Buy now, pay later costs a merchant somewhere between two and four times what a card transaction costs, and the entire business case for accepting it rests on whether the sale would have happened anyway. Published rate cards from the major providers cluster between roughly 2% and 6% of the order value plus a fixed per-transaction fee, against blended card acceptance costs that most US online retailers see in the 2% to 3% range. That gap is not a rounding error. On a $150 basket it is the difference between paying about $4 and paying about $9 for the same checkout, so BNPL merchant fees vs credit card costs only make sense as a question about incremental revenue, not about the fee line alone.
In short
- BNPL merchant discount rates published by Klarna, Afterpay and Affirm typically sit between roughly 2% and 6% of order value plus a per-transaction fixed fee, though rates are negotiated and change; verify current figures directly with each provider.
- Card acceptance for a typical US online retailer blends to roughly 2% to 3%, with regulated debit capped under the Federal Reserve’s Regulation II and digital wallets such as Apple Pay adding no wallet-specific fee on top of the underlying card cost.
- Provider claims of 20% to 40% average order value uplift are directional, not causal; a large share of BNPL orders are substitution from cards, and the only way to know the share for a specific store is a controlled incrementality test.
- The fee gap is covered when incremental gross margin from BNPL-attributable orders exceeds the extra cost of the substituted orders; in practice that works in categories with high margins, higher baskets and a young, credit-thin customer base.
- Hidden costs (refund fee retention, longer settlement, dispute handling and chargeback-equivalents) can add a further 0.3 to 1 percentage point to the effective rate and are the first thing to negotiate once volume is established.
What merchants actually pay across BNPL, cards and wallets
The clearest way to see the difference is to line up what each acceptance method takes from a single order. The figures below are drawn from publicly available rate cards and the providers’ own disclosures as of mid-2026, and they describe list pricing rather than negotiated pricing. Every one of them moves with volume, category risk and contract term, so treat the table as a map of relative cost, not a quote.
| Acceptance method | Typical merchant cost (US, list) | Cost on a $150 order | Who sets the price |
|---|---|---|---|
| Regulated debit card (bank over $10bn in assets) | $0.21 + 0.05% + $0.01 fraud adjustment interchange, plus processor markup | About $0.60 to $0.90 | Federal Reserve cap under Regulation II, processor adds markup |
| Credit card (Visa, Mastercard) | Roughly 1.5% to 3.5% interchange plus network and processor fees | About $3 to $5 | Card networks publish interchange, processors add margin |
| Standard payment processor (blended) | Around 2.9% + $0.30 (Stripe’s published US online rate) | About $4.65 | Processor list price |
| Apple Pay / Google Pay | No wallet fee to the merchant; underlying card rate applies | Same as the card behind the wallet | Wallet provider takes a fee from the issuer, not the merchant |
| PayPal checkout | Around 3.49% + $0.49 (PayPal’s published US standard rate) | About $5.70 | PayPal list price |
| Afterpay (pay in 4) | Published as up to 6% + $0.30 per transaction | Up to about $9.30 | Provider, negotiable at volume |
| Klarna (pay in 4 and financing) | Reported in the roughly 3.3% to 6% + $0.30 range depending on product | About $5.25 to $9.30 | Provider, negotiable at volume |
| Affirm (monthly financing) | Reported in the roughly 2% to 8% range depending on term and promotional APR | About $3 to $12 | Provider, heavily negotiated |
Two things stand out. First, BNPL is not a single price; a merchant-funded 0% APR financing offer over 12 months on Affirm costs far more than a pay-in-4 on the same platform, because the provider is giving up the consumer interest it would otherwise earn.
Second, digital wallets are essentially free at the margin. Apple Pay and Google Pay pass through the underlying card interchange and take their cut from the card issuer, which is why they are often the cheapest way to lift mobile conversion without changing the cost structure at all. For a fuller picture of how these rails fit together, the wider overview of how retail payments are changing across cards, BNPL and crypto sets out the landscape that this comparison sits inside.
How the debit cap works
The Federal Reserve’s Regulation II, which implements the Durbin Amendment of the Dodd-Frank Act, caps debit interchange for card issuers with more than $10 billion in assets at $0.21 plus 0.05% of the transaction, with a $0.01 fraud-prevention adjustment for qualifying issuers. The Board proposed lowering that base figure in late 2023, and the status of any revision needs to be checked against the Federal Reserve’s Regulation II page because the rulemaking has moved slowly. The practical point for retailers is that regulated debit is almost always the cheapest tender a customer can present, and a BNPL order that would otherwise have been a debit order is the most expensive substitution a store can make.
