Summary: Buy now, pay later (BNPL) entered US retail as a big-ticket financing tool, and most retailers still scope it that way. Marqeta platform data shows the average order value on BNPL transactions fell 9% year over year between January and May 2025, a shift toward smaller, non-discretionary purchases. That changes which retailers should care. BNPL exposure is now a function of how often customers visit, not how much they spend per visit.
Consider this example: a regional grocery chain evaluated BNPL in 2023 and passed on it. The reasoning was sound on its own terms: the average basket ran under $60, financing a week of groceries looked like a solution without a problem, and a per-transaction fee was difficult to justify in a category that already operates on thin margins.
Three years later, its customers are using BNPL on those baskets anyway. They are doing it through a third-party app, on a credential the grocer does not own, with the repayment relationship, the purchase data, and the brand impression sitting somewhere other than the store. The chain still has no BNPL strategy, because as far as its own systems are concerned, none of this is happening.
The 2023 decision was not a mistake. The terms it was made under have changed.
The assumption retail built its BNPL strategy on
BNPL arrived in US retail attached to a specific kind of purchase. Furniture, electronics, appliances, and travel were the anchor categories, and the anchor use case was the $900 mattress or the $1,300 laptop that a shopper did not want to put on a credit card in full. Retailers with high average order values built BNPL into checkout. Retailers with low ones concluded, reasonably, that it did not apply to them.
Nearly every scoping decision retail has made about BNPL since has inherited that framing. The problem is that the framing describes the category as it existed several years ago, and the behavior underneath it has moved.
What the transaction data shows
The clearest signal is not survey sentiment. It is volume. According to Marqeta's 2025 State of Payments Report, the average order value on BNPL transactions between January and May 2025 fell 9% against the prior year, a decline the platform data attributes to a shift toward non-discretionary, smaller-ticket purchases.
Average order value falling is not a sign of a category cooling off. It is a sign of a category broadening. The same report finds that within a single 30-day window, 77% of US consumers used a credit card and 69% used a debit card, alongside cash and P2P apps. Consumers are not selecting one method and holding it. They are running several concurrently and moving between them by purchase.
Marqeta's 2026 State of Credit Report shows the same pattern extending into everyday categories, with non-discretionary BNPL volume growing sharply year over year as consumers apply installments to groceries and food delivery.
The strategic implication is worth stating plainly. If BNPL usage is concentrating in smaller, repeat purchases, then a retailer's exposure to it scales with visit frequency rather than basket size. That inverts the original scoping logic. The categories that ruled themselves out of BNPL because their baskets were too small are now the categories where the behavior occurs most often.
Consumers are budgeting, not struggling
There is a version of this argument that no retail brand should want to make, which is that its customers are financing essentials because they cannot afford them. The research does not support that reading, and getting the distinction right matters for how a retailer positions any product it builds here.
The 2025 State of Payments Report finds that 23% of US consumers adjusted their payment behavior in response to economic pressure. That is a real cohort, and it is a minority one. The broader finding in the same research is that consumers are becoming more strategic about payment methods generally, switching between them to maximize spending power rather than to survive a shortfall.
What the data describes is a portfolio, and each method in it holds a job:
- Debit does spending control
- Credit does rewards and float
- BNPL does cash-flow timing, a scheduling function rather than a borrowing one
BNPL turns out to be as useful on a $70 grocery order landing three days before payday as on a $900 sofa.
A retailer accommodating that behavior is not extending distress credit. It is meeting a budgeting preference its customers have already adopted somewhere else.
Why the app wins a transaction your card should have
The 2026 State of Credit research points to a limit problem rather than a preference problem. Among consumers without a credit card, 23% turn to BNPL specifically when they cannot pay in full, using it to finance a purchase without taking on revolving debt. The report finds a comparable pattern among cardholders whose available credit does not accommodate the purchase in front of them.
That is a product gap sitting inside your own program. The customer wanted to transact, the card in their wallet could not do the job in that moment, and a third party was positioned to do it instead. The gap concentrates among younger consumers, who skew heavily toward BNPL usage and are the same cohort your credit program is least likely to have approved in the first place.
Each of those transactions was available to you. None of them reached you.
What staying out of it actually costs
The direct cost of third-party BNPL is well understood as a per-transaction fee, and retailers who offer it already model that line. What is less understood is what happens to the model when the behavior changes shape.
A fee scoped against occasional high-value purchases is one kind of expense. The same fee scoped against a behavior that recurs weekly is a materially different line item, and most retailers have not re-forecast it since the underlying frequency shifted.
The second cost does not show up on the P&L at all. Everyday, non-discretionary purchases are the highest-frequency and most predictive data a retailer can hold, because they reveal cadence, category mix, household composition, and budgeting behavior in a way an annual big-ticket purchase never will. That is precisely the set of transactions now routing through someone else's credential. The retailer sees a completed sale. It does not see who financed it, on what terms, or what that says about the customer.
For retailers who have declined to offer BNPL entirely, there is a third cost, which is the visit itself. Consumers who expect payment flexibility and do not find it will transact where they do find it, and that decision is made before the basket is ever built.
What a flexible credential changes
What is a flexible credential? A flexible credential is a single payment card that lets the cardholder choose between debit, credit, and BNPL at the moment of purchase. Rather than carrying separate cards and separate apps for each payment type, the customer selects how to pay at checkout on the same credential.
How companies build cards that let users choose how to pay at checkout:
- Unified card infrastructure:The issuer-processor platform supports multiple funding sources (debit, credit, installment) on a single card program and BIN
- Real-time decisioning:At authorization, the system presents available payment options based on the cardholder's eligibility, limits, and preferences
- Cardholder selection:The customer chooses their payment method via mobile app, digital wallet prompt, or pre-set rules before or during checkout
- Program management integration:The card program manager coordinates underwriting, compliance, and funding across all payment types within one credential
Applied to the behavior described above, the mechanics are direct.
Financing sits on the card the customer already carries, so it is available on a $70 basket as readily as a $700 one, without a separate app, a separate approval, or a handoff at checkout. The retailer accommodates the budgeting behavior instead of outsourcing it, which means the fee it was paying a third party becomes program economics it participates in. And because the credential is yours, the everyday transaction data stays with you, including the purchases customers make outside your stores.
This is also the answer to the scoping problem the grocery chain ran into. A flexible credential does not require a retailer to decide in advance which baskets deserve financing, because that decision moves to the customer, transaction by transaction. As we noted in an earlier post, State of Payments research shows 60% of consumers aged 25 to 44 want this financing inside a card they already carry rather than in a separate app. The demand is not for a new checkout option. It is for the option to live on the card.
The behavior already moved
BNPL did not stay in the categories retail assigned to it. It moved into the weekly shop, the delivery order, and the mid-month top-up, and it did so without requiring permission from anyone's category strategy.
The relevant question for a retailer is no longer whether its baskets are large enough to warrant financing. Customers have already answered that. The question is whether the financing happens on your credential or someone else's, and whether you can see it at all.
Marqeta powers flexible credentials and co-brand programs that bring debit, credit, and BNPL onto a single card, so retailers can meet everyday payment behavior inside their own brand experience. If your BNPL strategy was scoped to a different version of this category, it is worth revisiting.
Contact us to explore what a flexible credential could do for your program.
