I spend a lot of time inside subscription stores. The pattern that keeps showing up is this: the operator has spent twelve months tuning the acquisition funnel, three months tuning the welcome series, and about an afternoon on the cancel flow. The cancel flow is where the money is leaving. It is also where the single highest-leverage retention decision gets made, usually by a page that was built once and never revisited.
So when Recurly published its 2026 State of Subscriptions report in January and said in plain terms that pause-and-return had become a core retention lever rather than a nicety, I read it as a correction, not a trend piece. The industry has been treating cancel as a binary, and the people buying our work have been losing revenue to that binary for years without seeing it on a dashboard.
This piece takes the operational problem seriously. One expert’s framing, the numbers behind it, the levers that are moving, and the one that is not, and then what it actually costs to put pause before cancel inside a WooCommerce estate.
The expert framing: clearer choices, not lost interest
Joe Rohrlich, the CEO of Recurly, put the catch-22 on the record when the report landed: “Consumers have not lost interest in subscriptions. They are simply making clearer choices. That is leading to healthier long-term growth.” He said it in the 14 January 2026 announcement that accompanied the report, and then returned to it in the 15 January 2026 industry coverage: “The next chapter of subscriptions is about trust and relevance. When brands get that right, recurring revenue becomes more dependable.”
Read it twice, because the two sentences do a lot of work. The customer is not fatigued. The customer is managing their subscription portfolio in a way they did not two years ago, and most catalogues are not set up for that behaviour. Rohrlich’s co-author on the data, Brian Geier, Recurly’s VP of Business Intelligence, is more operational: “People are practical about subscriptions. When brands support flexibility, customers come back when the value is there.” That is the position I am going to test against the levers below.
Sizing the problem: the industry is not shrinking, it is sorting
The report draws on 2,200 subscription businesses and tens of millions of unique subscribers on the Recurly platform. The headline numbers that matter for a retention operator:
- Subscription growth has slowed to 12.6% and new acquisition sits around 3%. Nothing a retention team does will outrun that gap for long.
- Pause usage at top-performing merchants rose 337% year on year, and 75% of paused subscribers eventually came back.
- Nearly one in four new sign-ups is a former subscriber returning.
- Annual plans deliver 50 to 60% more revenue per user than monthly plans.
- Free-trial conversion has stabilised at 34%, down from 47% in 2021.
- More than 40% of consumers now say they are comfortable with AI managing parts of their subscription experience.
Nothing on that list looks like a trend article. Each is a lever a store can actually pull, or in one case, a lever a store cannot pull yet and should not pretend to. Taken together they describe a market that is not getting smaller. It is sorting the operators who understand the new behaviour from the ones still running 2022’s playbook.
Pause is the retention lever hiding behind cancel
Rohrlich frames pause as the fix for a measurement failure: a store that only offers “cancel” counts every “I will come back in two months” as churn, loses the customer’s address at the moment they wanted to pause, and then pays acquisition cost to re-engage someone who was never actually leaving. The 337% uplift in pause usage at top merchants is not a story about new behaviour. It is a story about operators finally offering the option the customer was already asking for.
The brands worth pointing at here are the ones that have taken this seriously for years. HelloFresh and Blue Apron built skip-this-week and multi-week pause into their cancel path on purpose, which is why their reported retention outperforms the acquisition slog they are running under. Audible famously keeps a pause option tucked behind cancel; it survives every pricing cycle because the churn figure it protects is enormous. The one brand that documents the counter-position in public filings is Bark, the dog-subscription operator: Bark counts paused subscriptions as churned in its own definition, and CEO Matt Meeker has told shareholders that “customer retention rates last quarter were at the highest level since going public” while using that stricter definition. His point is honest: how you count pause changes the number, so pick a definition and hold it.
Consumers have not lost interest in subscriptions. They are simply making clearer choices.
Joe Rohrlich, CEO, Recurly
Nearly one in four new subscribers is an old one
The report’s second lever is the quiet one. If 25% of new sign-ups are returning subscribers, your acquisition channel attribution is overstating cold acquisition by roughly that share, and your win-back programme is doing more work than your paid budget. Geier’s framing cuts through the attribution noise: flexibility makes return cheap. The customer who paused for six months comes back when the value returns; the customer who hit a hard cancel wall three years ago is still gone.
What this looks like in practice for a mid-market brand is unglamorous. Keep the ex-subscriber’s address segmented separately from the never-subscribed list. Treat the win-back email as a product, not a template, with its own quarterly review of subject line, offer construction and landing-page copy. Make the “resume where you left off” experience a two-click reactivation on a saved payment method rather than a three-step signup. Those three decisions move more revenue than most new acquisition experiments, and they almost always sit under a different team than the one measuring churn, which is part of why they get neglected.
Annual billing earns the attention it costs
Fifty to sixty per cent more revenue per user is not a small figure. The temptation is to treat annual billing as a discount mechanism and run a once-a-year 15% off upsell. The report’s data says that is the wrong framing. Annual customers behave differently: they engage with onboarding more deeply, they expect fewer transactional emails, they are more tolerant of a feature pause but less tolerant of a service outage, and when they renew they renew at a materially higher rate than a monthly resub.
