts.
Customer retention is 180 ÷ 200 = 90%. Subtracting all 40 new arrivals from the ending count would incorrectly produce 87.5%. Track the original customer group directly, or subtract only the new customers still present at the endpoint.
For revenue, suppose that the same original group starts with $100,000 MRR. By year-end it has lost $10,000 to full cancellations and $5,000 to downgrades, while adding $20,000 through expansion. Under this simple movement model:
- GRR: ($100,000 − $10,000 − $5,000) ÷ $100,000 = 85%.
- NRR: ($100,000 − $10,000 − $5,000 + $20,000) ÷ $100,000 = 105%.
These are illustrative calculations, not Corso client results. Document rules for account mergers, reactivations, currency conversion and usage-based revenue before comparing periods. Reconcile the result with the finance team’s underlying account records.
Statistics that need a narrower claim
“Increasing retention by 5% raises profits by 25-95%”
Do not treat a widely repeated range as a universal SaaS forecast. A retention change affects revenue over time, while profit also depends on margins, servicing costs, acquisition spending, discounting and the investment needed to change retention. The phrase “5%” also needs clarification: a relative increase and a five-percentage-point increase are different scenarios.
Build the scenario from your own economics. State the retained revenue, its contribution or gross margin as appropriate, incremental program costs and observation period. Do not count retained revenue and the full profit attributed to that same revenue as separate benefits.
“Training caused retention to rise”
A trained group may have started with more engaged accounts, better champions or higher contract values. A before-and-after comparison can also reflect product changes, pricing, seasonality or a different customer mix. Describe an association when that is what the design supports.
A randomized or carefully designed comparison can provide stronger evidence, but measurement requires attention to eligibility, exposure timing and follow-up. Small groups and incomplete renewal windows should remain visible in the report.
“This is the retention rate for our industry”
An industry label may combine annual enterprise contracts, monthly subscriptions, self-service customers and multiple product categories. A benchmark for consumer subscriptions should not be relabeled as enterprise SaaS. A median revenue-retention rate should not be presented as the percentage of customers retained.
A citation template for your own report
Use this structure when you reuse a statistic:
[Research organization], [publication and year], [specific metric and value], [measurement period], [population and metric sample if available], [method or material limitation], [link to the relevant page or table].
For internal decision-making, add your matching metric definition, current cohort size and the reason this benchmark is relevant. If the sample count is unavailable, say so. Do not borrow the total report sample when the metric uses a smaller subset.
Keep the original source link with a copied table. A third-party roundup is useful for discovery, but the primary publication is where a reader can verify the number and its context. An inaccessible or paywalled source is not automatically false; it is simply unsuitable for an unqualified claim that readers can verify from that page alone.
Turn the comparison into an investigation
Use the benchmark to locate a question: which accounts left, what revenue was lost, which outcomes were missing and whether a learning problem contributed. Then choose an intervention that addresses the observed reason. More training will not repair a missing product capability or an uneconomic contract.
The customer retention strategy guide connects those findings to practical next steps. If customers cannot perform an important workflow, Corso can help build and test the instruction around that task. Discuss the workflow with Corso.