Automatic Renewal Can Cost Subscription Firms Subscribers

In May 2026, the Federal Trade Commission announced a $35 million settlement with Shutterstock over alleged subscription and cancellation practices. The case puts a familiar problem back in view for companies built on recurring payments: when renewal becomes easy for the firm and hard for the customer, the commercial gain can invite customer pushback and regulatory scrutiny. 

A new research paper by HEC Paris professor Klaus M. Miller and fellow scholars adds a more immediate risk. It studies what happens when a promotional subscription automatically renews into a paid plan, compared with a contract that automatically cancels at the end of the trial. Subscription businesses often treat that choice as a retention setting. The study shows that consumers treat it as part of the offer itself.

A field test of a small contractual choice

The researchers ran a large-scale field experiment with a major European newspaper. More than 1.4 million readers who reached the paywall were randomly offered promotional subscriptions. The offers varied across three dimensions: auto-renewal or auto-cancellation, two weeks or four weeks of access, and a trial price of zero or EUR 0.99. All other steps were held constant, including the information readers had to provide. The researchers followed subscription and usage behavior for more than 20 months.

The central result is clear. Auto-renewal reduced trial take-up by 35 percent. In other words, for every 100 readers who would accept an auto-canceling trial, only 65 accepted an auto-renewing one. Auto-renewal then produced the short-term effect many companies expect: among those who did take the trial, post-trial subscription rates were 20 percent to 38 percent higher in the first months after the promotion.

That early advantage faded. After about a year, subscription rates were higher in the auto-cancel group. Across the full observation period, auto-cancellation generated 23 percent more total paid subscribers. Readers offered auto-renewal were also 7 percent less likely to subscribe in the 20 months after the promotion.

Why consumers avoid the offer

The study's contribution lies in explaining this pattern. Many consumers are inert, meaning they tend to remain in a subscription even when they would prefer to cancel. The common managerial assumption appears to be that these consumers fail to anticipate that inertia. Our evidence points in another direction. The model estimates that 35% to 55% of consumers are non-inert. The remaining consumers are inert, with an 81% to 85% monthly chance of failing to cancel a subscription they would rather drop. Among those inert consumers, 83% to 92% are sophisticated in the paper's terminology: they know their own tendency to delay or fail to act.

That awareness changes the sign-up decision. A reader may value the trial yet still reject it after weighing the risk of paying for months because cancellation will be postponed. In this setting, auto-renewal retains some customers after signing up and filters others before sign-up.

Subscriber quality and subscriber count

The usage data deepen the finding. Readers retained through auto-renewal behaved like low-engagement subscribers. More than half of the auto-renewal subscribers in the post-promo usage comparison registered zero visits to the platform. Among those who remained subscribed, auto-renewal subscribers visited 62 percent fewer pages than post-promo auto-cancel subscribers.

For a digital publisher, that matters beyond subscription revenue. The study's conclusion notes that unique subscribers, reach, advertising revenue and engagement-driven growth can all be affected when auto-renewal reduces the extensive margin. In the experiment, auto-renewal generated higher revenue in the medium term, but the benefits eroded over time as the subscriber count and engagement picture weakened.

The price and duration tests help isolate the role of the renewal default. Raising the promotional price from free to EUR 0.99 reduced trial subscriptions by 9 percent, with effects fading quickly. Extending the trial from two weeks to four weeks had a modest and more persistent effect. The researchers estimate that a four-week trial led about 0.1 percent of trial-takers to learn that they valued the subscription enough to remain subscribed. In this setting, learning from the trial played a limited role compared with inertia and consumers' anticipation of inertia.

What the study changes

The research reframes auto-renewal as both a customer acquisition decision and a churn decision. The renewal default affects who accepts the offer, who avoids it, who stays, and how much they use the product. A subscriber retained through inaction can support short-term revenue while contributing little usage. A customer who refuses an auto-renewing trial may represent demand lost before the funnel records a conversion.

For managers, the practical implication is to test the renewal default over a long enough horizon. Short-term retention and revenue can favor auto-renewal. Total paid subscribers, later subscriptions and product usage may tell a different story. The relevant comparison is how many potential customers engage with the firm over time, and whether retained subscribers actually use the service.

For regulators, the paper adds nuance. Consumers are affected by inertia, yet many anticipate it and protect themselves by avoiding contracts that expose them to it. That awareness can limit firms' ability to profit from inertia. At the same time, a smaller share of consumers remains vulnerable to paying for subscriptions they would prefer to end. The policy problem starts before cancellation, when the renewal default shapes the choice to subscribe.

The subscription economy has made recurring revenue a central business goal. This study shows that recurring revenue depends on the first contractual term customers see. The renewal default is visible to consumers, and consumers respond before the first recurring charge arrives.

For more on the topic, go to this Decoding special featuring professor Miller.