# A chargeback management system runs on records kept early

> A chargeback management system gathers evidence, decides which disputes to counter and tracks wins. The cases we studied point to records kept in advance.

Clawnify Resources · https://www.clawnify.com/resources/chargeback-management-system · 2026-09-25

## The five jobs of chargeback management software
A chargeback management system is the software, and the habits around it, that carry a card dispute from the moment your processor reports it to the moment it closes, and then learn from how it ended. People say chargeback management software when they mean the tool on its own, and chargeback management solution when a service that argues cases for you comes bundled in. Strip the labels off and it's five jobs.

- Intake. The processor tells you, by webhook or in its dashboard, and the clock starts. Stripe's documentation puts the response window at 7 to 21 days in most cases.

- Decide. Counter or accept, one dispute at a time, on the amount, the fees and the reason code.

- Assemble. Pull the order, payment, delivery and usage records into one file a stranger at an issuing bank can follow.

- Submit and confirm. Stripe's documentation says you get one opportunity to submit, so make sure the file is complete before it goes, then check that it arrived.

- Learn. Log each outcome by reason code and let it shape the next decision and your prevention.

Vendor pages mostly talk about job three and sell a win rate. The cases we researched kept pointing at jobs two and five instead. Those two decide whether the system pays for itself, because they control how much you spend fighting and how many disputes you invite next quarter. Job three still matters. It just only works when the records it pulls from existed before the dispute did, and a file thrown together from inboxes and screenshots in a panic isn't that.

One fact from Stripe's documentation frames the rest: the cardholder's bank decides the outcome, and Stripe has no say in it. Your processor carries the evidence. Someone else reads it. So we'd judge any chargeback system on two things: what it lets you show that reader, and what it lets you decide before a reader is involved at all.
## A record of use argues better than a folder of proof
Three businesses in our research did the natural thing when a dispute landed. They gathered every document they could find and sent the lot.

BusinessWhat was submittedResultNew business, first big sale of $55,000Signed delivery, photo proof, six months of useLostMarketing agency, 12+ yearsContract, texts and emails, links to two 40+ page sitesLost $20,000Service firm, two-month engagementDelivery emails, signed non-refundable agreement, client praiseLost $5,000, plus over $1,000 in legal fees
Two of them sold services. Their files argued that the work was agreed and delivered, and both owners built those files by hand after the dispute arrived. The agency owner was facing his first chargeback in over twelve years of running the agency, after five months of SEO work and two websites, and said he assumed he'd win. The service firm's owner reported spending several hours on his file. The bank refunded his client anyway. The first case is harder to wave away. The buyer filed six months after delivery while still using the product, the founders' appeal carried delivery and usage documentation, and they lost too.

The case that went the other way belongs to an indie software maker. He said he used to ignore disputes and lost almost all of them. Then he built a responder. When Stripe's webhook reports a dispute, his app pulls the user's details, the sign-up date and, in his words, "most importantly what they did in the app" into a PDF. His first large win, he reported, was a $1,199 dispute from someone who'd used the app for months and generated thousands of photos.

One win against three losses isn't a benchmark, and the first row of the table shows proof of use can still lose. What caught our attention is the kind of evidence. His file rested on a log the product kept while it ran, and it answered the exact question the claim raised: did this person use what they paid for? Nobody had to reconstruct that log. It was already there. Stripe's guidance leans the same way for card-absent fraud claims. Under Visa's Compelling Evidence 3.0 rules, Stripe checks your transaction history for eligibility, and says an eligible dispute typically has a significantly higher chance of being overturned.

That's the slice of job three a chargeback management system can actually own. It can't make an issuer agree with you. It can keep the record of what the customer did as it happens, and get it into the file in a form a bank can read.
## A won dispute still counts against you
Winning feels like the end of a dispute. Your account sees it differently. Stripe's documentation says all disputes, won or lost, count toward your dispute rate, and the $15 fee charged on receipt in the US isn't returned. A win returns the disputed amount and the separate $15 countering fee, but it does not return the count.

An ecommerce operator we followed put it bluntly to newer merchants: disputes filed count, win or lose, and that number decides your reserves, your fees and whether you keep your processor at all. He also claimed Visa cut its limit to 1.5% in April. We haven't confirmed that figure, and his argument doesn't need it. His advice was to stop pouring hours into evidence packets and spend them making sure the dispute never gets opened.

The software maker from the last section ended up somewhere close. Four months after his first big win, he reported that adding self-serve refunds inside his app had brought his dispute rate down. The responder won him disputes. Self-serve refunds meant there were fewer to win.

