Seasonal churn: what the January purge data actually shows#

The January subscription purge is mostly a story. In Antenna's Premium SVOD panel, Q2 was the lowest-cancellation quarter of both 2024 (30.0M cancels) and 2025 (34.1M), and Q1 was the annual peak in neither year. Churnkey's seasonality analysis puts peak churn intent in July, up 47% on the May baseline, with November the healthiest month of the year. What is genuinely seasonal is acquisition: 31% of 2025 streaming gross adds landed in Q4.

Every December a wave of posts warns merchants to brace for the January cancellation wave. Almost none of them cite a cancellation number. This post collects what is actually published, says clearly where the data is missing, and ends with the only planning move the evidence supports.

The cancel data does not show a January cliff#

Start with the one dataset that publishes quarterly cancellation volumes. Antenna's State of Subscriptions: Premium SVOD 2025 Year in Review tracks ten US streaming services through an opt-in panel of consumer transaction records. Its quarterly gross adds and cancels are printed on the chart, so they can be read directly rather than inferred.

Q1 cancels are higher than Q2, but they are not the annual peak in either year. In 2025 the peak was Q3, and Antenna attributes that directly to a news event rather than a season: Disney+, Hulu and HBO Max together saw 8.9M cancels in September 2025 alone during the Jimmy Kimmel controversy. Antenna's own framing of Q2 is "an 8% reduction in QoQ Cancels", which is the opposite of the seasonal-panic narrative.

Category churn is stable across the calendar boundary too. Antenna's weighted average churn rate was 4.6% in December 2025 against 4.8% in December 2024.

The seasonal effect that is real runs on acquisition#

The number that moves hard with the calendar is not cancellations. It is signups. Antenna reports that Premium SVODs drove 31% of annual gross adds and 57% of net adds in Q4 2025. Quarterly share of gross adds for 2025 ran Q1 24%, Q2 22%, Q3 23%, Q4 31%.

Black Friday is a large single slice of that: 8.9M attributed signups in 2025, up from 8.1M in 2024, roughly 25% of all November signups and about 13% of December signups.

That reframes January entirely. If Q4 loads an outsized cohort onto your file, then January and February will carry an elevated count of cancellations without any change in the rate. The cohort is bigger, and early-tenure churn is always the highest churn. Churnkey's voluntary churn benchmarks put monthly churn at 12.0% for subscribers under three months against 3.2% for subscribers past twelve. A December cohort hitting its month-one and month-two decisions in January is the whole effect.

Merchants who track raw cancel counts see a January spike. Merchants who track cohort-normalized rates usually do not. Reading cohort retention curves covers how to separate the two.

Churn intent peaks in July, not January#

The only vendor dataset that publishes an explicit month ranking points the other way. Churnkey's seasonality analysis (July 2025), drawn from "over $2 billion in revenue and hundreds of millions of subscriptions across SaaS companies", reports that "July has the highest churn intent rates across SaaS companies. Churn intent jumped 47% compared to May baseline." It also names November as the healthiest month, with December through February a secondary high period.

Two honest caveats before you plan around that. The dataset is SaaS-weighted, so it does not automatically transfer to a coffee subscription. And the metric is churn intent measured as normalized cancel-flow sessions per 1,000 subscribers, not completed cancellations, so it counts people who looked at the exit and then took an offer.

Even discounted, it is the closest thing to a published month ranking in this space, and it says the summer slump is a churn event while the winter is a growth event.

Where January is genuinely a purge#

One category does show the classic pattern, and it is worth being precise about which half of the funnel it happens in.

Placer.ai's foot-traffic panel measured fitness visits in January 2025 at 21.2% above December 2024, following a January 2024 month-over-month jump of 23.4%. That is the resolution surge, and it is enormous. What follows is decay rather than a cancellation cliff: across the four brands Placer.ai analysed, the share of repeat visitors (at least twice a month) peaked in Q1 2024 and fell consistently through the year to a Q4 low.

Pew Research surveyed 5,140 US adults between January 16 and 21, 2024 and found three in ten Americans made at least one resolution, rising to 49% of 18 to 29 year olds, with 79% of resolution-makers naming health. Of those, 59% said they had kept all of them so far. Note the fielding window: that reading is two to three weeks in, so it measures early enthusiasm, not annual adherence. Pew did not re-field it later in the year, and the widely circulated "80% fail by February" figure has no traceable primary source. Do not repeat it.

