Buy-now-pay-later at checkout: the real effect on conversion and refunds

Buy-now-pay-later at checkout: the real effect on conversion and refunds is a question many e‑commerce operators face, and the answer is rarely as simple as “it always helps.” In South Africa the BNPL market has exploded, with providers reporting double‑digit growth year on year, yet the same data set also shows higher charge‑back rates and longer settlement cycles. This article breaks down the numbers, shows you a step‑by‑step method to calculate the net impact on your store, and gives concrete actions you can take today.

The hidden cost of BNPL at checkout

When a shopper sees “Buy‑now‑pay‑later” next to the “Pay now” button, the immediate reaction is a boost in the conversion funnel. Offering buy-now-pay-later alongside card options reliably lifts checkout conversion, since it removes the single biggest objection for a hesitant buyer: paying the full amount today. That sounds attractive, but the headline hides two critical variables:

• Average order value (AOV) inflation. BNPL users tend to purchase higher-priced items, since spreading the cost changes what feels affordable, which typically pushes average order value up too.
• Refund and charge‑back exposure. Because the consumer’s commitment is spread over weeks, the likelihood of a post‑purchase dispute climbs. The figures used in the worked example below (1.8% for BNPL orders versus 0.9% for card orders) are illustrative, but reflect the direction seen consistently across BNPL providers.

Combine those two and the net profit effect can swing either way. A retailer that sells mostly low‑margin accessories may see profit erode, while a high‑margin fashion brand could actually benefit. The only way to know is to isolate the BNPL segment in your own analytics and apply a consistent measurement framework.

How to measure the true impact on conversion

The method below lets you calculate the incremental revenue (or loss) that BNBN at checkout generates. It works with any platform that can export order data – Shopify, Magento, WooCommerce, etc.

Step 1 – Pull a clean data set

Export all orders from the last 90 days, including these fields: order ID, payment method, order value, discount code, refund status, and date of purchase. Filter out test orders and any that were cancelled before payment.

Step 2 – Separate the cohorts

Create two groups:

• BNPL cohort. Orders where the payment method is a BNPL provider (e.g., PayFlex, AfterPay).
• Non‑BNPL cohort. All other orders (credit card, debit, cash on delivery).

Step 3 – Calculate baseline metrics

For each cohort compute:

• Conversion rate = (Number of orders ÷ Number of sessions that reached checkout) × 100
• Average order value = Total revenue ÷ Number of orders
• Refund rate = (Number of refunded orders ÷ Number of orders) × 100
• Net revenue = (Total revenue – Refund amount) – (Transaction fees)

Transaction fees differ: BNPL providers typically charge 2.5 % of the order value plus a R15 flat fee per transaction, while Visa or Mastercard charge around 1.8 %.

Step 4 – Determine incremental lift

Subtract the non‑BNPL metrics from the BNPL metrics. The key figure is the “incremental net revenue per session.” If the BNPL cohort adds R2.30 of net revenue per checkout session, you have a positive lift. If it subtracts R0.85, the program is costing you money.

Step 5 – Model long‑term effects

Because BNPL spreads payments, the cash‑flow timing changes. Use a simple 30‑day discount factor to bring future payments back to present value. For example, a R1 000 order paid over four weeks at a 5 % annual discount rate has a present value of roughly R995. Multiply that by the number of BNPL orders to see the true cash impact.

When you finish the calculation you will have a clear, numeric answer to the question posed in the title.

A hypothetical worked example

Consider a mid-size online fashion retailer that introduces BNPL. Here are illustrative numbers for a 90‑day period after launch:

• Total sessions that reached checkout: 45 000
• BNPL orders: 3 200 (7.1 % of checkout sessions)
• Non‑BNPL orders: 10 800 (24 % of checkout sessions)
• BNPL AOV: R2 450
• Non‑BNPL AOV: R2 180
• BNPL refunds: 58 orders (1.8 % of BNPL orders)
• Non‑BNPL refunds: 97 orders (0.9 % of non‑BNPL orders)
• BNPL transaction fee: 2.5 % + R15 per order
• Card transaction fee: 1.8 %

Step 1 – Baseline metrics

BNPL conversion = (3 200 ÷ 45 000) × 100 = 7.11 %
Non‑BNPL conversion = (10 800 ÷ 45 000) × 100 = 24.00 %

BNPL net revenue per order = (R2 450 × (1 - 0.025) - R15) – (R2 450 × 0.018 × refund rate) ≈ R2 380 – R44 ≈ R2 336
Non‑BNPL net revenue per order = (R2 180 × (1 - 0.018)) – (R2 180 × 0.009 × refund rate) ≈ R2 138 – R20 ≈ R2 118

Step 2 – Incremental lift

Incremental net revenue per BNPL session = R2 336 – R2 118 = R218
Multiply by the number of BNPL sessions (3 200) = R697 600 extra net revenue.

