BNPL Economics: Conversion Lift, Cash Timing, and Refund Risk
By Diosh — Founder, AHAeCommerce | eCommerce decision intelligence for $50K–$5M GMV operators
Buy now, pay later providers charge merchants 2–8% of the transaction value in fees — typically 2–3x what a standard credit card costs to process. In exchange, they promise higher conversion rates, larger average order values, and access to customers who prefer deferred payment. During the 2025 holiday season, BNPL crossed $20B in U.S. online spending (Adobe Analytics, 2025). That volume is real. The question is whether enabling BNPL converts that latent demand into profitable revenue for your business, or whether it converts cart completions at a margin cost that doesn't justify the checkout optimization.
The decision isn't binary, and it isn't the same for every product or customer segment. BNPL is a financing tool that changes buyer behavior in specific, predictable ways — both the ways that benefit the operator and the ways that don't.
The Default Assumption (and Why It Fails)
The standard framing for BNPL evaluation is a conversion rate test: enable BNPL at checkout, measure whether conversion rate improves, calculate the revenue gain vs. the fee cost, and deploy if the math is positive. This is directionally correct, but it measures the wrong thing.
Conversion rate improvement measures whether more buyers complete a checkout when BNPL is available. It does not measure whether those buyers are incrementally valuable or whether they are the same buyers who would have converted on standard payment terms — just now costing the merchant 2–4% more to process their purchase.
The incomplete metric the conversion test misses: post-purchase refund rate, repeat purchase rate, and average order quality for BNPL vs. non-BNPL customers. BNPL appeals disproportionately to buyers who are price-sensitive, who are making stretch purchases beyond their immediate budget, and who convert in a moment of intent that may not survive 30 days of post-purchase reflection. All three of these behavioral tendencies increase refund rate and reduce repeat purchase rate compared to the buyer who completed the same purchase with a credit card.
What the Decision Actually Hinges On
Fee Drag vs. Incremental Margin
The BNPL fee ranges from 2% (for high-volume merchants negotiating with Affirm or Klarna) to 6–8% for smaller merchants on default terms. Standard credit card processing runs 1.5–2.9% plus $0.30 per transaction. At 4% BNPL fee vs. 2.5% card, the merchant pays 1.5 percentage points more per transaction for BNPL-enabled purchases.
For a $150 AOV, that's $2.25 in additional processing cost per transaction. The BNPL is worth that incremental cost if — and only if — the buyer who used BNPL would not have completed the purchase without it. If 60% of BNPL checkouts would have converted on standard payment terms anyway, the effective cost of BNPL-enabled incremental conversions is $2.25 / 0.40 = $5.63 per incremental transaction. Whether that is worth paying depends on the contribution margin of those incremental transactions.
At a 45% gross margin on a $150 AOV, the gross contribution per transaction is $67.50 before BNPL fee, and $61.50 after. The incremental $6 in transaction cost is recovered if the BNPL-enabled incremental buyer has average retention and lifetime value. It is not recovered if BNPL buyers have meaningfully lower retention.
Refund Rate and Return Timing
BNPL buyers return products at higher rates than credit card buyers in most product categories, for a structural reason: BNPL removes the payment friction that typically acts as a commitment device. A buyer who pays $150 immediately has made a financial decision that is complete. A buyer who commits to four $37.50 installments has made a lower-friction decision that is easier to reverse if the product doesn't meet expectations.
The refund dynamic compounds with the payment schedule. When a BNPL buyer returns an item, the merchant typically refunds the amount received from the BNPL provider, while the provider handles installment refunds to the buyer. The merchant's refund cash flow arrives days to weeks before the provider's schedule with the buyer, but the merchant already paid the BNPL fee on the original transaction. On returns, the BNPL fee is either partially returned or not at all, depending on the provider's agreement terms.
For a product with a 20% standard return rate that runs 28% return rate for BNPL buyers, the additional 8% in returns costs: reverse logistics + the BNPL fee that isn't refunded + any inventory that returns as non-resellable. On a $150 product at 28% return rate, modeled against the full unit economics, the realized revenue per BNPL order is lower than the headline conversion rate suggests.
Customer Quality: LTV Distribution
The most important metric for evaluating BNPL is repeat purchase rate in the 90–180 day post-purchase window, segmented by payment method. This data exists in your Shopify, BigCommerce, or platform analytics if you segment payment method at the order level — it is rarely examined because operators default to ROAS or conversion rate as the evaluation metric.
If BNPL buyers repeat-purchase at 60% of the rate of credit card buyers, and your business model is built on LTV through repeat purchase, BNPL acquisition cost is effectively higher than standard payment processing cost even at the same headline fee rate — because you're acquiring customers who contribute less lifetime value per acquisition.
