The dispute says merchandise not received. But three weeks ago — you remember this — the customer emailed you. "It arrived Tuesday, love the color. Quick question about the strap." You answered the question. They thanked you. And then, a month later, they told their bank the package never came.
That email is not a nice-to-have. It is the case. A single timestamped message in the customer's own words, contradicting the claim they later made to their bank, does more work than pages of policies and screenshots. Yet most merchants never submit it — because they don't think of their support inbox as evidence, and by the time a dispute lands, the thread is buried under a thousand newer ones.
This article is about treating customer communication as what it actually is: the most persuasive category of chargeback evidence you own, and the one the card networks explicitly ask for.
The dispute is a memory. Your inbox is a record.
Here's the structural problem with every chargeback: it's filed late. Card network rules give cardholders a long window to dispute — up to 120 days from the transaction under most reason codes. So the claim a bank receives is rarely a fresh account of what happened. It's a reconstruction, assembled weeks or months after the fact, often prompted by a statement review or a budgeting app flagging an unfamiliar line item.
Psychologists have understood since Frederic Bartlett's work in the 1930s that memory doesn't replay events — it rebuilds them, and the rebuild bends toward whatever story we currently believe. A customer who has genuinely forgotten a purchase doesn't recall a blank; they recall not buying it. A customer who's since soured on your product doesn't remember a fine delivery with a later disappointment; they remember it arriving broken. None of this requires dishonesty. Reconstructive memory smooths the timeline into whatever the present feeling suggests must have happened.
Your inbox doesn't do that. An email sent on March 4th says what the customer thought on March 4th, frozen at the moment it mattered. When the dispute narrative and the contemporaneous record disagree, a reviewer has an easy choice — and it's not the memory.
Why records beat recollection in front of a reviewer
Courts have a longstanding preference for contemporaneous documentation — records made at the time of the events — over after-the-fact testimony, precisely because timing removes the opportunity for memory drift and motivated reshaping. Chargeback review isn't a courtroom, but the analyst at the issuing bank applies the same instinct. They're comparing two stories: the cardholder's dispute claim and your response. The story anchored to dated artifacts wins.
The networks have codified this. Visa's and Mastercard's evidence guidelines explicitly list correspondence with the cardholder among acceptable compelling evidence — for fraud claims, communication showing the cardholder engaged with the purchase; for service claims, threads showing the issue was raised and addressed. Stripe's dispute form has a dedicated customer_communication field for exactly this material. The infrastructure is waiting for your inbox. Most merchants just leave the field empty.
There's a second force at work, borrowed from the law of evidence: a statement that cuts against the speaker's own position carries unusual weight, because people don't say things that hurt their case unless those things are true. "It arrived Tuesday" costs the customer nothing in March. In April, it demolishes their dispute. The bank knows the customer had no reason to lie then — which is precisely what makes the message credible now.
Matching the thread to the claim
Not every email helps every dispute. The move is to find the message that contradicts the specific claim in the reason code.
Fraud / "I didn't authorize this." You're proving the cardholder participated. An email from the same address used at checkout, referencing the order — asking about sizing, requesting an invoice, confirming a delivery window — shows a person engaged with a purchase they now claim never to have made. Even a mundane "thanks!" ties the human to the transaction.
Product not received. The gold standard is the customer acknowledging arrival in their own words. Short of that: a delivery-day thread with no mention of a missing package, or a question that only makes sense if the item is in their hands.
Not as described / defective. The claim implies you failed to make it right. A support thread where you offered a replacement, a fix, or a refund — especially one the customer ignored or declined — reframes the dispute: this isn't a merchant who shipped junk and vanished, it's a customer who skipped the resolution you offered and went straight to their bank.
Subscription cancelled. The claim is "I cancelled and you kept billing." Your records either contain a cancellation request or they don't. A support history with no cancellation email — or one dated after the charge in dispute — is quiet, devastating evidence. So is the confirmation you sent when they signed up.
Excerpt, don't dump
A reviewer spends minutes on your case, not hours. Forwarding a forty-message thread and hoping they find the good part is how strong evidence loses.
Instead, quote the decisive lines in your written response — with the date, the sender address, and one sentence of framing: "On March 4, 2026, the cardholder wrote from the email address used at checkout: 'It arrived Tuesday, love the color.' This directly contradicts the claim that the merchandise was not received." Then attach the full message as backup, headers visible, with anything irrelevant redacted. You've done the reviewer's work for them: claim, contradiction, source, date. That's the whole argument in four lines.
One caution: never trim a quote in a way that changes its meaning. If the same thread contains a complaint, address it head-on rather than hoping nobody scrolls. A reviewer who catches selective editing discounts everything else you submitted.
Making your inbox evidence-ready before you need it
The merchants who win these disputes aren't luckier — their communication habits produce evidence as a byproduct.
Keep support on channels that log. Email, a ticketing system, in-app chat with exportable transcripts. The reassuring phone call that resolved everything leaves nothing behind; follow it with two lines in writing: "Great talking just now — confirming we're sending the replacement to the same address." That sentence is future evidence, and the customer experiences it as good service.
Connect identities. Evidence only works if you can prove the person emailing you is the person who paid. Use the checkout email as the support identifier where you can, and keep order numbers in subject lines or ticket fields so the thread links to the charge without guesswork.
And close loops in writing. When an issue resolves, say so on the record: "Glad the replacement sorted it out." If the customer replies with anything positive, you now hold a dated statement that the problem was fixed — the exact message that ends a "defective merchandise" dispute before it starts.
None of this is legal theater. It's ordinary, humane support that happens to leave a paper trail. The trail is the point.
The part that actually goes wrong
Here's the honest failure mode: the evidence exists, and it never gets filed. Stripe gives you a short window to respond — miss it and you lose by default, email or no email. Finding the right thread, matching it to the charge, framing the contradiction, and submitting before the deadline is exactly the kind of task that slips when you're busy running the business. That's the gap Argeback closes. It ingests your Stripe disputes the moment they arrive, drafts an evidence-backed response built around the strongest material — including the customer communication that wins these cases — and files it before the deadline, all from your phone. Your inbox already holds the winning argument; Argeback makes sure it actually reaches the reviewer. See how it works at argeback.lumenlabs.works.