Does EU law require a cancel button? For the statutory withdrawal right, yes, and recently. Since 19 June 2026, Article 11a of the Consumer Rights Directive has required traders to provide an electronic withdrawal function, a clearly labelled button available throughout the withdrawal period, in the same online interface where the contract was concluded. Its organising principle is easy to state and awkward for a certain kind of business model: withdrawing from a contract should be no more burdensome than entering one.
Recurring revenue is the most expensively valued line in e-commerce. A buyer applies a multiple to it precisely because it is supposed to represent demand that persists without being re-bought. Churn is the number that certifies this, and churn is a measurement of how many customers left. It is not a measurement of how many wanted to.
This is the exit-side companion to the support-obstruction piece on the same GP2 layer, and the specific case of the asymmetric-exit pattern catalogued in the dark-patterns synthesis. Where those describe friction as a cost suppressant, this one describes it as a metric fabricator, applied to the metric a multiple is paid on.
The Mechanism
Subscription friction is rarely a single obstacle. It is usually a sequence, each step individually defensible and cumulatively decisive:
- Channel asymmetry. Subscribing takes one click in the app or on the site. Cancelling requires an email, a phone call during business hours, or a support ticket, with the outbound path deliberately absent from the interface that took the payment.
- The retention gauntlet. Cancellation is nominally available, but routed through sequential offers, surveys, and confirmations, each an opportunity to abandon the attempt.
- Negative-option conversion. A free or heavily discounted trial converts to full price automatically, with the reminder either absent or timed so that noticing requires attention the customer has stopped paying.
- Renewal-window traps. Annual terms cancellable only within a narrow window, so a customer who decides to leave in month two must remember to act in month eleven.
- Dormant billing. Continued charges to customers whose usage went to zero long ago, a population that is revenue on the P&L and absence in reality.
Each of these converts a customer’s intention to leave into a customer who is still paying. In the reporting, that is indistinguishable from a customer who wanted to stay.
Where This Sits in EU Law: A New Button, an Old Prohibition, and a Distinction That Matters
The new obligation. Directive (EU) 2023/2673 inserted Article 11a into the Consumer Rights Directive, applicable from 19 June 2026. Traders must provide an electronic withdrawal function that is prominently displayed, clearly legible, easily accessible from the online interface used to conclude the contract, and continuously available throughout the withdrawal period. The function runs in two steps: an initial action expressing the intention to withdraw, then a confirmation. The directive’s title refers to financial services, which has misled some traders into assuming it does not reach them; the withdrawal function it inserted operates horizontally, across distance contracts concluded through an online interface where a statutory withdrawal right exists. Labelling matters as much as existence: the wording must signal the exercise of a statutory right rather than the start of a customer-service process, so framing the action as exploratory, a “review” or a “check”, carries its own compliance risk.
The distinction that does most of the work. Withdrawal and termination are not the same right, and conflating them produces bad diligence. The withdrawal right is the 14-day statutory cooling-off period after conclusion, and that is what Article 11a’s button serves. Cancelling an ongoing subscription in month seven is contractual termination, governed by the contract terms and by national law rather than by Article 11a. Some member states legislate that separately, Germany’s cancellation-button requirement being the most prominent example. A target can therefore be fully Article 11a compliant and still operate an exit path that takes a customer four screens and a phone call in month seven.
The prohibition that reaches the rest. That residual territory is not unregulated. Article 9 of the Unfair Commercial Practices Directive treats disproportionate non-contractual barriers imposed on a consumer seeking to exercise a contractual right, including terminating or switching to another trader, as a factor pointing toward an aggressive commercial practice. The support-obstruction analysis applies directly here: an exit engineered to be effortful is the same conduct as a support channel engineered to be unreachable, aimed at a different moment in the relationship.
Where this is heading. The Commission’s Digital Fairness Act names subscription cancellation and dark patterns among its targets, with a proposal expected during 2026. Nothing in it is law, and its scope is not yet knowable, so no valuation should rest on it. What the sequence establishes is direction: the withdrawal button arrived in June, national cancellation rules preceded it, and a dedicated regime is being drafted. The regulatory current runs steadily toward symmetry between joining and leaving, and a retention number that depends on asymmetry is positioned against it.
