Subscription Chargebacks Are Lost at Sign-up

The records that win a subscription dispute were captured at sign-up, or never.

Subscription businesses lose a disproportionate share of chargebacks, and they lose them for a specific reason. The argument the issuer wants to read on a Visa 13.2 (cancelled recurring transaction) or a Visa 13.7 (cancelled merchandise or services) dispute is not a story about what happened at the renewal. It is a story about what happened at sign-up, and the records that decide it were either captured then or do not exist now.

The cardholder who disputes a subscription renewal usually tells the issuer some version of the same story: they cancelled and were charged anyway, they never really signed up, or the renewal terms were not what they thought they had agreed to. The merchant has 30 days to rebut, and the rebuttal letter has to draw from records that captured the sign-up itself: what the cardholder clicked, what they were shown, what cancellation path was disclosed, when the trial-to-paid conversion was scheduled, and at what price. A merchant who built their subscription flow without those records in mind is not arguing against the cardholder. They are arguing against silence.

The strongest sign-up captures for representment are unambiguous about what the cardholder agreed to. Price and billing cadence sit in the same view as the consent action rather than scattered across separate screens of the funnel, the cancellation path is named in a single sentence the cardholder can find later, and a free-trial conversion to paid is treated as a separate consent action with the conversion date and the converted price stated alongside the click that authorizes it. Each of these is the cost of being able to argue, 14 months later, that the cardholder knew what they were signing up for.

Trial-to-paid is the single most common failure point. A trial that auto-converts to a paid plan after 14 days, where the buyer was shown the conversion language only inside a long terms document at sign-up, almost always loses at representment. The issuer reads the buyer's claim that they did not understand the conversion as plausible because the merchant's own records show the conversion was disclosed but not surfaced. The same trial, where the buyer was required to acknowledge the conversion date and amount as a discrete checkbox or click-through, almost always wins. The mechanical difference between the two flows is small. The gap in representment outcomes is large.

For Visa 10.4 (card-absent fraud) and the Mastercard 4837 equivalent, the SaaS evidence stack looks different from a physical-goods seller's. There is no shipping address to verify against the billing address. The proof that the legitimate cardholder was responsible has to come from the platform's audit log: the account creation timestamp matched to the cardholder's IP and device fingerprint, the welcome email delivery and any subsequent opens, the login pattern in the days and weeks after sign-up, and the cardholder's actual usage of features the legitimate buyer would have used. A SaaS or course business that does not log this data, or logs it but cannot retrieve it on demand at dispute time, is unable to argue 10.4 cases on the kind of evidence the issuer is equipped to read.

The cancellation log is the equivalent failure point for Visa 13.2. A subscription business that offers cancellation through a single documented path (a settings page, a specific email address, a chatbot flow) and logs every cancellation attempt against that path has a defensible record when a buyer claims they cancelled. A business that accepts cancellations through any channel that reaches a human, without a unified log, has to reconstruct the cancellation history from email archives, ticket systems, and expert memory at dispute time. The reconstruction is almost always incomplete, and the gap is what the cardholder's account fills.

The operational discipline that wins subscription chargebacks is unglamorous. Consent records have to be captured at the moment of sign-up in a format that survives a year, every cancellation request has to flow through a single logged funnel, trial-to-paid conversions need surfacing as their own consent actions, and usage data has to be retained at a granularity the dispute team can actually query when a notification arrives. None of this is visible to the buyer who never disputes, which is most buyers; it is the cost of being able to argue the small fraction of disputes that arrive, on the records the issuer is asking for, at the moment the issuer is asking.

The merchants who lose subscription disputes by default are not the ones with bad letters. They are the ones with sign-up flows that did not anticipate that one in a hundred renewals would arrive as a dispute 14 months later, with a buyer whose memory of what they agreed to is now adversarial. The case was built or lost at sign-up, and the letter that arrives 14 months later is downstream of that decision.