Luca Barberis

How does each cohort decay over time?

№ 182 · Retention curve by cohort

01

Definition

MetricShare of a cohort's starting ARR still retained at month N

Unit% of cohort starting ARR, plotted by months since contract start

Percent of subscribers retained at month N per signup cohort. Watch the M1 cliff (trial-to-paid leakage). Build from subscription event data, counting active subs per cohort per month. Step changes follow content launches. Example: Jan cohort 78% at M1, 62% at M6; Apr cohort 84% at M1 (improving).

02

Benchmarks

Bottom 30%MedianTop 30%
80% at month 12, roughly 64% at month 2488% at month 12, roughly 78% at month 2495% at month 12, roughly 90% at month 24

Private B2B SaaS, annual contracts, gross revenue retention measured on a cohort basis excluding all expansion. Month 24 figures are the month 12 rate compounded, which assumes a constant hazard and understates second-year risk in most books. · Month 12 values are grounded: Benchmarkit median GRR 88%, ChartMogul best-in-class B2B around 95%, KeyBanc and Sapphire 86% for 2023 recovering toward 90%. Month 24 is compounded rather than observed and should be treated as a floor, because ChartMogul's data shows churn tends to rise in year two once first-year expansion has been captured, so real curves usually bend down harder at the second renewal than a constant hazard predicts. Do not mix units on this card. Amplitude's product benchmarks put top B2B three-month retention at 15.6% against a 2.5% median, but that is monthly active users, not contracted ARR, and it belongs on a product retention card rather than a CS one. Cohorts also need enough accounts to be readable; below roughly thirty accounts a single logo makes the curve step rather than decay.

By category
CategoryBottom 30%MedianTop 30%
ACV under $10K (SMB)under 78%82% to 85%90% at month 12
ACV $25K to $100K (mid-market)82%88%93% at month 12
ACV above $100K (enterprise)87%91% to 93%96% at month 12
03

When it looks bad

The curve drops in a step at month 12 rather than decaying smoothly, and the step gets deeper with each newer cohort, meaning nothing is being lost during the term and everything is being lost at the renewal decision.

Cohort holds 98% through month 11, then falls to 81% at month 13 and 76% by month 15. The in-term line looks healthy all year, which is why the team reports green until the renewal window opens.

04

What to do about it

  • Plot adoption depth on the same axis as the retention curve for each cohort. Where a cohort's month 12 step is deep, the cause is normally visible in months one to three, and Amplitude's finding that 69% of products strong on early activation are also strong at three months holds the same direction for account-level adoption.
  • Separate the first-renewal cliff from ongoing decay by plotting cohorts against renewal number rather than calendar month. A book with a hard first renewal and a soft second is a qualification and onboarding problem; one that decays evenly is a value problem.
  • For card-billed cohorts, strip involuntary churn out of the curve before diagnosing. Recurly's network puts SaaS median annual involuntary churn at 1.06%, and on a low-ACV cohort that is a visible share of the month 12 step.
  • Where the newest cohorts sit below the older ones at every tenure point, stop working the curve and work the intake. Cohort quality is set before the CS team ever touches the account, and ChartMogul attributes much of the post-2021 retention compression to poor-fit customers acquired during the growth-at-all-costs period.
05

