The most expensive number in the chargeback industry isn't a fee. It's an average. A merchant gets their first dispute notification at 11 p.m., searches "is it worth fighting a chargeback," and lands on a statistic saying merchants lose most of the disputes they face. They look at the $340 on the line, weigh it against a fight the internet just told them they'll probably lose, and close the tab. The money is gone — not because their case was weak, but because a number that was never about them told them who they were.

If you've done some version of this, here's the uncomfortable part: the statistic was technically true and personally irrelevant. The gap between those two things is where merchants quietly bleed revenue, year after year, one unanswered dispute at a time.

Where the scary number actually comes from

Ask around and you'll hear that the average chargeback win rate sits somewhere between dismal and coin-flip. The exact figure moves depending on which industry report you read, which card networks it covers, and which year it was compiled — which should already make you suspicious of treating it as a law of nature.

But nearly every version of the number shares one structural feature that almost nobody mentions: it averages across every dispute, not every fight.

That denominator includes merchants who never responded at all — each one an automatic loss. It includes responses thrown together in ten minutes from a generic template. It includes disputes over charges the merchant had already refunded, and true-fraud cases where a stolen card was used and the merchant was always going to lose. All of it gets stirred into one pot and served as "your odds."

The average isn't really measuring how winnable disputes are. It's measuring how merchants, in aggregate, behave. Those are very different questions.

The blank-exam problem

Imagine a professor announces the class average on an exam was 41% — and only later mentions that a third of the students never turned in a paper, and every blank was scored as zero. The average is arithmetically correct. It's also useless for predicting how you'll do if you show up and write.

Statisticians would say you're reading an unconditional average when the question you actually care about is conditional: what happens given that a merchant submits a real, on-time, evidence-backed response? Dispute outcomes fall into at least three very different buckets — never answered, answered with generic boilerplate, and answered with evidence matched to the specific reason code — and their win rates are not remotely alike. Averaging them together is averaging apples, oranges, and empty bags, then telling you the fruit is bad.

Every unanswered dispute drags the published number down. Then the published number convinces the next merchant not to answer. The statistic doesn't just describe the problem; it manufactures more of it.

Learned helplessness, merchant edition

There's a name for what happens next, and it comes from real psychology, not marketing decks. In the late 1960s, Martin Seligman and Steven Maier ran experiments showing that animals exposed to outcomes they couldn't control later stopped trying to escape situations they could control. The finding became known as learned helplessness: it isn't the bad outcome that shuts down effort — it's the belief that outcomes and effort are disconnected.

Chargebacks are a nearly perfect machine for producing that belief. You upload evidence into a portal. Weeks pass in silence. A verdict comes back as a single word — lost — with no explanation of what the reviewer weighed or what was missing. Delayed, opaque feedback is exactly the kind that makes effort feel pointless. Lose two disputes like that and the rational-feeling conclusion is "the bank always sides with the cardholder." So you stop responding. Your personal win rate becomes zero. The industry average absorbs your zeros. And the cycle recruits its next believer.

Here's the twist worth knowing: when Maier and Seligman revisited the theory in 2016, they argued the arrow actually points the other way. Passivity is the default; it's the experience of control that has to be learned. You don't talk yourself out of helplessness — you act, watch outcomes vary with your actions, and let the evidence teach you that the lever is connected to something. For a merchant, that means the cure isn't optimism. It's a spreadsheet.

The only win rate that applies to you

Your real odds are conditional on things the industry average knows nothing about: your mix of disputes (true fraud you probably should lose versus friendly fraud you can contest), the evidence trail your product naturally generates (a SaaS with login records is not a boutique shipping fragile goods), whether your responses answer the specific reason code's question, and — above everything — whether you respond at all, on time.

A base rate is the right estimate only when you know nothing else about the case in front of you. You know plenty else. A "product not received" dispute against a merchant holding delivery confirmation is not a coin flip, no matter what the aggregate says.

The economics lean the same direction. A response costs you an hour — minutes, with the right tooling — against the full disputed amount. Even a middling personal win rate, applied across a year of disputes, recovers real money. And every response you file generates something the non-responder never gets: feedback that makes the next response better.

Your next moves

  • Compute your real number today: export the last 12 months of disputes from your Stripe dashboard and count wins, losses, and — in a separate column — disputes you never answered. That third column is your fastest lever, because those weren't losses, they were forfeits.
  • Segment your losses by reason code and find the single code you lose most often. Fix your evidence for that one code first — targeted improvement beats trying to overhaul everything at once.
  • Run a 90-day experiment: respond to 100% of disputes before the deadline, and log each outcome with the reason code and the evidence you submitted. You're building the experience of control the research says helplessness can't survive.
  • Autopsy your last three losses. Did your evidence answer the reason code's actual question — "was this authorized?", "was it delivered?" — or did it just argue, in general terms, that the charge was legitimate? Reviewers score the former.
  • Put a deadline safety net in place before you close this tab: a calendar alert or an email forwarding rule on Stripe's dispute notifications, so one busy week can't quietly become an automatic loss.

The average loses because the average doesn't answer

That's the whole argument of this piece in one sentence — and it's the failure mode Argeback was built to delete. It ingests your Stripe disputes the moment they're filed, drafts a response built around the evidence that dispute's reason code actually calls for, and files it before the 7-day deadline, straight from your phone. The scary industry number is mostly a census of merchants who were too busy, too discouraged, or too late. If you'd rather your win rate be decided by your evidence than by your availability, see how it works at argeback.lumenlabs.works.