Why blended card rates hide the spread
Most mid-sized retailers see one blended rate from their processor, which averages cheap debit and expensive rewards credit into a single number. That number is useless for judging BNPL, because BNPL customers skew toward the debit and credit-thin end of the mix. The right comparison is the cost of the specific tender the BNPL order displaced, usually debit or a basic credit card, against the BNPL fee.
Why BNPL providers charge what they charge
BNPL providers charge more because they carry costs that card networks push elsewhere. A pay-in-4 provider funds the full purchase price to the merchant within days, then collects four installments over six weeks, so the provider is extending short-term unsecured credit at scale, absorbing default risk, and running its own underwriting, collections and customer service. Card issuers do all of that too, but they recover it from cardholder interest and annual fees; most pay-in-4 products carry no consumer interest, so the merchant fee is the primary revenue line.
The Reserve Bank of Australia, which studied BNPL merchant costs closely in its 2021 review of retail payments regulation, reported that BNPL cost Australian merchants several times what card acceptance cost, with pay-in-4 fees around 4% against roughly 0.5% to 1% for debit and credit. The RBA also concluded that the no-surcharge rules most BNPL providers imposed on merchants were a policy concern precisely because merchants could not signal the cost difference to shoppers. The findings are Australian and dated, but the cost structure is the one US providers operate today.
Funding cost and loss rates
Two variables drive most of the fee. The first is the provider’s cost of funds, which rose sharply through 2022 and 2023 as interest rates climbed, and which sets a floor under what the provider must earn per loan. The second is the loss rate on the installment book; Affirm’s public filings disclose provisions for credit losses that run in the low single digits as a share of gross merchandise volume, and pay-in-4 books at other providers are understood to run similar or higher rates on younger customers. A 4% to 6% merchant fee on a six-week loan is, in annualized terms, an extremely high yield, but it is the yield the provider needs to cover losses, funding, fraud and a marketing engine that sends shoppers into the merchant’s app.
The marketing component of the fee
Providers argue, with some justification, that part of the fee buys distribution. Klarna, Afterpay and Affirm each operate consumer apps with tens of millions of users and in-app storefronts that route shoppers to partner merchants. The strategy Klarna and its rivals are pursuing, from shopping apps to bank charters, is covered in detail in the piece on the BNPL playbook for retail in 2026. Whether that distribution is worth paying for is a question a merchant can answer from its own referral data: if a meaningful share of BNPL orders originates from the provider’s app rather than from the merchant’s own site, the fee is partly a customer acquisition cost and should be compared to paid social, not to interchange.
Basket uplift and conversion lift: measured, not claimed
The headline claims are familiar. Providers routinely cite average order value uplift of 20% to 40% and conversion lift of 10% to 30% for merchants that add their button. Those figures are real in the sense that they describe what happened to merchants who adopted BNPL, but they are not causal, because the merchants who adopt BNPL are not a random sample and the customers who choose BNPL are not the same as the customers who choose a card.
Three biases inflate the claimed numbers. Selection bias: shoppers who pick a BNPL option at checkout already had a larger basket, so comparing BNPL baskets to card baskets measures who chose it, not what it caused. Substitution bias: a customer who would have paid with a debit card and now pays with pay-in-4 counts as a BNPL order in the provider’s dashboard but produced no incremental revenue and cost the merchant several dollars more. Timing bias: BNPL launches often coincide with checkout redesigns, new wallets and promotions, and the provider’s before-and-after chart absorbs all of them.
What the independent evidence says
Independent research paints a more modest picture. Federal Reserve surveys of BNPL users, including work from the Federal Reserve Bank of Boston and the Consumer Financial Protection Bureau’s 2022 and 2023 market reports, found that BNPL is used disproportionately by younger consumers, those with lower credit scores and those who report having been declined for other credit. That profile supports the view that some BNPL demand is genuinely new, because those customers could not have put the purchase on a credit card. It also supports the substitution view, because the same customers have debit cards and use them.
Academic work on installment credit, including studies of Affirm data by university researchers, has generally found positive but smaller effects than the marketing figures: measurable increases in spending among users, concentrated in discretionary categories, with a large share of orders that would plausibly have happened anyway. A more detailed reading of that evidence is in the analysis of how BNPL actually affects retail conversion rate, which separates checkout-completion effects from basket-size effects. The short version is that a merchant should expect real but partial incrementality and should plan to measure it rather than assume the provider’s number.
