The Great Subscription Audit

01 · The Audit Arrives From Two Directions

In 2026, Adobe agreed to pay $150 million to settle a federal case over how it treated customers who tried to leave. The complaint described the mechanics plainly: fees disclosed late, and a cancellation process routed through unnecessary steps and delays designed to wear the customer down. Adobe was not penalized for a bad product. It was penalized for making the exit hard.

That same year, 47 percent of consumers canceled at least one subscription, up from 31 percent two years earlier. People are combing their statements and finding charges they no longer recognize. The average household guesses it spends around $86 a month on subscriptions. The real figure is closer to $219. That gap is the quiet audit now underway at millions of kitchen tables, amid growing financial pressure.

Subscription businesses are being audited from two directions at once. Regulators are coming for the friction that makes leaving hard. New York City has adopted a rule, effective October 1, 2026, requiring that canceling be as easy as signing up, the first city in the country to do so. New York and California tend to lead on consumer legislation while other states follow, so one city today is a fair signal of where the rules are heading. From the other direction, customers are coming for the subscriptions that no longer serve them.

Planet Fitness has long counted on the member who signs up in January, stops going by February, and keeps paying anyway. To cancel, that member cannot simply click. In many cases the company requires cancellation in person or by mailed letter. The model depends on the customer not auditing, and on the exit being just annoying enough to postpone. That works until the customer runs the audit and the letter goes out. It is one reason the brand has faced class-action lawsuits over its billing and cancellation practices.

02 · Two Kinds of Retention

Two companies can post the same retention number and have a completely different brand relationship architecture. In the first, customers stay because the relationship earns it. They get more value the longer they stay, they would lose something real by leaving, and many of them recommend the brand without being asked. In the second, customers stay because leaving is a hassle. The contract auto-renews, the cancellation process is a maze, the switching cost is high, and they find it frustrating to get out.

At the far end of that second kind is extraction, where the brand does not just hold the customer in place but steadily pulls more value out of the relationship than it puts in. It is spreading into places you would not expect. In the auto industry, you can finance a car and still find capabilities built into it, remote start, driver assistance, added performance, sitting behind a monthly fee. When BMW charged a subscription to warm seats already installed in the car, the backlash was swift enough that the company backed off. The wider practice of gating pre-installed hardware has since drawn legislation in several states, where critics call it “subscription extortion.”

On the financial statements, both kinds of retention look healthy. The revenue recurs, the numbers hold. Underneath, one relationship is compounding, and the other is a reservoir held back by a wall. It looks stable until the wall is taken away.

Earned retention survives price increases, competitors, and a bad experience, because the relationship is worth keeping.

See how the Four Pillars tells earned retention apart from friction, alongside the metrics you already run.

03 · The Audit

Adobe is retention built on friction, made visible by a court. The friction was not a byproduct of the software. It was a designed feature of the exit. The product was excellent. The relationship, at the moment a customer tried to leave, was adversarial.

HP sits further along the same spectrum. It is not just making it hard to leave, it is almost daring customers to. Its firmware, which HP calls Dynamic Security, rejects third-party and refilled cartridges, and the company has defended the practice as protecting customers from malware, a risk security researchers have called practically nonexistent. A printer that stops working when a subscription lapses, or refuses ink the manufacturer did not sell, is not passive friction. It is aggressive extraction that accepts some customers leaving, because the ones who stay are worth more. The customer owns the hardware but rents the ability to use it. That is a dependency strategy, not a loyalty one.

It is fair to ask why this should matter to the CMO. Retention is retention, and a leader measured on it has every reason to protect it, friction included. The answer is timing. Friction retention is a liability that has not yet come due.

Disney offers a quieter tell, specific to streaming. A subscriber signs up for one series, watches it to the end, and cancels, because the relationship was never with Disney, it was with the show. When the service lost subscribers, the response was to discount them back. A brand with genuine relationship architecture rarely has to buy its customers back. The discount reveals what the retention was made of.

The console makers belong in the same conversation, moving ownership itself toward a license.

None of these brands is failing. Each shows strong numbers, and each has quietly moved a portion of its retention from earned to engineered, from a relationship the customer values to an arrangement the customer tolerates.

04 · The Number Moves Last

Consider a brand moving the other way. In its most recent quarter, Verizon reported its lowest consumer churn in years and its strongest subscriber additions in five, after pulling back from the handset subsidies that carriers have long used to hold customers in place. Its CEO framed the shift directly: the company is “earning long-term retention based on real value rather than subsidized promotions.” Whether that holds is a fair question, but it is worth watching.

Adobe shows the reverse. It built on friction deliberately, and for now it still works. Subscriber numbers are climbing. But the market has already cut Adobe’s value by more than half from its peak, not because customers have left, but because it can see how many alternatives now exist and how easily the switch can be made. Tools like Figma and Affinity, the latter now free for individuals, did not exist as credible replacements a few years ago. The retention number is still going up. The confidence underneath it is going down.

The friction that has been holding weak retention in place is being taken away from three directions now. Regulators are making the exit easier. Customers are auditing what they pay for. And competitors are offering the same value without the trap, the way Figma and Affinity did to Adobe. Most brands will not choose the moment they find out what their retention was made of. They will find out the hard way, when the friction is gone and the customers they were counting on are already leaving.