This is not universal. In some product categories — high-consideration purchase items where the primary barrier to conversion is financial, not behavioral — BNPL reaches buyers who are fully committed to the purchase and simply lack the immediate liquidity. These buyers have normal return rates and repeat purchase behavior once the initial conversion barrier is removed. The question is which profile describes your BNPL buyer cohort.
The Cost Reality
A practical model for a $100M GMV merchant enabling BNPL at 10% checkout adoption:
| Metric | Without BNPL | With BNPL (10% adoption) | |---|---|---| | Total GMV | $100M | $103M* | | BNPL-attributed GMV | — | $10.3M | | Standard processing cost (2.5%) | $2.5M | $2.3M | | BNPL processing cost (4.5%) | — | $464K | | Net additional processing cost | — | $208K | | BNPL return rate premium (8% above base) | — | $824K additional returns | | Total net impact before LTV effects | — | –$1.03M |
*3% GMV lift assumed from incremental BNPL conversion, per middle-range incrementality estimate
The incremental $3M in GMV at 45% gross margin produces $1.35M in incremental gross profit. After the additional processing cost ($208K) and return cost ($824K), the net incremental contribution is $318K — or 0.3% of GMV. Whether this justifies the checkout complexity, the customer service load from BNPL-specific disputes, and the cash flow timing impact of installment reconciliation depends on whether that $318K is the right return for the operational investment.
The model changes significantly with a higher LTV premium for BNPL buyers or a lower incrementality (more cannibalizing existing payment volume).
The Trade-Off Map
Enable Broadly
Enabling BNPL across all eligible products maximizes the checkout optionality and captures the full potential conversion lift. The operational overhead is low — most BNPL providers integrate via standard Shopify payments or API. The downside is undifferentiated exposure to the full BNPL cost stack on every BNPL-eligible transaction, including low-incremental conversions.
This approach makes economic sense for operators with high-AOV products (above $200) where the incremental conversion lift is more likely to represent genuinely new demand, or where the product category has a natural purchase-deferral dynamic (seasonal, high-consideration, gift purchases).
Enable Selectively on High-AOV Products
The more capital-efficient path: enable BNPL on products above a threshold AOV — typically $100–$150 or higher — where the incremental conversion benefit is most likely to represent new demand rather than payment method substitution. At lower AOV, the conversion lift from BNPL is often driven by buyers for whom the cash outlay is genuinely not the barrier (they would have used a credit card), and BNPL is simply a lower-friction checkout they prefer — at the merchant's cost.
Selective enablement reduces BNPL fee exposure on routine purchases while preserving the benefit on the purchases where BNPL's role as a conversion enabler is most defensible.
Disable and Reallocate
If measurement shows BNPL buyers have return rates more than 10 percentage points above average and repeat purchase rates below 60% of non-BNPL buyers, the channel is consuming margin on customers who don't justify the acquisition cost. Disabling and reallocating the processing cost differential to retention marketing for existing high-LTV customers is often a higher-return use of the same dollars.
What Operators Get Wrong Most Often
The first mistake is measuring BNPL conversion lift without a holdout group. If BNPL is enabled for all checkout sessions simultaneously, you have no control group to isolate the incremental conversion rate. The lift measurement requires an A/B test where a segment of shoppers sees BNPL options and a segment doesn't — run concurrently on similar cohorts. Without this, the reported lift is meaningless.
The second mistake is ignoring the refund economics in the BNPL evaluation. Merchants typically calculate fee cost and conversion lift; they don't model the return rate differential or the fee treatment on refunds. For return-heavy categories (apparel, electronics, home goods), the return adjustment can reverse a positive conversion ROI calculation.
The third mistake is treating BNPL as permanent once enabled. Most operators who enable BNPL to address a short-term conversion pressure keep it running without revalidating the economics. Quarterly measurement of BNPL vs. non-BNPL cohort LTV and return rate should inform a decision to maintain, adjust threshold, or disable.
BNPL can lift conversion. It can also lower cash quality and raise refund cost on the same lift. The operators who measure all three variables before making the decision get a model they can act on. The ones who measure only conversion rate discover the full picture in the quarterly P&L review.
Segment your last 90 days of BNPL orders against the same-period non-BNPL orders. Compare refund rate and 60-day repeat purchase rate. The decision is in that comparison.
AHAeCommerce is an independent eCommerce decision intelligence platform. No affiliate relationship influences this analysis. Drafted with AI assistance. Edited and claim-tested by Diosh.
Sources: Adobe Analytics, "2025 Holiday Shopping Season" — news.adobe.com/news/2026/01/adobe-holiday-shopping-season; see also: Payment Processing: The Margin Drain Nobody Audits, Customer LTV: What the Number Actually Means