What Suppressed Churn Does to the Model
The distortion compounds rather than sitting in one line. A suppressed churn rate lengthens the assumed customer lifetime. A longer lifetime raises lifetime value. A higher LTV improves the ratio against acquisition cost, which makes the marketing engine look more efficient than it is, and the same inflated retention feeds the recurring-revenue base that the multiple is applied to. One manufactured input propagates through unit economics, cohort analysis, and valuation together.
Three routes lead to normalisation, and each removes the mechanism rather than the customer’s underlying intent. The acquirer’s own counsel reviews the flows and requires symmetry. A national authority acts, whether through a sweep, a consumer body, or a competitor. Or the target implements the exit properly to satisfy Article 11a and finds the same interface work makes the whole exit path easier. When any of these lands, the customers who wanted to leave do so, in a cluster rather than a curve. The churn rate does not drift upward; it steps. The revenue that departs was, in the ordinary sense, never retained, but it was certainly valued.
There is a second exposure worth pricing separately. Where a negative-option conversion was never validly disclosed, or where the required withdrawal function was absent, the issue is not only prospective churn. It raises the question of refunds and unwound charges across an affected population, which is a contingent liability sitting behind the revenue rather than a margin adjustment ahead of it. Sizing that requires legal assessment per market; noticing that it might exist requires only looking at the flow.
Why This Slips Through Due Diligence
Churn arrives as a number in a data pack, calculated by the target from its own systems, and a number cannot show what a customer had to do to be counted in it. Cohort analysis inherits the same defect: it faithfully describes behaviour produced by an interface nobody examined. Legal diligence reads the subscription terms, which typically state a cancellation right correctly, because the friction lives in the implementation rather than in the text. And the target’s own team will describe the retention gauntlet as retention management, a discipline with vendors, conferences, and case studies behind it.
The gap is that no workstream owns the cancellation path as an object of inspection, even in deals whose entire thesis rests on recurring revenue.
What Outside-In Analysis Can Detect Before the Data Room
- Subscribing to the service and then attempting to cancel, recording the number of steps, the channels required, and whether the exit is available in the interface that took the payment
- Checking whether an Article 11a withdrawal function exists at all, how it is labelled, and whether it is accessible from the interface that took the payment rather than buried behind a login-and-search path
- Comparing the sign-up path with the exit path step for step, since the legal principle and the commercial signal are the same: asymmetry
- Reading the trial-to-paid conversion terms for how, when, and whether the consumer is reminded before the first full charge
- Reviews and complaint forums filtered for the vocabulary of trapped customers: “kept charging me,” “couldn’t cancel,” “had to call the bank,” “cancelled and was billed again”
- Checking whether cancellation is offered on all platforms the service is sold through, since app-store subscriptions and direct subscriptions often have different exit paths and only one may be governed properly
- App-store and marketplace review trends around billing complaints, which surface friction the target’s own channels never record
None of this establishes a breach, which is an assessment for counsel and for authorities. What it establishes is how much of the target’s retention depends on process rather than preference, and therefore how much of the recurring-revenue base would survive an exit path built to the standard the law is converging on.
The Pre-LOI Question Every PE Fund Should Ask
Churn is presented as a fact about customers. It is equally a fact about the cancellation flow, and only one of those is a durable asset. The question to put before the multiple is agreed: if cancelling took exactly as many clicks as subscribing, how many of these subscribers would still be subscribers in ninety days?
A target whose customers stay because leaving is easy and staying is worth it has recurring revenue, and the multiple belongs on it. A target whose customers stay because leaving is difficult has a retention number with a compliance countdown attached, and the acquirer is paying a multiple on the difficulty.
This analysis is part of Tronvik’s GP2 Fulfilment & Service Margin pillar. Nothing in this article constitutes legal advice. To initiate an outside-in subscription-friction screen on a specific acquisition target, contact info@tronvik.com.