Most Nepali shoppers abandon a cart at the last step: payment. The fix is rarely another gateway. It is the right mix of gateways, wired into a checkout that feels familiar.
Walk through any Kathmandu marketplace group and the pattern repeats. Someone wants to buy, asks "eSewa or Khalti?", and if the store answers with a bank deposit slip, the sale quietly dies. In 2026, Nepali ecommerce runs on five rails: eSewa, Khalti, Fonepay, ConnectIPS and cash on delivery. This is the guide to choosing which ones you need, and wiring them in without leaking sales at checkout.
The five payment rails in Nepal
- eSewa. The oldest wallet, with the deepest reach outside the valley, and close to mandatory for any store. Merchant accounts split into eSewa Merchant (wallet to wallet) and eSewa Business (cards and internet banking through their gateway).
- Khalti. The challenger wallet with a younger, urban base and the cleanest developer documentation. Khalti users convert well on mobile, and the merchant fee is competitive.
- Fonepay. A network that links dozens of bank mobile apps into one QR and one gateway. If your customers bank with Nabil, NIC Asia or Global IME, Fonepay often beats every wallet on convenience.
- ConnectIPS. Real time bank account to bank account transfers, run by Nepal Clearing House. Ideal for higher value orders, tuition fees and B2B invoices.
- Cash on delivery. Still the trust anchor for first time buyers outside the valley. It is not a gateway, but skipping it cuts your audience in half.
The short version: start with eSewa plus Khalti, add Fonepay if your customers live in banking apps, add ConnectIPS for high value orders, and keep cash on delivery until repeat buyers outnumber first timers.
How to choose your stack
Work backwards from your customers, not from the gateway brochure.
- Young, urban and mobile first: Khalti plus eSewa covers most of it. Add IME Pay if you sell to families receiving remittances.
- Bank loyal customers and B2B: Fonepay and ConnectIPS close the deals wallets cannot.
- Cross border sales: Stripe for international cards, plus the local stack for buyers inside Nepal.
Fees matter, but they are not the whole story. Compare the merchant discount rate, the settlement cycle (eSewa and Khalti typically settle within one to three working days, and cash on delivery pays out whenever the courier does) and the KYC effort. Every gateway in Nepal asks for company registration, PAN or VAT papers, bank details and a signed agreement. Budget two to four weeks for approvals if your paperwork is in order.
Checkout UX that stops abandonment
- Show every option on one screen. Do not hide wallets behind a "more" menu.
- Put the most used option first. For most stores that is eSewa or cash on delivery.
- Say the price once. Show the exact total and note any gateway fee up front. Surprises at the redirect screen are where trust dies.
- Handle failure gracefully. If a payment times out, keep the order as "payment pending" and let the customer retry from their order page, not from a blank cart.
- Confirm before the courier calls. Cash on delivery conversion improves when you verify the phone number with an SMS code before dispatch.
Mistakes we see every week
The most expensive errors are the boring ones. A store goes live with only Khalti and wonders why orders from outside the valley stall. A checkout adds a "handling fee" on the last step and carts vanish. A team treats the gateway as set and forget, so a failed webhook quietly drops orders. Measure payment success per gateway every week. If one rail converts below the rest, fix the flow or swap it out.
The gateway is the plumbing. The checkout is the product. Customers pay for the second and only tolerate the first.
Where to start
If you are launching, start with eSewa, Khalti and cash on delivery. Ship that, watch the data for a month, then add Fonepay or ConnectIPS where the orders justify it. If you already run a store, run a thirty day audit: which gateway gets picked, which gets abandoned, and what happens when a payment fails. That report is worth more than any new feature.