That's why the decision job starts before any chargeback exists:

- Answer inquiries. Stripe says an inquiry can be resolved without a dispute fee, and one left unanswered can escalate into a chargeback you're likely to lose.

- Act on early fraud warnings. Stripe says about 40% of Visa and Mastercard warnings become fraud disputes if the charge isn't refunded, and suggests refunding when the charge is roughly at or below the dispute fee. A refund stops the dispute. Stripe notes it doesn't remove the warning itself, which Visa still counts in its fraud monitoring.

- Be recognisable. The operator's list starts with a billing descriptor people recognise and a tracking email sent the moment an order ships.

- Make refunds easy. A customer who can get their money back without asking has one less reason to call the bank.

Refunding everything is the wrong answer too. Stripe's documentation says auto-refunding every early fraud warning isn't a good strategy, and that a refund is likely not worthwhile once the charge is more than about 35% above the fee. It comes down to one payment at a time, with the amount, the warning and the customer's history in front of you. That's the same record the evidence file draws on, consulted earlier.
## Clean numbers don't stop a freeze
So far we've treated the dispute rate as the thing to protect. It is. But the stores in our research show it won't protect you on its own. Each of these owners reported a low dispute rate or none at all, and each still had funds held, payouts blocked or payments shut off.

StoreReported dispute rateWhat happenedShopify store, 2,500+ orders since July0.16%$37,000 held for 120 daysShopify store, about 6,000 orders0.02% (one chargeback)Payouts blocked, no answer after documents sentFast-growing store, seven figures in 60 days0.01%Funds locked, no explanationShopify store0 chargebacksPayments disabled, appeal deniedStripe startupNo disputed chargebacksAccount closed after documents sent
We can't tell from the outside whether any of these reviews was fair. What the cases do show is how the owners pushed back. They argued from numbers a chargeback management system keeps: order counts and dispute rate, and in one case refund rate and response and shipping times. Every rate in the table is far below the 0.75% that Stripe's documentation calls excessive by industry standard, although Stripe also says a sudden spike can draw a network's attention before that line.

The 2,500-order store's appeal read like a monthly report. The owner put his 0.16% chargeback rate next to the 0.65% threshold he cited, which is lower than Stripe's figure. He gave a 1.5% refund rate and said it was deliberate: he refunds unhappy customers early, and in his view that's why disputes stay low. He said he answered customers within 24 hours and shipped within 24 hours with tracking. Then he asked for a human re-review, and for a partial release of funds against orders already delivered and undisputed.

That last request only works if you can show, order by order, which were delivered and which were disputed. None of these owners stopped a freeze with their records. What a record gives you is something to hand over when one happens, and it has to exist before the request for documents arrives. The days after a freeze are the worst time to start building it.

The fast-growing store's owner took a second lesson from his freeze. Set up a backup processor while you don't need one.
## Price each dispute before you counter it
Say the owner of a store that takes payments through Stripe comes to us after a month of dispute fees. Suppose there were a dozen disputes, most under $50 and a few in the low hundreds, and every one got countered out of habit. The first thing we'd look at is the arithmetic on each.

Stripe's pricing page lays it out for US disputes opened after 17 June 2025. There's a $15 fee when the dispute arrives, and you don't get it back. Countering costs another $15, and that one is returned if you win. The first fee is gone either way, so only the second belongs in the decision. Countering pays when your win rate for that kind of dispute beats the countering fee divided by the disputed amount plus the fee.

Disputed amountWin rate needed to break even$1060%$5023%$2007%$5003%
Staff time comes on top, so the real bar sits higher than the table. And the table is for Stripe accounts only: Shopify's help centre describes its own chargeback fee, which it may refund on a win depending on the region. These fees pile up faster than owners expect. A founder selling in France reported losing €300 in Stripe dispute fees in one day, and Stripe's page lists the EU and UK fees as 20 to receive and 20 to counter, in local currency.

In our imagined store, the small disputes are where the money leaks. Whether a bar of 23% or more is reachable depends on a number few dashboards show: the store's own win rate by reason code. Picture it at one win from three disputes of one type. A system that prints that as 1/3 instead of 33% is being straight with you about a sample too small to plan around.

Then the record. The disputes worth countering are the ones where order, payment, delivery and activity records already exist and match the reason code. Outsourced chargeback services commonly take a share of every dispute they recover. When the evidence comes from records you already hold, that's a recurring charge on your own data, and for most stores we think it's the wrong trade. We built chargeback evidence automation for Stripe and Shopify on that premise, and OpenDisputes is an open-source template that stages evidence for your review by default rather than submitting it automatically.

The chargeback management system worth having is the one whose record you'd trust on two particular days: the day a dispute lands, and the day your processor asks why it should keep your account.