Ecommerce chargebacks are payment reversals a customer’s card-issuing bank forces onto a merchant, not refunds the merchant agrees to give — the disputed amount leaves a Shopify Payments or Stripe balance the moment the bank files the case, before the merchant has said a word about whether the claim is true. The mechanism sits with the bank and the card network, Visa or Mastercard, not with the merchant or even the payment processor moving the money. What an operator actually deals with is the fallout: a fixed processing fee charged immediately, a short window to submit evidence, and, if the case is lost, the merchandise that already shipped gone along with the money.
What Actually Changes for an Operator When Ecommerce Chargebacks Hit?
The operator-level consequence of an ecommerce chargeback is not the fee on the invoice — it’s that money leaves the account before any person at the brand has decided whether the claim is legitimate, and the merchandise underneath that money is usually already in the customer’s hands. What’s left is a deadline rather than a choice: the only lever a brand has at that point is contesting the case with evidence before the window closes, not preventing the reversal itself.
Ownership of the problem inside the company changes as a direct result. A support team fielding a “where’s my order” email can check a tracking number and close it in one reply. A chargeback shows up as a dispute status on an order, not a support ticket, and it needs someone who can pull shipping confirmation, an IP address, delivery photos or a subscription agreement and assemble them into a formal response before a deadline that arrives with the case itself. Most brands at the $3M+ revenue floor this article is written for do not have a person whose job this specifically is; it lands on whoever in ops or finance happens to open the payments dashboard that week, which is a large part of why the deadline gets missed more often than a case gets lost on its actual merits.
Chargeback volume itself becomes the second operational change, once a brand runs any recurring-revenue programme. A brand fighting one chargeback a month can treat every case individually and still keep up. A brand running a subscription programme through Recharge, Skio or Stay AI, where a customer who forgot they were charged files a dispute instead of just cancelling, accumulates chargebacks as a rate rather than a series of one-off incidents — and it is that rate, not any single case, that a card network eventually measures and acts on.
What Does an Ecommerce Chargeback Actually Cost Per Order, Beyond the Flat Fee?
An ecommerce chargeback costs more than the flat $15 processing fee that most pricing pages quote as the headline number — the real figure has to weigh in the merchandise lost on every case a merchant loses and the staff time spent assembling evidence on every case regardless of outcome, at that merchant’s own win rate, not an average one.
Two named platforms show why the fee alone understates it, and by how much depends on which one a brand runs on. Shopify Payments charges a flat $15 chargeback fee in the US the moment a dispute is filed, and refunds it if the merchant wins (Shopify’s own chargeback-fee documentation). Stripe, run directly rather than through Shopify Payments, charges a separate $15 “dispute received” fee that is not refunded even on a win, plus an additional $15 “dispute countered” fee only if the merchant contests the case — and that second fee is refunded on a win, while the first never is (Stripe’s own dispute-fee documentation). Two providers, the same $15 headline number, and a materially different answer to whether winning gets it back.
The fee is also the smallest of three cost components once merchandise and labour are counted. On a lost case, the bank has already returned the transaction amount to the cardholder, and the goods that shipped against that order are gone — there is no return process for a chargeback the way there is for an ordinary refund. On every case, won or lost, someone has spent time pulling delivery confirmation, IP logs or a subscription agreement and formatting it into a one-shot response. That labour and that lost-merchandise value need to be multiplied by a merchant’s own win rate, not a borrowed industry-average one, because no processor or card network publishes a representative win rate for ecommerce disputes.
What follows is an invented, illustrative example, not a measured average — the inputs are assumed for the sake of showing the method, and every row is worked from them:
| Line | How it’s calculated | This month |
|---|---|---|
| Disputes filed | given | 40 |
| Assumed win rate | given, invented | 35% |
| Wins | disputes × win rate | 14 |
| Losses | disputes − wins | 26 |
| Fee cost (Shopify Payments, refunded on a win) | $15 × losses | $390.00 |
| Lost merchandise value | $85 average order value × losses | $2,210.00 |
| Evidence-assembly labour | 25 minutes × $28/hour loaded rate × 40 disputes | $466.67 |
| Total monthly cost | sum of the three cost rows | $3,066.67 |
| Cost per dispute, win-rate-adjusted | total ÷ disputes filed | $76.67 |
At this illustrative 35% win rate and $85 average order value, the win-rate-adjusted cost lands at $76.67 a dispute — more than five times the $15 fee alone, which is exactly why a brand that budgets only for the fee line underestimates its real exposure by that same multiple. A brand with a 60% win rate and a $40 average order lands on a different total entirely; the fee itself never moves, but it stops being most of the number once merchandise and labour are counted next to it. Digital goods change one row of this table, not the method — a lost dispute on a digital product has no shipped merchandise to lose, so that row drops to zero and the fee-plus-labour total is what the win rate is actually being weighed against.
