BNPL merchant fees and break-even calculator

The structure is the same as a card, a percentage of the order plus a fixed fee, and only the percentage differs. Published US merchant rates run from 4.5 percent plus 30 cents to 7.99 percent plus 30 cents, against 2.9 percent plus 30 cents on a typical flat rate card account. On a $120 order that is $5.70 to $9.89 against $3.78. The multiple moves with ticket size, which is why the BNPL Merchant Cost Calculator prices the basket you actually sell: Klarna is 1.75 times the card fee on a $25 order, 1.98 times on $120 and 2.04 times on $400, as the shared 30 cent fixed fee dilutes.

The awkward part of researching this is that most providers publish no rate card. Klarna's own developer documentation says the rate is returned per transaction by its Price Plans API and is dynamic, so there is nothing to read, and Affirm and Zip quote on application. Sezzle is the exception: 6.1 percent plus 30 cents in its own US merchant agreement, with a caveat that the rate can be higher by industry classification. PayPal is the other, publishing Pay Later at 4.99 percent plus $0.49 against 3.49 percent plus $0.49 for PayPal Checkout, which puts the BNPL premium on that platform at exactly 150 basis points. The rest of the rates below are real published US prices, just published by somebody else: Stripe prints a rate for every BNPL method it resells, and Square prints Afterpay at 6 percent plus 30 cents beside its own 2.6 percent plus 15 cent card rate. Treat those as the list price, because a direct contract at volume can be lower.

One trap before the arithmetic starts. A lower percentage is not automatically cheaper, because the fixed fees are not equal. PayPal Pay Later charges 49 cents where every other method here charges 30. On a $15 basket that makes Pay Later $1.24 and Klarna $1.20, so the lower rate is the more expensive method. The two are identical at $19 and the ranking flips either side of it.

BNPL is not sold as a payment method, it is sold as a growth channel. The pitch is more conversions and bigger baskets, and the price is roughly double a card. So the question is not what it costs. It is how much new revenue it has to create before it stops being a voluntary discount on sales you already had.

The arithmetic has five terms and no mystery in it. Take the extra fee per switched order, the BNPL fee minus the card fee, and multiply by the number of orders that switched. That is the deficit. Then take the contribution one genuinely new order makes, the order value times gross margin less the BNPL fee. Divide the deficit by that contribution and you have the extra orders a month the method must create. On the defaults the deficit is $400.68, a new order contributes $46.51, and the answer is 8.61 orders, which rounds up to nine. Against 1,164 baseline orders that is a lift of 0.74 percent.

That number is low. At a 45 percent gross margin, a BNPL method that creates almost any incremental demand at all pays for itself. The margin does the work: a new $120 sale carries $54.00 of contribution and the fee takes only $7.49 of it, so a handful of extra orders covers a lot of cannibalized ones. That is the honest case for BNPL.

Which is why the margin is the input to stress first, not the fee. Run the same scenario at a 25 percent gross margin and the break-even share nearly doubles to 14.15 percent of BNPL orders. At 15 percent margin it is 26.09 percent, the required basket uplift is 34.30 percent, and the month turns net negative at the entered mix. Underneath all of it is a hard floor: when gross margin is at or below the BNPL rate, every order on that method loses money whether it is new or not.

Here is the mistake. A customer arrives at your checkout intending to buy, sees a BNPL button, and uses it. You have not gained a sale. You have taken the sale you were already going to make and paid roughly twice the fee to process it. That order belongs in the cost column, and on the defaults there are 108 of them a month against 36 genuinely new ones.

Almost every published BNPL return figure gets this backwards, in the direction that flatters the seller. Compute the margin on all 144 BNPL orders instead of the 36 incremental ones and the same month returns $6,697.44 instead of $1,273.68. That is a factor of 5.3, both look reasonable, and nothing tells you which one you are reading. The Consumer Financial Protection Bureau, in Buy Now, Pay Later: Market Trends and Consumer Impacts, describes lenders deploying models, product features and marketing campaigns to increase the likelihood of incremental sales. Incremental is a word to earn, not assume.