Running a clean incrementality test on BNPL
The only way to know what BNPL is worth to a specific store is to run a controlled test in which some shoppers see the BNPL option and others do not, and then compare revenue and margin across the two groups rather than comparing BNPL orders to card orders. This is a standard holdout design, and most checkout platforms can support it with a feature flag or an A/B tool at the payment-method step.
- Define the unit of randomization. Randomize at the visitor or customer level, not the session level, so that a shopper does not see BNPL on one visit and not on the next. Persist the assignment in a cookie or customer record.
- Hold out a meaningful share. A 50/50 split gives the fastest read; a 90/10 split protects revenue if BNPL is already large but needs several times longer to reach significance. Run for at least two full purchase cycles for the category.
- Measure the right outcomes. Track conversion rate, average order value, revenue per visitor and, critically, gross margin per visitor after payment costs. Revenue per visitor captures both conversion and basket effects in one number.
- Attribute the payment cost per group. Apply actual fees by tender to each order in each group. The test group will show higher payment costs; the question is whether it also shows higher margin after those costs.
- Check for cannibalization of returns. Compare return rates and refund volume by group over a window long enough to capture the return policy, because BNPL orders in some categories return at higher rates.
- Look at repeat behavior. If BNPL customers acquired through the provider’s app come back and pay by card on the second order, part of the fee was acquisition and the lifetime view will look better than the first-order view.
Reading the result
Suppose the holdout shows revenue per visitor of $4.20 with BNPL and $3.90 without, a 7.7% lift, on a store with a 45% gross margin. The lift is worth about $0.135 in gross margin per visitor. If BNPL orders in the test group are 15% of revenue and carry a fee 3 percentage points above the displaced card cost, the extra payment cost is roughly $0.019 per visitor. The test clears comfortably.
Change the margin to 25% and the BNPL share to 35%, and the lift is worth $0.075 against extra cost of about $0.044; still positive but no longer comfortable once refund retention and settlement drag are added.
Common ways the test goes wrong
The most frequent error is measuring BNPL orders against card orders inside the same group, which reintroduces every bias the holdout was designed to remove. The second is stopping early because the BNPL group looks good after a week; payment-method effects are small relative to daily noise, and a test that has not reached its pre-set sample size is not a result. The third is forgetting that the provider’s app may drive traffic that never enters the test; that traffic is real value but belongs in a separate acquisition-channel measurement.
Categories where the maths works and where it does not
Whether the fee gap is covered depends on three things: gross margin, basket size relative to the customer’s comfortable one-off spend, and how credit-constrained the customer base is. Categories that score high on all three can absorb a 5% fee easily; categories that score low on any of them usually cannot.
| Category | Typical gross margin | Basket profile | Incrementality tendency | Fee gap covered? |
|---|---|---|---|---|
| Fashion and footwear (D2C) | 50% to 65% | $80 to $250, discretionary | Moderate to high; young, credit-thin buyers | Usually yes |
| Beauty and skincare | 60% to 75% | $40 to $120, repeat purchase | Moderate; strong app-driven discovery | Yes, especially on bundles |
| Consumer electronics | 10% to 25% | $200 to $1,500, considered | High on financing terms, low on pay-in-4 | Only with negotiated rates or vendor-funded promos |
| Furniture and home | 35% to 50% | $300 to $2,000, considered | High on longer financing | Yes for monthly plans, marginal for pay-in-4 |
| Grocery and consumables | 20% to 30% | $50 to $150, habitual | Low; almost pure substitution | Rarely |
| Travel and experiences | 10% to 20% (agency) | $300 to $3,000 | Moderate; timing-driven | Marginal, watch refund rules |
| Fitness equipment and outdoor | 30% to 45% | $150 to $1,200 | High; purchase often deferred otherwise | Usually yes |
The pattern is that BNPL earns its fee when it removes a genuine affordability barrier for a purchase the customer would otherwise defer or downsize. It fails when the purchase is habitual and the customer would have paid anyway: the option is used, the fee is paid, and the baskets look exactly like the debit baskets they replaced.
Electronics and the vendor-funded promotion
Consumer electronics deserves a note because it is the category where BNPL fees are highest and margins thinnest, yet it is also where financing is most established. The way the economics close is that the brand, not the retailer, often funds the 0% APR promotion; Apple, Samsung and Peloton have all subsidized installment offers through Affirm and others. A retailer evaluating a 0% APR offer should ask who is paying the merchant discount rate before assuming it will land on its own P&L.
Negotiating rates once volume is established
List rates are a starting point. BNPL providers negotiate on rate, on fixed fee, on settlement timing and on refund treatment, and they negotiate hardest with merchants who can credibly threaten to shift volume to a competitor or to turn the button off. The three major US players are compared on pricing posture, underwriting and merchant tools in the piece on Klarna, Afterpay and Affirm compared for US merchants, and the differences in how each one approaches enterprise pricing are material.