The actual win-rate-adjusted cost per dispute for a specific catalogue, average order value and win rate is — metric to confirm — because that combination differs for every brand and nobody publishes a representative figure across them. Running the same four-row calculation against a brand’s own dispute log and average order value, rather than borrowing a range from a vendor page, is what turns the number from a guess into a fact a brand actually owns.
How Do You Submit Evidence for a Chargeback in Shopify Payments?
A Shopify Payments chargeback is submitted through the same order the dispute is attached to, not a separate chargebacks page — open the order, filter by chargeback and inquiry status if it needs finding, and Shopify surfaces the case, its reason code and its deadline directly on that order (Shopify’s own chargeback-management documentation).
The deadline varies from 7 to 21 days after the chargeback or inquiry is filed, and Shopify states it as a range rather than a single number because the actual window is set by the card network the transaction ran on, not by Shopify — the due date shown on that specific order is the one that governs, not a general rule of thumb applied across every case. Evidence has to be a PDF, JPEG or PNG file; Shopify caps each individual file at 2MB and the combined evidence for a case at 4MB, so a full-resolution phone photo of a signed delivery slip can use up most of that combined limit before a second document is even attached. A merchant can submit early by selecting “Submit now,” but the trade-off is real: evidence can’t be edited or added to after an early submission, the same one-shot rule Stripe applies to its own dispute-response form.
For a “product not received” dispute specifically, Shopify includes AI-generated insights in the evidence package by default, pulled from the store’s own delivery and fulfilment data — a merchant can opt out, but the default is on. This is one of the few places a chargeback response is partly automated rather than assembled by hand, and it is specific to that single reason code; a “not as described” or “unauthorised” dispute still needs the same manual evidence-gathering it always has, because the underlying claim is different and the automation only covers one of them.
Two outcomes follow from the submission. A won chargeback returns the full disputed amount to the merchant’s balance, along with the fee if that provider refunds it on a win. A lost chargeback leaves both amounts debited, and Shopify may publish an issuer response document explaining the bank’s reasoning — useful for spotting a pattern across future cases, less useful for the case it explains. Stripe’s own documentation is explicit that a merchant “can’t appeal a chargeback decision or submit additional evidence after a decision has been made,” and Shopify Payments runs on the same finality.
What Chargeback Ratio Do Visa and Mastercard Actually Enforce?
Visa enforces a 1.50% chargeback-and-fraud ratio against a merchant’s card-not-present volume as of 1 April 2026, and Mastercard enforces a 1.50%–2.99% chargeback ratio combined with 100 to 299 monthly chargebacks before escalating a merchant into its own monitoring programme — both are numeric thresholds measured monthly, not vague warnings a merchant only hears about after the fact.
Visa’s Acquirer Monitoring Program (VAMP) calculates its ratio as reported fraudulent transactions plus total disputes, divided by total settled card-not-present transactions. The merchant-level “excessive” threshold tightened from 2.20% to 1.50% on 1 April 2026, and Visa separately monitors the acquiring bank itself, at 0.50% (“above standard,” effective 1 January 2026) and 0.70% (“excessive”) — so an acquirer can face its own consequences even when no single merchant it processes for individually crosses the line, which is part of why an acquirer sometimes drops a merchant proactively rather than wait for a network penalty. Mastercard’s Excessive Chargeback Program (ECP) instead calculates its ratio as chargebacks received in a given month divided by sales processed the prior month, and enrols a merchant at two tiers rather than one.
| Visa Acquirer Monitoring Program (VAMP) | Mastercard Excessive Chargeback Program (ECP) | |
|---|---|---|
| Ratio formula | (fraud reports + disputes) ÷ settled transactions, card-not-present only | Chargebacks this month ÷ sales the prior month |
| Merchant threshold | 1.50% (from 1 April 2026; was 2.20%) | Excessive Chargeback Merchant: 1.50%–2.99% ratio AND 100–299 chargebacks |
| Higher tier | Not tiered the same way at merchant level | High Excessive Chargeback Merchant: 3.00%+ ratio AND 300+ chargebacks |
| Exit condition | metric to confirm | Below the lower tier’s thresholds for 3 consecutive months |
Crossing either threshold does not itself close a merchant account — it moves the cost of the dispute rate from a per-case fee to a standing one. Both networks pass a monitoring cost down through the acquiring bank once a merchant is enrolled, and it’s the acquirer’s own agreement, not the network’s public rulebook, that states what that cost actually is and when it can lead to account termination. The exact fine schedule at each tier is — metric to confirm — neither network publishes a public rate card for it, and every specific dollar figure circulating for it traces back to a chargeback-recovery vendor’s own blog rather than to Visa or Mastercard directly, which is why none is stated here as a fact.
