Buy Now, Pay Later Statistics 2026: The honest US numbers

Buy Now, Pay Later statistics in 2026 are a minefield of numbers that look precise and aren’t comparable. This roundup keeps every figure US-scoped where the data is strongest, attaches a named source and publication date to each one, and flags the widely-repeated stats that fall apart under scrutiny — including the “global market size” number that doesn’t really exist.

The honest headline is a gap. US BNPL transaction value reached an estimated $70 billion in 2025 — real money, but only about 1.1% of what Americans put on credit cards. Meanwhile 47% of users say they paid late on a BNPL loan in the past year, a figure that sits far above the sub-2% delinquency rate the industry reports. Both numbers are “true.” They just measure different things, and most coverage never runs them side by side.

Below: what the Richmond Fed and the CFPB actually measured, who is borrowing, the three risk metrics that disagree by design, the vendor average-order-value claims ranked by how much evidence backs them, where regulation stands as of early July 2026, and a closing list of the stats we refused to print. For the merchant-side decision — whether to add BNPL at checkout at all — see our companion buy-now-pay-later decision matrix.

The strongest single US number comes from the Federal Reserve Bank of Richmond, whose February 2026 economic brief fit a log-linear model to the CFPB’s loan-origination data across the six largest providers. That model puts 2025 US BNPL transaction value at roughly $70 billion — about 1.1% of total US credit-card purchase volume. Growth has settled to around 20% a year in inflation-adjusted terms, a sharp deceleration from the pandemic-era surge (origination volume rose more than 300% from 2019 to 2020, then roughly 186% the following year before cooling).

The regulator’s own loan-level dataset only runs through 2023, its most recent full year. It covers Affirm, Afterpay, Klarna, PayPal, Sezzle and Zip, and shows nominal origination climbing from $2.2 billion in 2019 to $43.9 billion in 2023, with the average loan size rising from $111 to $131. Anything after 2023 — including the $70 billion 2025 figure — is a model estimate, not a CFPB-confirmed count, and we label it that way throughout.

Roughly 47% of Americans say they’ve used a BNPL service at least once, per LendingTree’s March 2026 survey, including 10% who’ve used one six or more times. Usage skews young and, counterintuitively, not only low-income: 61% of Gen Z (ages 18-29) and 59% of people earning $100,000-plus a year report having used it. A separate J.D. Power study found 37% of consumers made a BNPL purchase in the past 90 days — a 5-point jump in a year — rising to 50% among consumers under 40.

Provider usage overlaps heavily, because many people hold accounts with several at once. Among users, PayPal leads, followed by Affirm, Klarna and Cash App Afterpay. These shares are not mutually exclusive, so they sum well past 100% — a reminder that “most-used provider” is a usage-incidence figure, not market share.

The demographic spread matters for merchants. BNPL is no longer a fringe checkout option for the young and cash-strapped; the near-60% adoption among six-figure earners tells you it has become a budgeting-and-convenience tool as much as a credit tool. That is the same dynamic driving abandonment-recovery interest — see the cart-abandonment data BNPL is meant to solve.

Most BNPL coverage cites one risk number and treats it as the risk number. That is the single biggest source of confusion in the category, because there are at least three in wide circulation and they genuinely measure different things. The CFPB measures dollars actually written off. The industry trade group reports its own delinquency definition. LendingTree asks consumers whether they were ever late. Line them up and the “alarming 47%” and the “reassuring under-2%” stop contradicting each other.

Put crudely: the self-reported late-payment rate is more than 20 times the reported delinquency rate — 47% versus under 2% — yet neither is wrong. A payment three days late that gets cured is a “late payment” a consumer will remember and report, but it rarely becomes a delinquency and almost never a charge-off. The CFPB’s independently measured charge-off rate of 1.83% sits in the middle and is the number to anchor on for actual credit loss.

The self-reported late-payment rate has climbed steadily: 34% in 2024, 41% in 2025, 47% in 2026 — a 13-point rise in two years. The trend, not the level, is the story, because the survey definition hasn’t changed. Alongside it, borrowing behavior has intensified. 63% of users have held multiple BNPL loans at the same time, and 25% have juggled three or more at once, up from 23% a year earlier. Gen Z and millennial users are the most likely to stack.

