BNPL is an external credit decision, not an extra button. This article explains the states, refunds, settlement, and how to measure margin after fees, not conversion alone.
Buy-now-pay-later looks like an option on the checkout page: one more button beside the card. In reality it is a credit decision made by an external provider, a different order lifecycle, a different settlement, and a different refund policy. If you add it the way you add an ordinary payment method, you will discover the difference at the first declined order or the first return.
This article separates what you need to know before enabling it: how it appears at checkout, which states you must design, how refunds and settlement work, and how to measure its effect on margin, not conversion alone. It does not mention fees or rates, because those depend on your contract, volume and product category.
What actually changes
With a card, the decision is almost instant: authorisation, then approval or decline. With BNPL, there may be an eligibility check, then a decision, then an electronic signature, then settlement. Some orders stay pending because the provider asks for more information. That middle state causes most customer problems, because the store does not know whether the order is paid.
So you need an explicit order state for each stage: pending_provider, approved, declined, cancelled, captured, refunded. Relying on the provider's page is not enough. Your store is the source of truth for the customer, and it must show a clear state rather than “processing” forever.
| State | What to define | Common failure |
|---|---|---|
| Before selection | Is the option shown for every basket or above a threshold? What text explains the instalments? | Showing the option, then declining it at the next step |
| Eligibility check | Does it happen before the order is created or after? Who bears the check cost? | Creating an order, then cancelling it on decline |
| Approval | Is the order paid immediately or after signing? When is stock reserved? | Reserving stock for an order that never completes |
| Decline | What message, what alternative, and can the order still be paid by card? | A generic message with no alternative path |
| Cancellation | Who cancels, when is stock released, and how is the provider notified? | Cancelling in the store without cancelling with the provider |
| Refund | Does the customer get money back or is the instalment plan adjusted? | A full refund for a partial return |
| Settlement | When does the store receive the money, and how is it matched to orders? | A gap between the provider report and the store record |
Checkout is not a billboard
BNPL may raise conversion, but that is no reason to hide its terms. Show the total, the number of instalments, any fees, and the financing provider. Do not design a button that looks like a card and then sends the customer into financing. The trust you gain at checkout is lost at the first complaint.
Put the option in its natural place among payment methods, with clear copy. If the customer is not eligible, tell them before they fill in the address, not after. If the order needs a signature, explain that in advance. Transparency is not a legal detail; it is what stops an order becoming a support ticket.
Measuring impact is not only conversion. Compare average order value, return rate, cancellation rate, cost after fees, and margin. Conversion may rise while returns rise with it, turning the gain into a loss. Baymard's 70.19% cart abandonment average makes you want every payment method, but margin is what keeps the store alive.
Refunds are the integration test
What exposes a weak integration most is a partial return. The customer bought three items and returned one. With a card, you refund the value of that item. With BNPL, the amount may already be paid to the store, or split across instalments. Who bears the difference? Is the plan adjusted or is cash refunded? These are not technical details; they are policy that must be written before enabling the method.
From reading to shipping
Want this built, not just explained?
Bring us the problem behind the article. We will tell you what it takes, what it costs, and whether we are the right team.
Test refunds before you test conversion. Create an order, approve it, refund part, refund the rest, then cancel an order before shipping. Watch what happens in the store and in the provider dashboard. If the two records do not match, you have an accounting problem that will appear at month end.
The checklist before enabling it
Do not enable the method because a competitor did. Use this list to decide whether it fits, and how it will be operated.
Customer: does your audience actually buy in instalments, or do promotions attract buyers who never return?
Margin: can your current margin absorb fees, refunds and cancellations?
States: do you have an explicit state for each stage, and do you show it to the customer?
Refunds: is your policy written for full, partial and cancelled orders?
Settlement: who matches the provider report to the store record, and when?
Support: who answers instalment questions, the store or the provider?
Testing: have you tested approval, decline, pending, refund and cancellation?
Measurement: do you know how to judge profit after fees, not conversion alone?
How to choose the provider
Do not compare logos. Compare five things: coverage for your customers, contract clarity, settlement speed, support quality when an order is declined, and integration with your store and accounting system. Ask for a live demonstration of a declined order and a refunded order, not only a successful one.
Read the data clause. What does the provider see from the customer's basket? Does it retain the purchase record? Does it use it for marketing? These questions concern customer privacy and appear in the store policy. Saudi checkout conversion fixes puts them in the wider context of the checkout page.
Frequently asked questions
Does BNPL increase sales? It may increase conversion, but the effect on profit depends on fees, returns and order value. Measure margin, not the button.
Who approves the order? The provider, under its credit model. The store does not see every decision detail, and should not promise the customer approval.
What happens if the order is declined? It should remain payable by another method, with a clear message. Do not empty the basket.
How is a partial return handled? According to the provider contract and your policy. Write the rule before enabling it, and test it with a real order.
Do we need one provider or several? Start with one. A second provider doubles the states and settlement, and should only be added for a clear commercial reason.
Do you support integration? Yes, from state design to testing and reconciliation, inside the store scope.
Sources
- Baymard: checkout usability — 11.3 fields on average, and 22% who abandoned because checkout was too long.
- Baymard: cart abandonment rate — the 50-study 70.19% average.
- Ecommerce development — scope, process, and delivery.
- Rasad's published price bands — the figures this article works from.
Cover image: Shixart1985, CC BY 2.0, via Wikimedia Commons.

