Ask anyone who has run payments at scale what chargebacks actually consist of, and you'll get the same answer: a thin layer of genuine fraud sitting on top of a thick layer of confusion. Someone sees a line on their statement they don't recognise, and in the age of banking apps, the dispute button is one tap away. No phone call, no friction, no moment where they might remember. Tap, dispute, done.
Follow the money on that tap. The merchant loses the sale, loses the product, pays a dispute fee, and takes a hit to a ratio that card schemes watch closely. Accumulate enough disputes and you enter monitoring programmes where the fines make the original transactions look like rounding errors. All of it traceable, in most cases, to a customer who simply didn't recognise a name.
Which is why I find it strange that the statement descriptor, the actual words that appear on the customer's statement, is treated as plumbing. It's the last brand touchpoint in the entire transaction, it's read at the exact moment of maximum suspicion, and most companies have never once looked at theirs.
Local matters enormously here, in ways that are invisible until you operate in a market. A descriptor that renders cleanly and recognisably in a South African banking app. A reference on a mobile money receipt that matches what the SMS said at purchase time. A URL on the statement line that leads to a page that answers "what is this charge" in plain language before the customer's thumb finds the dispute button. Receipts that arrive in the language the customer actually reads. Refunds that land back on the original payment method fast enough to end the conversation.
Then there's the part of this that isn't optional. Subscriptions attract specific obligations: clear consent before the first charge, notice before a trial converts to paid, notice before renewals, and cancellation that a consumer can actually execute. In South Africa, consumer protection law gives subscribers real cancellation rights, and the card schemes impose their own notification and cancellation requirements on top. You can treat these as compliance overhead, or you can notice that every one of them is also a dispute-prevention mechanism. A customer who got a renewal reminder and a working cancel button yesterday doesn't dispute the charge today.
The one that seems to hurt founders most: easy cancellation. The fear is churn, and the fear is wrong. Customers blocked from cancelling don't stay, they escalate: to the bank as a chargeback, to the regulator as a complaint, to the internet as a warning. Honest retention, a genuine pause option, a fair offer at the exit, outperforms a hidden cancel link on every metric that survives contact with reality, including revenue.
We built our whole model around carrying these obligations as the seller of record, so I'm not a neutral observer. But the underlying point holds whoever does the carrying: consumer protection done properly isn't the tax you pay on revenue. It's part of how the revenue survives.