≈ R32 434 lost a yearthe cheapest rail at these numbers
A R650 basket, R150 000 a month, 3% refunded, 4 payout runs, incl 15% VAT. Indicative until re-verified; the chargeback line is an editorial estimate, not a published fee. Open these numbers in the full calculator →
8 providers tracked, 13 rails modelled
Rates ex VAT unless a figure says otherwise
No provider owns us, and nothing is ranked for pay
Every fee carries the date we verified it
Five ways money reaches you
Not every rail costs the same.
A gateway can bundle several rails with different prices. Compare the actual method mix, basket, refunds and payout frequency rather than assuming the card headline controls the result.
Cards
Visa and Mastercard still take most online payments, and every gateway here processes them. They are also the only rail that can be charged back.
The customer pays straight from their bank at checkout. Cheaper than cards almost everywhere, settles in real time, and cannot be reversed by a chargeback.
Real-time bank payments by proxy ID, plus SnapScan and Zapper at the counter. Merchant pricing is set per provider rather than published as a rate card.
Payflex, PayJustNow, Float and MoreTyme lift conversion on bigger baskets, and the provider carries the credit risk. You pay for that in the merchant fee.
5.5% + R2.00 via MoreTyme · highest here
MoreTyme merchant fee, published by PayFast
Read off the provider's own published pricing page.
Local gateways settle in rand only. Selling abroad or invoicing a foreign client means a second set of tools, and a second set of fees on the way home.
The card and instant EFT figures span every rail this site models, ex VAT, computed from the same engine as the calculator. The BNPL figure is what PayFast publishes for MoreTyme on its own fees page: no BNPL provider publishes a merchant rate directly. PayShap, QR and cross-border pricing is set per provider or per corridor, so those two cards carry no rate rather than a borrowed one.
The tool nobody else built
Headline rates lie. Effective rates don't.
Set your basket and your monthly volume. We add up everything each provider quietly takes per R100, then rank them by what actually reaches your account.
R650
R150 000
O
Ozow pay-by-bankCheapestPay-by-bank · no chargebacks
R2 703/mo
1.80% effective · keeps R98,20/R100
Pe
Peach Pay by BankPay by Bank · no chargebacks
R3 075/mo
2.05% effective · keeps R97,95/R100
Ps
Paystack EFTEFT · no chargebacks
R3 553/mo
2.37% effective · keeps R97,63/R100
St
Stitch Capitec PayCapitec Pay · no chargebacks
R3 563/mo
2.38% effective · keeps R97,62/R100
iK
iKhokha Instant EFTInstant EFT · no chargebacks
R3 563/mo
2.38% effective · keeps R97,62/R100
PU
PayUNo published pricing
We did not find public South African rates or pricing terms on the PayU pages checked. Obtain a current written quote from PayU.
Ranked on the same engine as the full calculator (incl 15% VAT), holding refunds at 3% and payouts at 4 runs a month. 8 further rails sit below these five. The full tool opens those two inputs and shows the receipt behind the cheapest option, including the chargeback provision, which is an editorial estimate rather than a published figure. Open the full calculator →
Local gateways, side by side
Eight providers, one comparison table.
The five cheapest providers at our standard scenario. The full table carries all 8, with settlement, watch-outs and the date each figure was verified.
The lowest modelled rate is not a universal recommendation. These are editorial starting points for five common scenarios, with the effective rate computed from the assumptions shown on each card.
I sell physical products on WooCommerce or Shopify.
Y
Yoco
keeps R96,95 per R100 · 3.05% effective
At R450 a basket and R60 000 a month, 3% refunded, incl 15% VAT.
Yoco is the simplest place to start: no monthly fee, no fixed fee per transaction, and one account covering online and in-person. PayFast is the alternative when you would rather have every payment method a shopper might reach for.
Alongside it
PfPayFast, the widest method coverage, if reach matters more than rate
OOzow, add the bank rail once volume justifies a second integration
On Shopify, the platform adds its own 0.6% to 2.0% third-party gateway fee on top of whatever your gateway charges. That stacking is what pushes real cost toward 4% to 5.5%, and no gateway choice removes it.
At R350 a basket and R200 000 a month, 2% refunded, incl 15% VAT.
Recurring billing is a tokenisation problem before it is a rate problem. Peach publishes card-on-file and recurring-payment capabilities, plus 3-D Secure 2 support. Confirm retry behaviour and current written terms for your implementation before choosing a gateway.
Alongside it
PsPaystack, a capable subscriptions API with better docs
StStitch, debit orders, where fixed monthly collections run at scale
Much of Peach’s pricing is plan-based or quoted rather than published. Get a written quote before you build against it, and re-check the rate here against what you are actually offered.
I am writing the checkout myself, in React or Next.js.