Where Do Operators Get Ecommerce Chargebacks Wrong?
Operators most often get ecommerce chargebacks wrong by treating the win rate as free money rather than a rate that still has to clear the labour cost of fighting every case, including the ones that lose anyway. Fighting every dispute regardless of amount or reason code costs the same evidence-assembly time whether the order was worth $18 or $180, and a brand fighting all of them at a 20% win rate is spending eighty percent of that labour on cases that were never going to succeed.
The second mistake is not separating reason codes before deciding how to respond. A dispute coded as fraud, where the cardholder says they never authorised the charge, needs evidence of identity and delivery. A dispute coded as “product not received” or “not as described” needs shipping confirmation and a description match instead, and submitting fraud-type evidence against a not-received code wastes the single submission a merchant is given.
The third mistake is watching the raw chargeback count and ignoring the ratio the networks actually enforce. A brand growing fast can double its order volume and its chargeback count together and still be improving on the metric Visa and Mastercard measure, while a brand with flat order volume and a slowly rising chargeback count looks unremarkable on raw numbers and is heading toward a monitoring programme regardless.
The fourth mistake is treating a subscription dispute like a first-order one. A friendly-fraud chargeback on a recurring charge looks identical to a stranger’s stolen-card dispute in a queue of cases sorted only by reason code, but it needs entirely different evidence — the accepted subscription agreement and a run of prior undisputed charges on the same card, not an IP-address match on a single transaction. Sorting disputes by subscription status before assigning them, not just by reason code, is what keeps that distinction from getting lost in a shared inbox.
How Is an Ecommerce Chargeback Different From a Refund or an Inquiry?
A refund and an inquiry both surface next to a chargeback in the same order-status view, but only a chargeback carries a fixed fee and a fixed evidence deadline the moment it is filed. A refund is money a brand sends back by its own choice, with no fee and no deadline attached to it. An inquiry is a question raised before either applies — Stripe’s own documentation describes it as a pre-dispute stage, “typically issued when an account owner doesn’t recognize a transaction,” and responding at that stage can resolve the question before it escalates into a fee-bearing dispute at all.
Shopify’s admin groups a chargeback and an inquiry under one “chargeback and inquiry status” filter, because an inquiry that goes unanswered becomes a chargeback and the same order needs the same attention either way; a refund never appears in that filter, since a refund the brand initiates has no dispute state to track in the first place.
PayPal draws a similar distinction with an extra stage the card networks don’t have, because PayPal sits between the merchant and the bank rather than being the bank itself. A “dispute” in PayPal’s Resolution Center is the customer attempting to reclaim funds directly through PayPal; a “claim” is that dispute escalated for PayPal itself to investigate and decide (PayPal’s own documentation on disputes, claims and chargebacks). A chargeback is what happens after that, when the customer goes around PayPal’s Resolution Center straight to their card issuer, and the case moves out of PayPal’s process into the network’s entirely. A merchant who resolves the PayPal-level dispute or claim never reaches a chargeback at all, which is one more reason a PayPal dispute deserves as fast a response as a formal chargeback, not a slower one.
None of the three terms describes the same event, and a support team that considers a case settled after issuing a goodwill refund can still watch the disputed amount leave the account anyway — a refund does not cancel a chargeback already in motion, because the two are separate transactions moving in the same direction for entirely different reasons.
None of this is a payments problem once the volume is more than the odd one-off case — it’s a fraud-and-chargebacks systems problem, the same way a rising failed-payment rate is a systems problem rather than a customer-service one. A capable brand can win most of the cases genuinely worth fighting; the money gets lost in the ones nobody classified correctly, the deadlines nobody tracked against a calendar, and the evidence nobody assembled before the one-shot submission closed. That is the class of work we build fraud and chargeback response systems for — reason-code routing, evidence assembly pulled from the systems that already hold the answer, and deadline tracking that doesn’t depend on someone remembering to open the payments dashboard that week.
Sources
The dispute-response process, deadline range, fee structure and appeal finality are drawn from Stripe’s own published documentation. The Shopify Payments chargeback fee, evidence file requirements, submission workflow and AI-generated evidence detail are drawn from Shopify’s own Help Center. The Visa Acquirer Monitoring Program thresholds are reported by the Merchant Risk Council, an independent payments-industry association, and are labelled as reporting on Visa’s own rule change rather than as Visa’s direct publication. The Mastercard Excessive Chargeback Program thresholds are drawn from Braintree’s (PayPal’s) own developer documentation of the network programme its merchants operate under. The PayPal dispute-claim-chargeback distinction is drawn from PayPal’s own published definitions. No per-order win rate, average order value or network fine schedule is quoted as a fact, because none of the three is published by a processor or card network in a form specific enough to state as a single number; each is marked in the body as an item to calculate from a brand’s own numbers instead.