Step 3 – Cash‑flow timing

Present value of future BNPL payments (using 5 % annual discount) reduces the R697 600 by about 0.5 %, giving R694 200. The retailer still gains roughly R694 k over the period, which translates to an additional R15.43 per checkout session that saw BNPL.

The example shows a clear positive impact, but note the margin cushion. If the retailer’s profit margin were only 5 %, the extra R218 per order would be largely eaten by the higher fee structure, turning the lift negative. That’s why the same calculation must be run on a per‑margin basis.

Applying the findings to your store

Now that you have a repeatable framework, follow these steps to embed it in your regular reporting cycle.

Integrate into your analytics dashboard

Most BI tools (Google Data Studio, Power BI, Tableau) can pull the raw export and apply the cohort logic automatically. Build a “BNPL Impact” card that refreshes daily and shows:

• Conversion lift (percentage points)
• Incremental net revenue per session
• Refund rate differential
• Cash‑flow timing adjustment

Set alerts for any metric that moves beyond a pre‑defined threshold – for example, a refund rate above 2 % for BNPL orders.

Test different placement and messaging

The same data set can be sliced by placement (above the fold vs. hidden in a modal) and by copy (“Pay later, no interest” vs. “Split your payment”). Run A/B tests that last at least two weeks to capture enough BNPL orders for statistical significance. Record the incremental lift for each variant and keep the best performer.

Negotiate fees with providers

If your analysis shows that the fee structure is eroding profit, approach the BNPL partner with the numbers. Many providers will lower the flat fee or offer a volume‑based discount once you prove a certain transaction volume.

Consider a hybrid approach

If your margin is thin on low‑ticket items, you can hide BNPL behind a minimum order value filter (e.g., only show it for carts above R1 500). The same calculation can be run on the filtered cohort to confirm that the restriction improves overall profitability.

Common pitfalls and how to avoid them

Even experienced operators can stumble when adding BNPL to the checkout flow. Below are the most frequent mistakes and practical ways to sidestep them.

Ignoring the refund lag

Refunds for BNPL orders often take longer to process because the provider must reverse the installment schedule. If you simply count refunds as a percentage without accounting for the extra days, you will underestimate the cash‑flow strain. Add a “refund lag” column to your spreadsheet and apply a daily discount factor for the extra days.

Mixing cohorts without proper segmentation

Some stores tag orders by payment method but also use a “payment gateway” field that groups all third‑party processors together. This can blend BNPL with other deferred‑payment options, diluting the analysis. Create a dedicated “BNPL” tag in your order management system to keep the data clean.

Relying on a single provider’s data

If you offer multiple BNPL options, each will have its own fee schedule and refund behaviour. Treat each provider as a separate cohort; aggregate only after you have the per‑provider lift figures.

Assuming conversion lift equals profit lift

Higher conversion does not automatically mean higher profit. Always subtract the incremental fees and the higher refund cost before celebrating a lift. The method in this article forces you to do exactly that.

Neglecting mobile vs. desktop differences

Mobile shoppers are more likely to use BNPL, but they also have higher cart‑abandon rates. Run the calculation separately for each device type. If mobile shows a negative net lift, you may choose to hide the BNPL button on mobile while keeping it on desktop.

Quick checklist for BNPL optimisation

• Export checkout data for the last 90 days with payment method, order value, and refund status.
• Separate BNPL and non‑BNPL cohorts.
• Calculate conversion, AOV, refund rate, and net revenue for each cohort.
• Apply the transaction‑fee formula (percentage + flat fee) to both cohorts.
• Subtract non‑BNPL metrics from BNPL metrics to get incremental lift.
• Adjust for cash‑flow timing using a discount factor.
• Integrate the calculation into a daily dashboard and set alert thresholds.
• Run A/B tests on placement, copy, and minimum‑order filters.
• Negotiate fee reductions if the incremental net revenue is marginal.
• Review mobile vs. desktop performance separately.
• Schedule a quarterly review to re‑run the analysis as fees and shopper behaviour evolve.

Understanding the real effect of Buy‑now‑pay‑later at checkout requires more than a quick glance at conversion numbers. By isolating the BNPL cohort, accounting for higher fees and refund rates, and adjusting for cash‑flow timing, you can decide whether the feature adds true value to your bottom line. If you need a fast, data‑driven assessment of your own store, run a free scan of your store and see where your checkout is leaking.

For further reading on the mechanics of deferred payment options, see the Wikipedia entry on Buy now, pay later, which outlines the global market trends and regulatory considerations.