Ps
Paystack
keeps R96,38 per R100 · 3.62% effective
At R600 a basket and R150 000 a month, 2% refunded, incl 15% VAT.
Paystack publishes a popup checkout, signed-webhook guidance and failure diagnostics. It is a Stripe subsidiary, but it remains a separate provider with its own South African terms and eligibility.
Alongside it
StStitch Express, enterprise rails self-serve, with native BNPL since 2026
Paystack’s published dispute process and scheme charges can change. Read the current dispute documentation, response deadline and charge schedule, and confirm your commercial terms at onboarding.
At R8 000 a basket and R400 000 a month, 1% refunded, incl 15% VAT.
On a large basket the percentage is the whole cost, and the bank rail undercuts every card rate here. It also settles in real time and cannot be charged back, which removes the single largest risk on a high-value order.
Alongside it
StStitch, pay-by-bank with enterprise orchestration behind it
PePeach, Pay by Bank alongside a card rail on one contract
Pay-by-bank is not a complete checkout on its own. Some customers will only pay by card, so budget for a card rail beside it and expect that rail to cost more per rand.
My customers are overseas and pay in dollars or euros.
W
Wise
not a rand rail, so not modelled here
For invoicing foreign clients directly, compare regulated services that may provide supported local receiving details and disclose their exchange-rate margin. For eligible software sales, a merchant of record may administer specified indirect-tax functions within its contract, but does not automatically remove every obligation.
Alongside it
PsPaystack, keep it for the rand side of the business
PdPaddle, a merchant of record handles global VAT for digital goods
PoPayoneer, where a marketplace dictates the rail
Stripe currently labels South Africa an extended-network market rather than a standard direct market. Do not assume a South African entity qualifies for direct onboarding. A foreign structure raises fact-specific tax, exchange-control, banking and substance questions, so obtain professional advice first.
Editorial starting points use the stated capability research, not payment or ranking position: the calculator sorts on price alone and may disagree. Every displayed rate is computed at the scenario printed on the card. Verify current eligibility and terms before deciding.
Four things that cost merchants money
Things everyone gets wrong.
Stripe works in South Africa.
Stripe currently lists South Africa as an extended-network market associated with Paystack, not in the same category as its standard direct markets. Do not assume that a South African entity and bank account qualify for direct Stripe onboarding. Compare eligible local providers, a merchant of record within its supported scope, or a properly advised foreign structure, and obtain written confirmation from the provider for the exact entity.
The rate on the pricing page is the rate you pay.
Every rate published by a South African gateway, and every rate on this site unless a figure says otherwise, is quoted excluding VAT. Add 15% to model what actually leaves your account. On top of that sit fixed per-transaction fees, per-transaction minimums, payout fees and refund losses, none of which appear in the headline percentage. That gap between the advertised rate and the effective rate is the reason the calculator on this site exists.
If you refund a sale, you get the fee back.
Almost never. Most gateways keep the original processing fee when you refund a customer, and several charge a further fee to process the refund itself. A R500 sale refunded in full can still cost you R15 to R20 with nothing to show for it. At a 3% refund rate this quietly moves the ranking between providers, which is why the calculator treats refund losses as a first-class input rather than a footnote.
Shopify Payments will sort out my fees.
For a South African store using a third-party payment provider, Shopify says third-party transaction fees may apply and depend on the plan. Those charges can sit alongside the provider's own rate. Confirm current Shopify Payments eligibility, the store's plan, order treatment and any exclusions in the admin and contract before modelling the combined cost.
Beyond the rand
Selling past the border changes the maths.
Local gateways settle in rand only. The moment you sell into Lagos or invoice a client in London, you are on someone else's rails, at someone else's rates.
The figures on those two pages are directional, not verified. They are third-party, non-South-African and corridor-dependent, so they were never part of the pricing pass behind the rest of this site and they carry no Verified badge anywhere.
Evergreen, updated as it moves
The rules are changing in your favour.
The SARB is opening clearing and settlement to non-bank PSPs. More providers, lower costs and new payment methods, from 2026 onward.
15 Jun 2026
Draft Authorisation Framework consultation closed
The May 2026 activity-based authorisation framework and related notices were published for consultation. They are proposals, not current licences or final law, and may change before implementation.
7 Jul 2026
QR+ Standard 1.2 published
SARB published the QR+ Standard version 1.2 overview for interoperable domestic QR payments. Publication of the standard does not mean every wallet or merchant is already interoperable; implementation and participant readiness still matter.
still ahead
No confirmed date
Possible direct non-bank authorisation
The draft framework proposes activity-based authorisation for some non-bank payment institutions. No provider should be described as having a future licence, and no implementation date or market outcome is guaranteed until final instruments are issued.