They bought it, got it, and disputed it anyway.
Most disputes are not stolen cards. They are real customers who did not recognise a charge, forgot a subscription, or found disputing easier than asking for a refund. Which is good news, because that is addressable.
What this page settles.
Three questions, in the order most people ask them.
What it actually is
A chargeback raised by a real customer, on a transaction they really made, that they really received.
Merchants expect disputes to be stolen cards. For most businesses they are not. The larger category by some distance is customers disputing their own purchases — and the reasons are mundane rather than criminal.
The term is unhelpful, because it bundles together a customer who genuinely did not recognise a charge on their statement with one who deliberately kept the goods and got their money back. Those are different problems. The first is a process failure on your side and is fixable this week. The second is a small minority and has to be fought.
Which is the useful reframe: most of your dispute volume is something you are doing, not something being done to you.
Why it happens
Four causes, in roughly the order they occur.
They did not recognise the charge. The single biggest cause. A customer sees a line on a statement a month after buying, the name is a holding company or an abbreviation or the name of a payment platform rather than your shop, and they call their bank. From their point of view they are reporting fraud. Billing descriptors.
They forgot a subscription. Recurring billing that stopped being noticed, often after a trial, often at a renewal. The charge arrives, it looks unfamiliar, it gets disputed. Subscription pricing.
They could not reach you. A customer who wants a refund and cannot find a phone number, or emails and gets no reply in three days, disputes instead. Their bank is responsive and you were not. This is the cause merchants most underestimate.
They decided to. Kept the goods, got the money back, knew what they were doing. Real, a minority, and the only part of this that is genuinely adversarial.
Note that three of the four are process problems. That is the opportunity.
The four controls that work
In order of return per hour spent.
Fix the descriptor. What appears on the customer’s statement should be the name they would recognise — the trading name, not the legal entity, not an abbreviation, not a payment platform’s name — plus a phone number if there is room. Every provider lets you set this and most merchants have never looked at theirs. One phone call, and for some businesses it removes a third of their disputes.
Make yourself easy to reach. A visible phone number and an email address answered within a day. A customer who can get a refund from you does not need to get one from their bank — and a refund costs you the sale, while a chargeback costs you the sale plus a fee of $15 to $35 plus a mark against your ratio. Being reachable is cheaper than being right.
Email before you charge, for anything recurring. A renewal notice three days before the charge, with the amount, the date and a one-click way to cancel. It feels like inviting cancellations. It is cheaper than the disputes, and far cheaper than a ratio problem.
Keep evidence as a matter of routine. Delivery confirmation, tracking, signature on higher-value items, the terms they accepted and when, and for services a record of usage. Evidence collected at the time wins disputes; evidence reconstructed afterwards usually does not. How to fight a chargeback.
A fifth, worth it at volume: pre-dispute alerts. Network services that notify you of a pending dispute and let you refund before it becomes a chargeback. Each alert costs a fee and avoids a chargeback fee and — more importantly — a ratio hit.
When to fight and when to refund
A commercial decision, not a moral one.
Refund when the amount is small relative to the chargeback fee and your time; the customer has a reasonable point; your evidence is thin; or you can catch it before it becomes a chargeback. A pre-dispute refund keeps the transaction out of your ratio entirely, which is worth more than the money on anything but a large sale.
Fight when you have real evidence — delivery confirmation, accepted terms, usage records, correspondence showing they received and used what they bought; the amount justifies the effort; or the pattern suggests a customer doing this repeatedly.
Always fight where losing would push you toward a monitoring threshold, regardless of the amount. At that point you are not defending the sale.
Remember you pay the fee either way. Most processors charge the chargeback fee whether you win or lose, so “winning” recovers the sale and not the fee. Chargeback fees and how they work.
The ratio, which is the real risk
The money lost to friendly fraud is usually survivable. The ratio is what ends accounts.
Both major networks monitor chargebacks as a ratio to transactions, and breaching the thresholds puts you in a remediation program with fees, scrutiny and — if it continues — termination. Visa’s VAMP merchant ratio dropped from 2.2% to 1.5% on 1 April 2026 (CEMEA remains at 2.2%). Mastercard’s ECM threshold is 1.5% with at least 100 chargebacks a month over two consecutive months; HECM is 3% with 300 a month. Confirm current figures with your acquirer, because these move. Chargeback monitoring programs.
Two consequences worth holding onto:
Refunding before a dispute is filed keeps it out of the ratio. A chargeback you prevented is worth more than one you won.
The ratio is per merchant identifier. If you run several locations under one MID, one site’s problem is everybody’s ratio. What is a MID.
And the end state is worth naming: an account terminated for chargebacks frequently comes with a MATCH listing, which blocks new accounts for five years. The MATCH list.
What does not work
Arguing that the customer is lying. Issuers decide on evidence, not on tone. A representment that reads as aggrieved and contains no documents loses.
Blocking the customer and hoping. The dispute is already filed; the bank does not care that you have blocked them.
Ignoring it. An unanswered dispute is a loss, plus the fee, plus the ratio.
Terms and conditions as a defence on their own. “Our policy says no refunds” does not win a dispute. Evidence that the customer agreed to a specific term, at a specific time, sometimes does — which is why the acknowledgement matters more than the policy.
Over-tightening fraud rules in response. Rejecting every AVS mismatch to stop friendly fraud blocks legitimate customers to prevent disputes that mostly have other causes. Fix the descriptor first. Selling online.
What we charge
Interchange + 0.50% + 10¢ in person, interchange + 0.75% + 25¢ keyed, with no contract, no monthly fee and no monthly minimum.
The highest-value thing we do on this topic takes five minutes and does not involve switching anything: look at your billing descriptor and tell you whether your own customers would recognise it. For a surprising number of businesses that single change is the whole dispute problem.
While we are on the subject
What is friendly fraud?
A chargeback raised by a genuine customer on a purchase they really made — because they did not recognise the charge, forgot it, could not reach you, or decided disputing was easier than returning.
Is friendly fraud really fraud?
Sometimes deliberate and often not. A customer who genuinely does not recognise a charge is not committing fraud, which is why the descriptor fix works so well.
Can I win a friendly fraud dispute?
Often, with evidence — delivery confirmation, the terms they accepted, usage records, prior correspondence. The win rate depends almost entirely on what you can document.
Should I just refund instead of fighting?
Below the point where the fee plus your time exceeds the sale, usually yes — and a pre-dispute refund avoids the ratio hit, which matters more than the money.