October 2, 2026 · Disputed Team
We hear this question a lot on calls with prospects.
And yes, you can fight everything. Any dispute that lands in your queue can get an evidence packet and a response. We don't think you should fight every chargeback you get, and our reason is as much about how you want to treat your customers as it is about the math.
Let's walk through it.
Ideally, someone from your team would look at every chargeback and handle it the way that case needs. The problem is that at scale the bespoke approach stops being possible. A merchant receiving tens of thousands of chargebacks a month would need an enormous team to review each one.
This is why most in-house teams we've talked to choose to just fight everything. It's faster to use the same basic response, even if the customers who were right end up paying for it. And at the end of the day, you still probably get a positive ROI to show finance.
Even if you're using an outsourced tool, in our experience most of them just give you two options:
So, you end up in the same situation: it's just faster to fight everything.
The question we ask about every chargeback is, "Who is right? The merchant or the customer?" (We walk through our full method for handling disputes in how Disputed manages chargebacks.)
A lot of chargebacks are friendly fraud, and those deserve a fight. But not every chargeback is friendly fraud.
Say a customer tries to cancel their subscription and the cancel button throws an error. They email support, hear nothing for ten days, and the next renewal hits their card. You have the terms they accepted, the renewal reminder you sent and a login from last month. That's a decent evidence packet, and you might win. But the customer did everything you asked them to do, and the thing that failed was your cancel flow. A win would keep one renewal's worth of money and leave the broken button in place for the next customer who tries to leave.
True third-party fraud, where the card really was stolen, is the clearest case of the cardholder being right. If your own fraud screening already tells you an order was fraudulent, fighting the chargeback means arguing with the victim. Some victims give up because the amount isn't worth the effort, which looks like a win on your side but leaves the cardholder angry at your brand.
Fraud fighters tell a familiar story here: when a fraud victim loses a chargeback, the bank tends to point at the merchant, and the frustration ends up in social media posts about your brand. The fix for these is acting on your fraud screening before the order goes out.
This is why we believe that when the customer is right, you should accept the dispute and then find out why they went to their bank instead of coming to you. We've built our whole workflow around doing this without overloading a fraud team.
A chargeback is often a complaint that should have reached you first.
Maybe it went to your support team and didn't get resolved, or the customer couldn't find a way to reach you at all. Sometimes a refund was on its way, but it was going to take longer than a phone call to the bank. This is why your support team has to be faster than the bank.
The accepted cases deserve a closer look than the ones you fight. Was there a support ticket before the dispute? Did the customer try to cancel and fail? Was a refund already in progress? Does your billing descriptor match the name they know you by?
The causes people find when they dig in can be surprisingly specific. In one review, a consultant found a couple of support agents promising refunds and never issuing them, because processing the refund was too much work. A second consultant traced a run of "not as described" disputes to a single product, and the answer was a fuller description on the product page. Other causes come down to a customer who couldn't find a phone number, or didn't recognize the name on their statement.
Whatever you learn needs a way back to the rest of your team, or nothing changes.
A short weekly list works if every accepted dispute gets the same four columns: what happened, why, how to resolve it, and who owns the fix. Send it to whoever owns support and billing policy. Because the columns never change, the lists can be compared week to week, and the causes that keep coming back are easy to spot. The goal is for support to fix the problem before the next customer reaches for a chargeback.
Fight or accept is a decision you make one case at a time. The larger lever is the pattern across cases. When a run of disputes traces back to a single policy, billing descriptor or authorization flow, fixing it removes disputes you would otherwise have to decide on every month.
Preventing a dispute is far better than winning one. A won dispute still counts against your ratio, and it still started with a customer unhappy enough to call their bank.
This is the part of the work Disputed is built for. We trace dispute volume back to the policies, billing descriptors and authorization flows that generate it, and to who inside the company owns changing each one. Those patterns are hard to see from inside the queue, where every dispute looks like its own case.
You can copy and paste your way through a full queue with a generic evidence packet. The first problem with that is the cases where the customer was right, which a generic packet fights anyway. The second is where your team's hours go. Every hour spent on a case you were unlikely to win is an hour that could have gone into better evidence on a case you were likely to win.
Generic evidence also tends to be less persuasive. A response that doesn't tell a clear story for that specific case gives the reviewer less to work with.
Buyers raise this themselves. One asked us: "Are are you saying that, if we continue to take this route of just defending everything and sending a cookie-cutter type evidence packet, that our performance may suffer over the long run?"
We think the answer is yes. Generic evidence on every case means fighting customers who had a point, and it means thinner evidence on the cases that were winnable.
When a vendor runs your disputes, fighting every one doesn't cost your team time. It doesn't change your ratio either, because a won dispute still counts against it.
We'd still encourage you to have your vendor fight selectively, and to report back consistently on the disputes they chose to accept and why. If your vendor can't do that, it's worth asking them why. Those accepted cases are a direct signal about what's going wrong upstream: which policy, which product page, which part of the support experience pushed a customer to their bank. Use the accepted cases to look for how those disputes could have been prevented in the first place.
That's how Disputed works when it runs a merchant's disputes. Every case gets a recommendation to fight or accept, with the reason behind it. When the customer is right, the case is accepted and the cause goes on a fix list for the team that owns it.
Some of the return on accepting a dispute never shows up in recovery numbers. A customer whose dispute you accepted, and whose problem you then fixed, may buy from you again. A customer who lost to your evidence packet is less likely to.
That's why we treat this as a question of how you want to run the business as much as a question of numbers.
Some processors charge you to respond. Stripe, for one, adds a $15 fee in the US each time a dispute is contested, on top of the fee for receiving it (it's waived when Stripe's own Smart Disputes tool responds), and only returns it if you win. That fee is separate from anything your chargeback vendor charges, so every response to a case you had no real chance of winning costs you money. Check your processor agreement for that line.
Some processors also screen responses before passing them on. PayU, for example, has a document-review stage before a response is submitted to the acquiring bank, and a status for responses "rejected for lack of sufficient documentation for representment." With a processor like that, your response gets judged twice, and a generic packet has to get past both.
The opposite mistake is accepting every chargeback by default. Some merchants do this, looking at each case only long enough to close it out or flag the customer's account. That leaves money on the table that's rightfully yours, and it teaches customers that a chargeback is the easiest way to get a refund from you.
If you won't fight anything, at least do the root-cause work. The accepted cases still tell you what's going wrong.
Fight hard when the customer isn't right. Who is right decides whether a case deserves a fight. Your evidence and the order size decide whether it's worth one.
Repeat friendly fraud from identifiable customers and organized abuse both belong here, and contesting them carries a deterrence value that a single case's recovery doesn't capture. High order values push the same way, because a weak case on a large order can still be worth the effort. Repeat filers are also where reading cases together pays off, since a claim that looks ordinary on its own stands out once you group claims by account.
All this boils down to:
Fight the disputes where the customer isn't right, with the best evidence you can build. Accept the ones where they are, and send each of those back upstream to find out why it happened.