Most payment guides for Vietnam put six providers in a table and ask you to crown a winner. That framing is the problem. A store that sells to Vietnamese buyers and to buyers abroad is not picking one gateway. It is picking two rails that break in different ways, then living with the paperwork both of them generate.
Here is the order I would actually decide in for 2026 — buyers first, fees last.
Split your buyers before you look at a single fee
Write down how your orders divide between three groups.
- Urban buyers in Hanoi and Ho Chi Minh City who are comfortable with a wallet or a card.
- First-time and rural buyers who still want to hand cash to a courier.
- Foreign buyers, expats and tourists who recognise a global brand and get nervous about an unfamiliar local one.
The split tells you where to spend integration effort. If most of your volume is domestic, an international-first setup taxes every order. If most of your volume is cross-border, a wallet-only checkout quietly kills conversion for the people who cannot use it.
The local rail: wallets and bank QR
MoMo is the default for general Vietnamese consumers, with the largest user base of the group — the article this is based on cites more than 31 million users. Fees sit around 1-2%, money moves instantly between wallets, and QR codes let you bridge an offline counter and an online shop.
ZaloPay matters when your selling already happens inside Zalo. Fees run about 1-1.8%, and payment links can sit in the same conversation where the customer asked about the product. If you run a Zalo official account for sales and support, this is the least friction you can add.
VNPay is the bank-side option. QR acceptance is wide, integration reaches a large number of Vietnamese banks — the source guide puts it at over 40 — and the fee band of 1-1.5% is the cheapest of the three. The trade-offs are slower settlement and documentation that is less friendly if your developers expect a modern API.
Picking between MoMo and ZaloPay is not a technical question. Go MoMo when you sell to the general public; go ZaloPay when your funnel is conversational. Plenty of stores run both and let the customer choose.
The international rail: PayPal and Stripe
Both charge 2.9% + $0.30, and both are expensive next to a local wallet. You pay that premium for reach, not for processing.
PayPal wins on trust. Buyer protection is the reason a first-time foreign customer will pay a seller they have never heard of, and the multi-currency support saves you from explaining exchange rates. The costs show up later: currency conversion, and support that is not built around a Vietnamese merchant.
Stripe wins on engineering. Recurring billing, a clean API, marketplace payouts and fraud tooling make it the sane choice for subscriptions or a custom platform. It also asks more of you up front — a technical team, and a bank setup that can be awkward from Vietnam.
A practical pattern: PayPal for one-off international orders, Stripe when you need recurring charges or you are building the checkout yourself.
Cash on delivery is not a legacy feature
COD still carries a large share of Vietnamese e-commerce, and the source guide puts it at roughly 40-50% of transactions, concentrated among first-time and rural buyers. Treat it as a trust product rather than a payment method. It costs you 2-3% in logistics and returns, and it slows your cash cycle.
The useful move is not removing it. It is making prepayment slightly more attractive — a small discount for paying online — so the customers who can pay digitally do, while the ones who need reassurance still convert.
Run the cost comparison on your own basket
Take a month of sales and price it twice. At $10,000 in monthly volume, a domestic wallet lands somewhere around $100-$200 depending on the band, while an international card rail lands near $290 plus $0.30 per transaction.
That gap is not a rounding error. If the bulk of your buyers are Vietnamese, routing them through an international gateway can cost four figures a month that you never needed to spend. The point is not that one rail is better — it is that each order should travel the cheapest rail that the buyer will actually use.
The cost nobody puts in the comparison table
Every provider keeps its settlement and reconciliation reports inside its own dashboard, not inside your accounting system. That is fine until two things happen: a customer disputes a charge weeks later, and someone asks you to prove a specific month's revenue.
At that moment you need the original settlement file, the transaction reference and the matching invoice. Gateways keep downloadable exports for a limited window. Under the Law on Accounting 2015, Article 41, those records are accounting evidence that has to survive for at least 10 years.
So build the boring habit early: export each month's settlement file, store it somewhere you control, and keep more than one version. A synced folder with file versioning is enough — what matters is that the file is yours and that a dead laptop does not take it with it. Two details worth checking before you commit to any storage plan: whether the price is genuinely one-time or renews higher later, and whether the capacity covers several years of exports rather than only the current one.
Mistakes that cost real money
- Assuming every buyer wants to tap a card. A large share of Vietnamese shoppers reach for a wallet or a bank transfer.
- Stacking every gateway on the checkout page. Four options is already a decision tax; three or four well-chosen ones cover almost everyone.
- Shipping a desktop-only checkout. Mobile is where most of the traffic is — the source guide puts it above 70%.
- Leaving refunds undefined per method. Wallet refunds can be immediate, bank and card refunds take days, COD refunds need a human. Say so before the order, not after.
- Storing card data yourself. Let the gateway tokenise it and keep your servers out of scope.
What changes next
Four shifts are worth tracking. Buy-now-pay-later providers are pushing instalments into mainstream checkout. A small number of merchants are experimenting with crypto settlement. Payment is being absorbed into super apps, so the checkout increasingly lives inside an app rather than a website. And machine learning fraud scoring plus real-time settlement are steadily shrinking the gap between paying and being paid.
None of that changes the core decision: two rails, chosen by buyer mix, reconciled monthly.
A decision you can make this week
- Count your orders by buyer group for the last month.
- Pick one local rail and one international rail that cover the top two groups.
- Add COD only if first-time or rural buyers are material to your volume.
- Test each one in sandbox, including a refund, before it goes live.
- Walk the checkout on a phone, on a slow connection.
- Put a monthly settlement export on the calendar.
The full comparison — fee bands, integration difficulty and per-business-type recommendations — lives in the original guide on my site.
Read the full comparison on Tân IT365 — fee bands, integration difficulty and per-business-type recommendations.