The takeaway for a wellness, fitness or courses merchant: January is a January acquisition problem. You will add a large, weakly committed cohort whose churn lands in March and April.

Events beat seasons#

Two of the largest churn movements in the public record were not seasonal at all.

• The Q3 2025 streaming cancel spike above was a news controversy, worth 8.9M cancels in a single month across three services.
• Antenna's panel estimates for Netflix show US churn moving from 1.8% in December to 2.5% in January after the January 2025 price rise, decaying back to 2.0% by May. Read as an estimate from panel data rather than a company disclosure, and note that this is a January spike caused by a pricing decision the company made in January, not by the month.

If your own January looks bad, check what you changed in January before you blame the calendar.

The holiday-cohort trap#

The most useful and least repeated finding in the Antenna report is what happens to promotional cohorts on a delay.

2024 Black Friday signups retained 66% of subscribers through month six, a 10-point lead over other 2024 promotional signups, and Antenna's summary reports 9-month retention of 57% against 43% for 2024 signups overall. By month twelve, Black Friday signups and the overall category both land at 36%. Antenna attributes the early lead to promotion structure: Disney+ and Hulu ran 12-month discounts in 2024, HBO Max a 6-month one.

The cohort does not retain better. It retains later. The discount window holds people in place, and the churn arrives when the window closes. A merchant running a 6-month holiday discount has not avoided the January problem; they have scheduled it for June.

That is the mechanism worth planning around, and it is a duration decision rather than a depth decision. How deep should a save discount go covers the margin side of the same trade.

What the calendar actually does#

Where the public data is simply missing#

This post would be better with e-commerce numbers, and they do not exist publicly in usable form.

Recurly's 2024 State of Subscriptions and its churn benchmark research both slice churn by industry and by price point but publish no month or quarter analysis at all. Recurly's benchmarks do show education carrying the highest median churn of any industry at 4.99%, above ecommerce at 4.25% and SaaS at 3.22%, which is consistent with a school-calendar business being structurally leakier. That is a level, not a season, and it should not be presented as one.

Recharge's published reports were unreachable during this research. No platform we could reach publishes cancellations by calendar month for physical subscriptions. If you see a monthly seasonality chart for subscription boxes, check whether it has a named dataset behind it before you plan against it. Most do not.

How to plan around it#

Five moves the evidence supports, in order.

1. Normalize before you panic. Divide cancellations by subscribers at the start of the month and split by cohort tenure. If January's rate is flat and only the count moved, your Q4 acquisition worked and nothing is wrong. The churn rate calculator does the denominator correctly.
2. Set your promotional discount duration deliberately. The holiday cohort churns when the discount expires. Pick the expiry month on purpose and staff for it.
3. Do not stack a January retention discount on top of a December acquisition discount. You will be conceding margin twice to the same subscriber, and the Antenna data says the first concession was already carrying the retention.
4. Move save-flow attention to your own peak, whatever it is. For SaaS-shaped products the published evidence points at July. For resolution categories it points at month three of the January cohort. For most stores it points at whatever month follows your largest promotion.
5. Check what you changed. Price rises, packaging changes and renewal-reminder launches all produce month-shaped spikes that look seasonal in a year-over-year chart and are not.

The one thing not to do is treat the calendar as a reason. A season is a scheduling fact about when your cohorts arrive and when your discounts expire. It is never the cancellation reason, and the offer that saves a subscriber in July is the same offer that saves them in January.

What's next#

• WooCommerce churn benchmarks for the baseline rates that seasonal swings move around.
• Reading cohort retention curves for separating a cohort-size effect from a rate effect.
• The first 90 days for why a large December cohort produces a loud February.

Sources: Antenna, State of Subscriptions: Premium SVOD 2025 Year in Review (2026) and Antenna insights pages, accessed August 2026 · Churnkey seasonality analysis (July 2025) and voluntary churn benchmarks (November 2025) · Placer.ai, Fitness Starts Strong in 2025 (February 2025) · Pew Research Center, New Year's resolutions: Who makes them and why (January 2024) · Recurly churn rate benchmarks (July 2026 network data).