The four-character clue everyone skips
When a Stripe dispute lands in your inbox, your eye goes straight to the amount and the deadline. You start gathering everything — the order, the emails, the shipping confirmation — and you upload it all, hoping that volume wins. It rarely does. Buried in that same notification is a short string of characters, something like 10.4 or 13.1 or 4853, and that string is the single most important thing on the screen. It is the reason code, and it is not a label. It is the rulebook the issuing bank will use to judge everything you submit.
Most merchants treat the reason code as background noise. The ones who win read it first, because the code tells you exactly which question the bank is asking — and a brilliant answer to the wrong question loses every time.
What a reason code actually is
A chargeback is not a freeform complaint. When a cardholder disputes a charge, their bank has to file that dispute under a specific, pre-defined category defined by the card network — Visa, Mastercard, American Express, Discover. Each category has a code, and each code carries its own definition, its own evidence requirements, and its own time limits baked into the network's operating rules.
Visa overhauled its system a few years ago under an initiative called Visa Claims Resolution, sorting every dispute into four buckets: Fraud (codes beginning 10), Authorization (11), Processing Errors (12), and Consumer Disputes (13). Mastercard uses four-digit codes for a similar set of categories — 4837 for no cardholder authorization, 4853 for a cardholder dispute, 4834 for a point-of-interaction error. Stripe surfaces whichever code the issuer assigned, sometimes translated into plainer language like "fraudulent" or "product not received," but the underlying network code is what governs the case.
Think of the code as the venue. It decides which court you're arguing in, which means it decides which evidence is even admissible.
Why the same evidence wins one case and loses another
Here is the part that trips people up. Imagine two disputes for the same $200 order. One comes in as a fraud code — Visa 10.4, "Card-Absent Environment," meaning the cardholder claims they never authorized the purchase. The other comes in as 13.1, "Merchandise/Services Not Received," meaning they admit they bought it but say it never arrived.
For the 13.1 case, your shipping confirmation and delivery tracking are the whole ballgame. Proof the item reached the cardholder's address effectively ends the argument. But upload that same tracking number against the 10.4 fraud claim and it does almost nothing — because the cardholder isn't saying the package failed to arrive. They're saying they never made the purchase at all. Delivery to an address proves the box moved; it doesn't prove the right person clicked buy.
To win a 10.4, you need a different kind of proof entirely: evidence linking the actual cardholder to the transaction. A matching billing and shipping address, a verified IP and device, a history of prior undisputed purchases from the same account, an AVS or CVV match at checkout. Visa even built a specific provision — often called Compelling Evidence 3.0 — for showing a pattern of prior legitimate activity from the same customer.
Same order, same dollar amount, same merchant. Two completely different winning arguments, and the only thing that told you which was which was those few characters in the code.
A field guide to the codes you'll actually see
A handful of reason codes account for the overwhelming majority of disputes a typical online business faces. Learning to recognize them on sight changes how fast and how well you respond.
Fraud (Visa 10.x, Mastercard 4837). The cardholder denies authorizing the charge. Your job is identity and authorization: prove the genuine cardholder transacted. Lean on AVS/CVV results, device and IP data, login records, and any prior purchase history tied to the same person.
Merchandise or services not received (Visa 13.1, Mastercard 4855). The customer acknowledges the purchase but says nothing arrived. Tracking, delivery confirmation, and — for digital goods — access logs, download timestamps, or login activity are your evidence.
Not as described or defective (Visa 13.3/13.5, Mastercard 4853). The customer received something but claims it didn't match what was promised. Here your product descriptions, screenshots of the listing, terms of service, and any support correspondence showing you tried to resolve it carry the weight.
Canceled recurring transaction (Visa 13.2). The customer says they were billed after canceling. Your cancellation policy, the timestamp of their cancellation request (or absence of one), and your records of what they agreed to at signup are what matter.
Credit not processed (Visa 13.6, Mastercard 4860). The customer expects a refund they say never came. If you already refunded, proof of that refund usually resolves it. If you didn't owe one, your refund policy does the work.
Authorization and processing errors (Visa 11.x and 12.x). These are more technical — a charge made after an authorization was declined, a duplicate charge, an incorrect amount. They're often won or lost on transaction records rather than customer behavior.
Read the code, then read the clock
The code doesn't just shape your evidence — it interacts with the deadline. Card networks give merchants a tight, non-negotiable window to respond once a dispute is filed; on Stripe that's typically around a week, and the exact count depends on the network and the issuer. Miss it and you forfeit by default, regardless of how strong your case was.
That's why reading the code early is a practical advantage, not an academic one. The moment you know you're facing a 13.1, you know to go pull tracking immediately. The moment you see 10.4, you know tracking won't save you and you start assembling identity evidence instead. The code lets you skip the scramble and gather the right things on the first pass, while the clock still has time on it. Merchants who don't read the code often waste two of their seven days assembling evidence that was never going to count.
The mistake that quietly costs the most
The most expensive habit in dispute response isn't laziness — it's submitting a generic packet to every case. Networks and issuers are increasingly automated in how they review evidence, and a response that doesn't speak directly to the reason code's specific question reads as noise. You can attach ten documents and still lose because none of them addressed the actual claim being made.
The discipline is almost embarrassingly simple: find the code, look up what it means, and ask one question — what would convince a stranger that this specific claim is wrong? Then submit that, and only that, organized so the reviewer can see it in seconds. Precision beats volume, every time.
Where this gets handed off
The trouble is that decoding a reason code, mapping it to the right evidence, pulling that evidence from your records, and filing it cleanly — all inside a one-week window — is a real job, and it tends to arrive on your busiest day. That's the gap Argeback is built to close. It ingests the Stripe dispute, reads the reason code for you, drafts an evidence-backed response tuned to what that specific code requires, and files it before the deadline — from your phone, while you get on with running the business. The codes stop being a foreign language, and the deadline stops being a threat.
If you'd rather spend your week building than deciphering Visa's rulebook, you can see how it works at https://argeback.lumenlabs.works.