The vendor uplift statistics have the identical flaw. Klarna's business site claims a 40 percent increase in average order value and a 20 percent increase in conversion. Afterpay's US retailer page claims 58 percent, attributed to internal Afterpay data. Sezzle claims 50 percent or more and Zip a 46 percent average order value lift. PayPal claims 93 percent higher average order value for enterprises and 62 percent for small businesses, both footnoted to its own internal data. Not one is a controlled measurement. Every one compares people who chose BNPL against people who did not, which tells you who self-selects into installments and nothing about what would have happened without the button.

There is one way to find your own number and it is not a spreadsheet. Turn the method off for a randomized share of traffic, or off entirely for a fixed period against a comparable one, and measure total orders and total revenue rather than BNPL orders and BNPL revenue. If total orders do not move, incrementality is zero. The 25 percent default here is a guess, not a measurement.

The genuine structural advantage of BNPL is that the provider buys the receivable. You are paid for the order and the provider chases the installments, so a shopper who stops paying is the provider's loss. Klarna states that you always get paid upfront and in full and that it assumes credit and fraud risk. Afterpay states that the money arrives in one to two business days. Zip's loans are originated by WebBank subject to credit approval, so a missed repayment is the lender's exposure, and Affirm carries the credit risk on the loan it underwrote. That is a real transfer of risk and it is worth something.

It is narrower than the marketing implies, and the boundary is the same on every product: the provider takes the shopper's ability to pay, the merchant keeps everything about the goods. Adyen's integration guidance is the clearest public statement of it. On Afterpay, the provider performs risk checks on the shopper and, where they succeed, takes on the full fraud risk for the payment, but the chargeable dispute reasons remain product not received, product unacceptable and credit not processed, and the merchant pays the dispute fee on those regardless of outcome. Shoppers can raise a chargeback with Afterpay up to 120 days after the payment, the cancellation, or the expected receipt of the goods. On Klarna, the same guidance says Klarna assumes the risk of non-payment only if the merchant meets Klarna's order fulfillment rules, and can charge the payment back where it does not.

Three published terms decide the rest, and none is visible in a rate. Dispute fees: on Stripe a lost Klarna dispute costs $15.00 against $3.00 on Zip, a fivefold gap on an otherwise similar product. Refunds: Sezzle charges 2 percent plus 30 cents to initiate one and retains the original processing fee, so a returned item costs the fee twice. Settlement: Sezzle publishes one to seven business days after order capture against Afterpay's one to two, and Affirm and Zip publish no window at all.

Offer it when the arithmetic clears with room to spare. Higher margin categories are the natural fit, because the fee is a smaller share of contribution and the break-even lift is tiny. Considered purchases with a real affordability barrier are the other fit: furniture, equipment, electronics, travel, elective health and dental, anything where the ticket is large enough that a shopper genuinely abandons rather than downgrades. That is where the incremental share is plausibly high, because there is a specific reason the sale would not otherwise happen.

Do nothing when your gross margin is thin, when your average ticket is small, or when you cannot say why a customer would fail to buy without installments. A grocer at a 5 percent gross margin offered a 5.99 percent method is not facing a marginal decision, they are facing a method that loses money on every order. A $25 ticket makes the effective rate 7.20 percent at Klarna's published rate, and nobody abandons a $25 cart for want of four payments of $6.25. In both cases the honest output is to leave things alone, and this calculator will say so.

There is a middle path that gets skipped: BNPL does not have to be on for everything. The fee is a percentage, so the cost concentrates on your largest orders, and the affordability argument is strongest there too. That makes a minimum order threshold the highest leverage setting most providers offer, because turning the method on above a floor cuts the cannibalization on small everyday baskets, where incrementality is lowest.

And do the boring comparison first, because it usually wins. On the worked example the extra fee bill from turning BNPL on is $400.68 a month at 25 percent incrementality and $534.24 at zero. Shaving 30 basis points off the card rate on the same $139,680 of monthly card volume is $419.04 a month, with no incrementality assumption and nothing to measure. The two decisions compete for the same money, and only one of them needs a holdout test to prove.