Sources

  1. Benchmarkit (with Pavilion)2025 B2B SaaS Performance Metrics Benchmarks · 2025 · n=583 private B2B SaaS companies, CY-2024 actuals; metric-level samples smaller (NRR n=228, GRR n=225, expansion ARR n=81, expansion CAC n=21) Median NRR 101% (down from 105% in CY-21, 103% in CY-22). Median GRR 88%, down from 90% over three years. Expansion ARR is 40% of total new ARR at median, 58% at $50M to $100M ARR, 67% above $100M. Expansion CAC ratio $1.00 versus new customer CAC ratio $2.00. Subscription gross margin 81% median. GRR and NRR both rise as ACV rises. 5242563.fs1.hubspotusercontent-na1.net ↗
  2. ChartMogulSaaS Retention Report 2023 · 2023 · n=2,100+ SaaS businesses, aggregated anonymised billing data Top quartile NRR by ARR band: 94% at $1M to $3M, 99% at $3M to $15M, above 105% at $15M to $30M. 41.1% of businesses with ARPA above $500 per month hold NRR above 100%, versus 2.7% below $10 ARPA. Best-in-class GRR is around 95% for B2B; 35.7% of businesses above $500 ARPA hold GRR above 85%. Top quartile customer retention is 90% at three months and 70% at twelve months, rising to 93% and 77% at $15M to $30M ARR. chartmogul.com ↗
  3. ChartMogulThe SaaS Retention Report: The New Normal For SaaS · 2024 · n=2,500+ SaaS businesses, aggregated anonymised billing data, H1 2021 to H1 2024 Expansion drives up to 40% of growth for companies at $15M to $30M+ ARR, versus 30% in early 2021. Even the top quartile at $15M to $30M+ ARR failed to reach 100% NRR in 2024. Companies at or above 100% NRR draw over half of added revenue from expansion; companies below 60% NRR draw 70% from new business and 15% from expansion. New customers expand most during year one as they ramp usage, and churn tends to rise in year two. Median churn in the low NRR bucket is 7%, double that of companies at or above 100% NRR. chartmogul.com ↗
  4. SaaS CapitalWhat is a Good Retention Rate for a Private SaaS Company in 2025? · 2025 · n=1,000+ private B2B SaaS companies, 14th annual survey, ARR above $1M Companies with ACV between $25,000 and $50,000 report median NRR of 102%, top quartile 111%, bottom quartile 97%. NRR correlates positively with ACV. Companies with the very highest ACVs report the highest GRR. NRR defined as December 2024 MRR from customers present in December 2023, divided by total December 2023 MRR. saas-capital.com ↗
  5. Amplitude2025 Product Benchmark Report, B2B technology cut · 2025 · n=2,600+ companies, behavioural data September 2023 to September 2024 At three months, top B2B products retain 15.6% of users versus 2.5% at the median, a gap above 6x. 69% of products strong on early activation were also strong three-month retention performers. For 90th percentile products, activation runs about 21% at day 1, 12% at day 7 and 9% at day 14. amplitude.com ↗
  6. KeyBanc Capital Markets and Sapphire Ventures16th Annual Private Company SaaS Survey · 2025 · private SaaS companies, published November 2025 Gross retention is expected to approach the 90% threshold after declining to 86% in 2023. Net retention has remained above 100% through the same period with modest improvement expected. YoY ARR growth expected to accelerate from 15% in 2024 to 20% in 2025. sapphireventures.com ↗
  7. RecurlyChurn rate benchmarks: SaaS, media, retail and more industries · 2026 · Recurly subscription network median annual rates, July 2026 Median annual involuntary churn across industries is 1.25%, ranging from 0.18% for high-ARPC businesses to 1.69% in higher-risk sectors. SaaS sits at 1.06%. Involuntary churn falls 87% from the lowest to the highest ARPC tier and is recovered through intelligent retry logic, automated dunning and account updater services. recurly.com ↗
  8. Vista Point AdvisorsHow to Improve SaaS Retention Rates (with Benchmarks and Averages) · 2021 · companies represented by the bank over the preceding five years Median gross revenue retention 88% with a long tail toward lower retention. Median logo retention 89%. Enterprise-focused software should reach at least 90% logo retention to be considered best in class. Logo retention behaves as a qualification checkbox rather than a valuation driver. vistapointadvisors.com ↗
  9. Drivetrain (citing RevOps Squared B2B SaaS Benchmarks)Logo retention rate: what it is, how to calculate it and why it matters · 2022 · RevOps Squared 2022 B2B SaaS Benchmarks, private B2B SaaS companies Median logo retention across companies is 85%, ranging from 89% at $1M to $5M ARR down to 83% at $50M to $100M ARR. Logo retention is most informative below roughly $10K per customer; above that, gross dollar retention is the better read. drivetrain.ai ↗