QR is the cheapest acceptance device ever invented — a printed square. Whether it scales into real interoperable payments or stays a walled-garden gimmick depends entirely on the rail behind it and whether anyone agrees on one standard. South Africa is finding that out the hard way.
A QR payment encodes payment instructions — who to pay, sometimes how much — in a scannable square. It collapses the cost of card acceptance to near zero: no terminal, no PED, no dongle. For a micro-merchant a printed sticker is the entire hardware budget. That is why QR exploded in markets with large informal economies and cheap smartphones, and why it underpins some of the largest payment systems on earth.
But QR is only a transport for instructions. It does not move money; the rail behind it does — a card scheme, an account-to-account instant rail, or a closed wallet balance. The square is the easy part. The economics, fraud profile and interoperability all live in the rail, which is exactly where the interesting fights happen.
EMVCo publishes two QR specifications, defining a standardised TLV (tag-length-value) payload so a single code can carry multiple domestic and international payment options:
| Mode | Who shows the code | Who scans | Typical use |
|---|---|---|---|
| MPM — Merchant-Presented | Merchant (sticker or terminal screen) | Consumer, with their app | The common case: shop counter, market stall |
| CPM — Consumer-Presented | Consumer (app shows a code) | Merchant, with a scanner | Higher-throughput retail, transit gates |
MPM codes are either static (one printed code per merchant, consumer keys the amount) or dynamic (a fresh code per transaction with the amount embedded — less error-prone, needs a screen). The EMVCo MPM specification has been adopted as a national QR standard in several countries, including Singapore and South Africa. The standard makes the format interoperable; whether the payment is interoperable is a separate, harder question of who honours whose code.
The transformative QR systems are the ones where a single national standard sits on top of a cheap, real-time account-to-account rail, so any app can pay any merchant.
India’s UPI is the canonical example: a unified QR over the UPI instant rail, with effectively zero merchant cost on person-to-merchant flows, processing billions of transactions a month. Brazil’s PIX reached comparable scale astonishingly fast by pairing an instant rail run by the central bank with a standardised QR and aggressive defaults. The lesson both teach: QR adoption is not a QR problem, it is a rail-plus-mandate problem — cheap real-time settlement, one standard, and a regulator willing to push.
A single national QR every app must read and write — no walled gardens. This is the part fragmented markets keep failing.
A2A instant settlement at near-zero merchant cost is what makes QR beat cash, not the code itself.
UPI and PIX both rode deliberate central-bank or regulator pressure. Voluntary interoperability tends to stall.
South Africa got QR early and fragmented. SnapScan (Standard Bank-backed) and Zapper built genuine merchant footprints years ahead of most markets — but as separate walled gardens. A merchant’s SnapScan code did not work in the Zapper app, and vice versa, and Masterpass added a third silo. Consumers learned which sticker matched which app, which is precisely the friction that interoperability is meant to remove.
PayShap — the rapid payments programme run by PayInc (the rebranded BankservAfrica, in which the SARB took a 50% interest in late 2025) — arrived as the low-cost A2A instant rail QR has been waiting for. Three years in, PayShap has pivoted explicitly toward merchant payments, and a standardised QR over PayShap is the ambition: let a merchant accept PayShap the way they accept SnapScan or Zapper today, from any bank app.
The SARB is now pushing a unified QR+ standard to end the “works with SnapScan but not Zapper” mess, and the 2026 budget confirmed PayInc as the shared, open national payments infrastructure. Whether QR+ over PayShap actually displaces the incumbents depends on the same trio India and Brazil nailed: one standard, near-zero consumer cost, and an experience that genuinely rivals a tap.
EMVCo MPM standardises the code’s bytes, not who accepts it. South Africa had “standard” QR codes that still only worked in one app. Format interop is necessary, not sufficient.
Static codes let the consumer key any amount, and stickers can be physically overlaid with a fraudster’s code. Dynamic QR and merchant-display verification matter more than they look.
Incumbents with a merchant footprint have little incentive to interoperate. Without a regulatory push, “we’ll integrate later’’ means never.
If QR carries merchant fees or consumer cost while cash is free, informal merchants stay on cash. UPI and PIX won on near-zero cost, not technology.
Consumer-presented mode needs the merchant to have a scanner — reintroducing the hardware cost QR was meant to remove. MPM is the low-cost default for a reason.
Use QR (MPM, static) when you need the cheapest possible acceptance for micro-merchants and the rail behind it is cheap and real-time. In an emerging market with smartphone penetration and an A2A instant rail, QR is the most capital-efficient way to bring merchants into digital payments — nothing else matches a sticker. Use dynamic MPM wherever the merchant has a screen and amount-entry error or sticker-swap fraud is a concern.
Skip QR as a strategy if you cannot get to one standard. Launching a fourth walled-garden QR wallet in South Africa adds friction, not access — the market does not need another app-specific sticker. If you are a bank or PSP here, the bet is QR+ over PayShap, not a proprietary code. Be honest about CPM: it reintroduces scanner hardware, so reserve it for high-throughput retail where that cost is justified.
Cost of being wrong: build a proprietary QR garden and you may win a few years of merchant lock-in, then strand it when a regulator-backed interoperable standard arrives — which in South Africa it now is. Price it above cash and you get stickers on counters that no one uses. The expensive error is treating QR as a product when it is a feature of the rail; bet on the rail.