Three ways to take money, with genuinely different economics. Most of the advice you will read argues for whichever one the author sells. The honest answer is that the right mix follows from a handful of facts about your business, and it is usually more than one.
| Cards | Open banking | Crypto | |
|---|---|---|---|
| Cost | 1.2 to 3%, plus a fixed fee per transaction | Pence, flat, whatever the value | Network fee, processor cut, conversion spread |
| Settlement | Two to five working days. Rolling reserve if you are new | Seconds to minutes | Minutes, plus a banking delay if you convert |
| Reversals | Chargebacks up to 120 days, fee either way, ratio watched | None. Disputes are between you and the customer | Irreversible |
| Friction | Lowest. A saved card is one tap | Smooth on mobile, awkward on desktop | Highest, unless your audience already holds it |
| Recurring | Mature. Network tokens survive card reissue | Exists, but support is uneven between providers | Impractical without custody you should avoid |
Cards are the most expensive per transaction and the correct default for almost everyone, because conversion beats cost at every volume most businesses operate at. A 2% fee on a sale you make is better than no fee on a sale you lose.
The two cases where that stops being true: very high volume where basis points are material, and average order values high enough that the percentage becomes an absolute number worth engineering around.
The saving is real and it is large. On a fifty pound order, cards might cost you a pound and open banking twenty pence. At scale that is a business decision rather than an optimisation.
What it costs you is conversion, and the cost is not evenly distributed. On mobile it is genuinely good and frequently better than cards: the redirect hands off to the banking app, the customer approves with a fingerprint, and they are back. No card number, no expiry date, no three-digit code.
On desktop it is noticeably worse. There is no app to hand off to, so you get one of two things: a QR code the customer scans with their phone to finish on another device, or a full bank login in the browser with whatever authentication their bank demands. Both are several steps more than clicking a saved card, and the device switch in particular loses people.
That matters more than it sounds, because desktop skews towards higher-value and more considered purchases, and B2B is heavily desktop. The saving is largest on exactly the orders where the friction costs most. Look at your own device split before assuming the arithmetic works.
Beyond that: some banks handle the flow badly, recurring is less mature, and you need a provider willing to onboard you, which below a certain volume is its own problem.
The right posture is both. Offer open banking alongside cards, incentivise it if the saving justifies it, and let the customers who prefer it use it. Treating it as a replacement is how businesses discover their conversion rate was load-bearing.
One genuine advantage beyond cost: no chargebacks. For a business with a high dispute rate, or one in a category where friendly fraud is common, that can be worth more than the fee difference.
The honest position: if your customers already hold crypto and expect to pay with it, accepting it removes friction and wins business you would otherwise lose. If they do not, offering it adds a payment method almost nobody uses and a set of operational questions you did not need.
Irreversibility cuts both ways. No chargebacks is genuinely valuable. But a mistaken or fraudulent payment cannot be undone, and a customer who sends to the wrong address has lost their money and will still contact you about it.
If you do accept it, the important decision is whether you hold the asset or convert immediately. Holding is a treasury position and a volatility exposure, taken deliberately or not. Most businesses should convert on receipt and treat it as a payment method rather than an investment.
That last point is the one that survives every other argument here. The specific answer today matters less than whether you can act on a different answer in eighteen months.
We build multi-method payment layers where adding or dropping a method is configuration rather than a project. If you are weighing this up, we can give you the arithmetic for your own volume.