The demand signals underneath are the part worth interpreting. 68% of users agree BNPL causes them to overspend, and 54% say they wouldn’t be able to make ends meet without it — a share that rises to 62% among parents of kids under 18. The fastest-growing use case is telling: 29% used BNPL for groceries in 2026, up from 14% two years earlier — more than double — now the third most-common category behind clothing and tech. When installment credit migrates from discretionary electronics to weekly groceries, it stops being a convenience feature and starts being a cash-flow patch, and that shift is what the rising late-payment line is quietly tracking.

None of this has yet shown up as broader consumer-credit stress. Credit-card delinquency and charge-off rates had stabilized and begun to flatten heading into the second half of 2025, which underpins the Richmond Fed’s conclusion that BNPL has not caused system-level strain to date. The projection worth watching is whether the self-reported late-payment line keeps rising while grocery use expands — if both continue, the gap between “user-felt” strain and “lender-measured” loss is the leading indicator to monitor into 2027, well before it would surface in charge-off data.

The hardest volume numbers come from the checkout data itself. BNPL contributed $20 billion in US online spend during the 2025 holiday season (November 1 to December 31), up 9.8% year over year, per Adobe. Against a record $257.8 billion in total online holiday spend, that works out to roughly 7.8% of holiday e-commerce dollars (our calculation: $20B / $257.8B). Cyber Monday 2025 was the single biggest BNPL day ever, crossing $1 billion for the first time at $1.03 billion, up 4.2% year over year. Mobile did the heavy lifting — smartphones drove 82.2% of holiday BNPL purchases and 79.4% of the Cyber Monday total.

At the provider level, the SEC filings are the firmest ground. Both leaders are growing merchandise volume in the low-to-mid 30s percent, and Klarna in particular has crossed into profitability. The provider numbers below are the hardest available data points on the category’s trajectory. For a deeper cut on peak-season payment mix, see our holiday e-commerce statistics roundup.

The number merchants most want — how much BNPL lifts average order value — is also the least trustworthy, because nearly every published figure is a marketing claim from a provider with an incentive to inflate it. We found the same claimed lift stated as anything from +17% to +210% depending entirely on who was measuring and how. The table below ranks the claims by disclosure quality rather than by size, so the biggest number doesn’t win by default.

Our own published e-commerce data carries the most conservative figure — a +17% AOV lift on a $135 average US BNPL transaction — and that is the number we’d put in a board deck. It is not that the higher claims are impossible; a well-placed, high-consideration product genuinely can see a larger lift. It is that none of the 40%-plus figures come with a controlled, third-party methodology, so treating them as a planning assumption is how merchants end up disappointed. Reconcile against our essential e-commerce statistics before you model a BNPL business case.

The regulatory picture reversed direction over the past year. The CFPB withdrew its 2024 BNPL interpretive rule on May 12, 2025 — the rule that had treated pay-in-four products as credit cards under Regulation Z — and stated it does not intend to reissue it, calling the original interpretation procedurally defective. The Bureau has also said it will not prioritize enforcement actions tied to digital account access to BNPL loans, redirecting supervision toward what it frames as more pressing consumer threats. As of early July 2026, no reversal of that non-enforcement posture has been announced, though the stance can shift and is worth re-checking against the CFPB’s own newsroom before relying on it.

With the federal step-back, states are moving. New York is building the first state BNPL licensing regime: the NY Department of Financial Services published a proposed rule on February 23, 2026 to implement the 2025 NY BNPL Act. On the announced timeline, the rule would take effect 180 days after final adoption, with lenders given a further 45 days to obtain a license. Importantly, as of the July 2026 cutoff this rule is still proposed, not final — BNPL licensing is not yet in force in New York, and coverage that implies otherwise is ahead of the facts.

A stats roundup is only as honest as the numbers it leaves out. These three circulate widely and look authoritative, but none survived our sourcing bar — so rather than print them with a quiet hedge, we’re naming them and explaining why. If you’ve seen these quoted elsewhere, this is the context that’s usually missing.