Leverage usually arrives at three thresholds. Below roughly $1 million in annual BNPL volume, most merchants are on standard pricing through a platform integration and have little room to move. Between $1 million and $10 million, providers will typically discuss tiered rates, reduced fixed fees and faster settlement in exchange for placement commitments such as showing the provider’s messaging on product pages. Above $10 million, custom contracts are normal and the conversation covers exclusivity, co-marketing budgets and, for larger brands, revenue share on the provider’s app traffic.
What to ask for, in order
- Rate tiering by product. Pay-in-4 and long-term financing should be priced separately, and a merchant that mostly sells sub-$200 items should not pay a blended rate that reflects financing costs it never uses.
- Fixed fee reduction. The $0.30 per transaction matters far more on a $40 basket (0.75 points) than on a $400 basket (0.075 points); low-ticket merchants should push on the fixed fee first.
- Refund fee return. Ask that the merchant discount rate be refunded in full on full refunds within the return window; some contracts retain part or all of the fee.
- Settlement timing. Move from T+3 or longer to T+1 or T+2; the working-capital cost is small but real, and it is often conceded easily.
- Dispute handling terms. Clarify who bears the loss on item-not-received and not-as-described claims and how evidence is submitted.
Using a second provider as leverage
Running two BNPL providers is operationally messy, but the credible option to do so is the most effective negotiating tool a mid-sized merchant has. Several checkout platforms now let a merchant route by basket size, sending sub-$200 orders to one provider and financing-eligible orders to another, so the providers compete on the segment each one wants.
Hidden costs: refunds, disputes and settlement timing
The published rate is not the effective rate. Several smaller costs accumulate around BNPL orders that do not appear on the rate card and that can add between 0.3 and 1 percentage point to the true cost of acceptance, depending on the category’s return rate and the contract’s refund terms.
Refund fee retention is the largest. If a contract allows the provider to keep the merchant discount rate on a refunded order, a category with a 25% return rate is effectively paying the fee on a quarter of its BNPL revenue twice, once on the sale and again in lost recovery. Dispute handling is the second: BNPL providers run their own dispute processes rather than the card networks’ chargeback system, and the evidence standards, timelines and fee structures differ from what a merchant’s fraud team is used to. The mechanics of that are laid out in the piece on BNPL refunds and disputes: what happens when an order goes wrong, and the practical advice is to model dispute cost per order by provider, not to assume it matches card chargebacks.
Settlement and cash timing
Card settlement for most US online merchants lands in one to two business days. BNPL settlement varies from next day to several days, and some providers batch weekly for smaller merchants. At a 10% cost of capital, three extra days on $5 million of annual BNPL volume is worth roughly $4,000 a year, which is small but is also nearly free to negotiate away. More important than the interest cost is the reconciliation load: BNPL payouts arrive net of fees, refunds and adjustments, and finance teams that do not build a reconciliation process for each provider end up carrying unexplained variances for months.
Regulatory and compliance overhead
Regulation of BNPL has moved unevenly across markets, and the merchant-side implications are still settling. In the US, the Consumer Financial Protection Bureau issued an interpretive rule in May 2024 stating that pay-in-4 BNPL lenders are card issuers for purposes of Regulation Z’s dispute and refund provisions; the Bureau subsequently signaled in 2025 that it would deprioritize enforcement of that rule, and its current status should be checked directly with the CFPB. In the UK, the Financial Conduct Authority’s regulation of BNPL took effect in July 2026, bringing affordability checks and dispute rights that providers have said may reduce approval rates. Merchants selling into those markets should expect providers to pass some of the compliance cost through and should read any contract amendments carefully.
This section is general information about how BNPL and card acceptance costs are structured and is not legal, tax or financial advice. Fee rates, regulatory thresholds and settlement terms change and vary by contract; any merchant evaluating specific terms should verify current figures with the provider and consult a licensed payments consultant, attorney or accountant about its own situation.
How to decide: a working framework
The decision reduces to a comparison a finance lead can run in a spreadsheet once the incrementality test has produced a number. The inputs are measured revenue lift per visitor, gross margin, the BNPL share of orders, the fee spread against the displaced tender, the return rate and refund treatment, and the acquisition value of provider-app traffic. The output is margin per visitor with BNPL versus without it.
