How are subscribers leaving each period?
№ 081 · Churn cohort waterfall
Definition
MetricMonthly gross subscriber churn
Unit% of subscribers cancelling per month
Start subscribers, minus voluntary churn, minus involuntary (payment failures), minus active termination, end subscribers. Read involuntary bar size as easy money on the table. Build by classifying each cancellation by reason per period. Example: 1M subs, -68K voluntary, -42K involuntary, -8K terminated = 882K.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| Above 6.5% monthly; specialty streamers have ranged to 9%+ | 4.5-5% monthly | Under 3% monthly (large established bases; Netflix runs under 2%) |
US premium SVOD, cancels in a month divided by subscribers at end of prior month, excludes free tiers and MVPD/telco distribution, 2023-2025 data (Antenna); corroborated by Recurly digital media network churn. · Definitions diverge sharply. Antenna measures monthly account churn; Recurly's July 2026 page labels its figures annual medians while its methodology historically reports churn measured monthly, so treat 4.14% as a monthly-equivalent with caution; Deloitte's 39-41% is the share of consumers cancelling any service in six months, not a monthly rate. Do not mix the three.
Category split omitted: The service-category churn split (premium vs specialty vs sports vs vMVPD) is published only by Antenna; Recurly's vertical split does not cover the same categories, so the two-source test fails.
When it looks bad
Cancel bars growing to nearly offset gross adds each period, with visible cancel spikes in the month after every tentpole or season finale.
100k gross adds against 85k cancels for 15k net; churn jumping to 9% the month after a finale versus a 4.5% baseline.
What to do about it
- Convert event sign-ups into habit before the cancellation window: longer seasons and weekly episodes rather than a binge drop; Q4 tentpoles drove 31% of premium SVOD gross adds and 57% of net adds in 2025 (Antenna), so the retention play is in the 8 weeks after
- Rebuild the cancel flow with pause and downgrade-to-ad-tier offers before confirmation; a saved downgrade retains most blended revenue as ad monetization scales
- Run structured win-backs against new season drops; return acquisitions are 27.7% of digital media acquisitions and cost less than cold acquisition (Recurly)
- Split the waterfall into voluntary and involuntary before acting; payment-failure churn responds to billing fixes, not content (see card 082)
Sources
- Antenna Weighted average monthly churn stabilized at 4.6% in 2025; flat or lower churn in 7 of 11 months vs prior year antenna.live ↗
- Antenna Monthly churn: specialty 6.6% (range 6.6-9.2% since 2023), sports 5.1% (range 4.4-12.0%), vMVPD 4.5% (range 4.1-7.1%) antenna.live ↗
- Antenna via Deadline Premium SVOD churn hovered around 5% since January 2023; Netflix consistently under 2% deadline.com ↗
- Recurly Digital media & entertainment total churn 4.14%, mid-pack among six verticals; education highest at 4.99%, SaaS lowest at 3.22% recurly.com ↗
- Deloitte 39% cancelled at least one paid SVOD in six months, above 50% for Gen Z and millennials; 24% churn-and-return deloitte.com ↗
- Deloitte Six-month churn stable around 40%; 61% would cancel their favorite service at a $5 monthly price increase deloitte.com ↗
- SubJolt Corroborates Recurly and Antenna ranges; digital media recovered nearly $100M of failed payments in 2025 (Tier 3 corroborator only) subjolt.com ↗