The uncomfortable half of this lever is renewal risk. An annual customer who leaves takes twelve months of margin with them, so the renewal moment carries more weight than any single monthly decision. Recurly’s own data on this is specific: payment-recovery automation inside the annual renewal window recovers meaningful revenue that would otherwise walk out as involuntary churn. More on that in the next section.
Failed-payment recovery is a retention line, not an ops cost
Rohrlich’s team puts a number on this one that is hard to ignore. Software merchants on the Recurly platform recovered more than $155 million in revenue through automated payment-recovery tools in 2025, and digital media operators recovered close to $100 million through similar workflows. That is not optimisation at the margin. It is a line of defence that most WooCommerce stores still run manually, through CSV exports and dunning emails that fire on a schedule rather than on card issuer behaviour.
The right place to put this on an operations plan is as a retention metric, not a finance metric. If 0.9% of your monthly subscribers are churning involuntarily on a card failure, your retention programme is funding your card-network partners’ failure rates. Recurly reports that top-performing merchants recovered a 53% share of failed monthly payments and a 23% share of failed annual renewals when the automation was configured end to end. Those are real recovered subscribers, not statistical noise.
The lever that is not working yet: free trial conversion
Format D rules require that I name the lever that is not pulling, because every honest operational post has one. Here it is: free-trial conversion has stabilised at 34%, down from 47% in 2021. Thirteen points of conversion have evaporated across the industry in four years, and the Recurly data does not pin it to a single cause.
My read, from inside the builds, is that free trial is now a two-sided problem. Consumer behaviour has shifted toward portfolio management, so a trial is more often an evaluation than an intent-to-buy signal, which pulls conversion down. At the same time, operators have piled on friction at the credit-card capture step to filter the drive-by trial, which pulls activation and later conversion down with it. Nobody has shipped a trial design in the last two years that reliably reverses the slope. The honest answer today is that free trial is a weaker lever than it was and should carry proportionally less of the acquisition budget, not that free trial is broken.
What this costs to ship on a real WooCommerce estate
Here is where we earn the piece. Recurly operates its own billing layer, so a merchant on Recurly inherits most of the above as platform behaviour. On a WooCommerce estate running WooCommerce Subscriptions, the levers above are a build, not a toggle, and the cost of that build is predictable if you plan it as one project.
The honest baseline matters. WooCommerce Subscriptions does not expose a customer-facing pause action out of the box: the wc-on-hold status exists on the admin side and the extension model allows pause as a store-manager action, but a self-service “pause for six weeks” flow is a build. That is also true of the self-service dashboard improvements, where third-party packages hook into ssd_pause_subscription and sibling actions to give customers the lever directly. If you want to publish pause-before-cancel to your store this quarter, you are commissioning that build; you are not configuring an existing switch.
A workable scope for a mid-market store with roughly 2,000 active subscribers looks like this. Four weeks of engineering to add a customer-facing pause action with a configurable window (one, three or six weeks) on the My Account subscription view, plus the admin-side reporting change so pause does not get counted as churn in the renewal cohort. Two weeks on the win-back pipeline: a segmentation query against subscriber lifecycle status, a Mailchimp or Klaviyo flow for ex-subscribers at 30, 60 and 120 days, and a reactivation landing page on the saved payment method rather than a fresh checkout. Two weeks on the failed-payment side: pre-dunning email timed to known card-cycle behaviour, a retry schedule configured on the action_scheduler queue rather than a fixed cron, and an admin dashboard surface so the retention lead sees recovered revenue weekly instead of discovering it at quarter end.
Eight weeks of engineering, roughly, plus two weeks of analyst and copy work on the lifecycle emails, plus one line of invoice we are increasingly explicit about: the AI usage for drafting and iterating win-back copy itself, metered honestly rather than hidden inside a day rate. That ships pause, win-back and recovery for a store that today runs none of the three.
WordPress is the right implementation layer for this because the entire chain is already instrumented. Customer records are first-party, segmentation runs against a database you own, the Subscriptions extension’s hook surface is enough to carry the self-service dashboard, and Action Scheduler is more than sufficient for the retry mathematics. The alternative, a hosted billing platform plus a thin content layer, is cleaner on day one and more expensive on day 365 because the customer data lives somewhere you rent.
The part that is not yet answered
Rohrlich’s framing is strong on what to build. It is quieter on what to measure. If pause-and-return is a lever rather than a cancellation, the churn metric needs to carry a reconciliation window: a subscriber who paused on 1 October and returned on 15 December was not churned, but the standard month-end calculation treated them that way for two reporting cycles. Meeker’s point at Bark is the companion warning, pick a definition and hold it. Our own position is to carry two: gross cancel and net-of-return at a six-month window, both published to the retention team each month. Neither is the number on its own.
If you are evaluating a platform, or a partner, on the strength of its retention story, this is the question I would put to them: show me the pause state in your data model, show me the win-back pipeline you built last quarter, and show me the recovered-revenue line on a monthly dashboard. If the answer involves exporting a CSV at any step, that is the state of the practice, not the state of the art. The gap between those two is where the next two years of subscription revenue is sitting, waiting for someone with a build team to pick it up.
That is the build team conversation we are having with clients right now, and it is the one I would happily have with you.
Last modified: October 15, 2026