Three rules of thumb fall out of running that model across typical inputs. At gross margins above 50%, almost any measured lift above 3% to 4% covers a 5% fee even with pessimistic substitution. At gross margins between 25% and 50%, the case depends heavily on the BNPL share of orders and on refund terms; a merchant in this band should negotiate before launching, not after. Below 25% gross margin, pay-in-4 is very hard to justify on first-order economics, and the case has to rest on financing for high-ticket items, vendor-funded promotions or a demonstrated lifetime-value effect.
The broader shift in how shoppers pay, from cards to wallets to installments to account-to-account rails, is not going to reverse, and the strategic overview of how retail payments are changing across cards, BNPL and crypto makes the case that merchants will need to offer more tenders, not fewer. The discipline is to pay for the ones that produce incremental margin and to negotiate hard on the ones that do not.
FAQ on BNPL merchant fees
How much do BNPL providers charge merchants compared with credit cards?
Published list rates for pay-in-4 products from Klarna, Afterpay and Affirm generally fall between roughly 2% and 6% of the order value plus a fixed fee of around $0.30, with longer financing terms and merchant-funded 0% APR offers priced higher. Card acceptance for a typical US online retailer blends to about 2% to 3%, and regulated debit is far cheaper. As a rule of thumb, BNPL costs two to four times what a card order costs, but rates are negotiated and change, so current pricing should be confirmed directly with each provider.
Do Apple Pay and Google Pay cost merchants more than cards?
No. Apple Pay and Google Pay do not charge the merchant a wallet-specific fee. The transaction runs on the card stored in the wallet, so the merchant pays the same interchange and processor fees it would pay on that card presented directly. That makes wallets the cheapest way to improve mobile checkout conversion, because they lift completion rates without changing the merchant’s cost of acceptance.
Is the 20% to 40% average order value uplift from BNPL real?
The figures describe what providers observe across their merchant base, but they are not causal. Shoppers who pick BNPL already had larger baskets, many BNPL orders replace debit or credit orders that would have happened anyway, and BNPL launches usually coincide with other checkout changes. Independent research finds positive but smaller effects, concentrated in discretionary categories and among younger, credit-constrained customers. The only reliable number for a specific store comes from a holdout test.
How do you run an incrementality test for BNPL?
Randomize visitors into a group that sees the BNPL option at checkout and a group that does not, persist the assignment across sessions, and run for at least two full purchase cycles or until a pre-set sample size is reached. Compare revenue per visitor and gross margin per visitor after actual payment costs across the two groups. Do not compare BNPL orders to card orders within the same group, because that reintroduces the selection bias the test exists to remove. Track returns and repeat purchases for both groups.
Which retail categories benefit most from offering BNPL?
Categories with high gross margins, discretionary baskets in the $80 to $500 range and a younger customer base see the clearest benefit: fashion, footwear, beauty, fitness equipment and outdoor gear. Furniture and consumer electronics benefit from longer financing terms rather than pay-in-4, often with the brand funding the promotion. Grocery, consumables and other habitual purchases see mostly substitution, where the fee is paid for sales that would have happened on a debit card anyway.
Can merchants negotiate BNPL fees?
Yes, and they should. Below about $1 million in annual BNPL volume most merchants are on standard platform pricing, but above that providers will discuss tiered rates by product, lower fixed fees, faster settlement and full refund of the merchant discount rate on returned orders. The strongest leverage is a credible plan to route volume to a second provider or to switch the button off. Merchants should negotiate before a launch or a renewal, not after volume has been committed.
What hidden costs come with BNPL acceptance?
The main ones are retention of the merchant fee on refunded orders, which hits high-return categories hardest; dispute processes that differ from card chargebacks and can carry their own fees and evidence rules; slower settlement than cards, which adds a small working-capital cost; and reconciliation load, because payouts arrive net of fees and adjustments. Together these typically add 0.3 to 1 percentage point to the effective rate, depending on category and contract.
Can a merchant surcharge BNPL orders to recover the fee?
Most BNPL contracts prohibit surcharging or otherwise steering customers away from the BNPL option, and providers have historically enforced those clauses. Regulators in Australia flagged no-surcharge rules as a concern in the Reserve Bank of Australia’s 2021 payments review. In the US, card surcharge rules vary by state and by network, and they do not automatically extend to BNPL. Any surcharging plan should be checked against the specific provider contract and against state law, ideally with legal advice.
What to read next
The fee comparison is one input into a larger decision about which tenders a store should offer. The piece on the BNPL playbook for retail in 2026 covers where the providers are heading and what that means for merchant leverage, and the deeper look at how BNPL actually affects retail conversion rate goes further into the evidence on incrementality that this analysis leans on.