BNPL became regulated credit on 15 July in the UK
2026-08-06
Buy now pay later stopped being an unregulated product in the UK. Deferred payment credit is now inside the FCA perimeter, and the Consumer Duty applies to it. Here is why it is a trading story.
- Ecommerce
- Payments
- Checkout
- Ecommerc Strategy
Buy now pay later was, for ten years, the most commercially successful form of unregulated consumer credit in Britain's retail sector. This situation has now come to an end. Deferred payment credit is now within the FCA's jurisdiction, and the majority of the industry coverage has treated it as a matter of compliance: checking your promotions, making sure your lender is authorised, and then moving on.
That framing misses the interesting part. The rules do not merely sit alongside the commercial mechanism that made BNPL work. Two of them act directly on it.
What actually changed
The FCA published its final rules in PS26/1 on 11 February 2026, with 15 July 2026 as the date they took effect. Deferred payment credit, the interest-free instalment model that sits under most BNPL propositions, became a regulated activity. Lenders had to be authorised or registered under the Temporary Permissions Regime, and the Consumer Duty now applies to them.
For consumers the practical changes are straightforward: there will be pre-contract information on the amounts to be paid, the dates of payment and what will happen if a payment is missed, proportionate affordability checks before any credit is given, access to support and guidance towards free debt advice for people who are having difficulties, access to the Financial Ombudsman Service, and Section 75 protection in the case of new agreements, and backdated interest is prohibited.
For retailers, the situation is more complicated than the headlines indicate. When a retailer refers customers to a third-party lender, they are usually not carrying out regulated credit broking, though there are some exceptions, such as in the case of the sale of homes. Retailer-backed instalment plans that charge no interest or fees are still outside the current regime, at least until a review takes place. However, the entire customer journey and all financial promotions relating to BNPL, whether on the website, in the app, by email, through the CRM system or on social media, fall within the lender's compliance responsibilities, and the retailer is responsible for those channels. In reality, the retailer handles the day-to-day operational tasks while the lender assumes the liability, which is an unsatisfactory arrangement to manage without a clear agreement existing.
The FCA's decision to take action is due to the scale of the issue; the Financial Lives survey showed that one in five UK adults, about 10.9 million people, had used deferred payment credit in the 12 months up to May 2024, compared with 17% and 8.8 million in 2022.
Why the instalment lifts the basket
If you are going to reason about what regulation does, you need to be precise about what BNPL was doing before it, and the best evidence here is not from a payments vendor.
Stijn Maesen and Dionysius Ang published a study in the Journal of Marketing using transactional data from a major US retailer that partnered with a BNPL provider for the first time. Their difference-in-differences analysis found that adoption of instalment payments was associated with an increase in purchase incidence of roughly 9 percentage points and a relative increase in purchase amounts of roughly 10%, and the effects held across the whole post-adoption period rather than fading after novelty wore off.
The mechanism they propose, then test in three preregistered experiments, is that segregating a payment into instalments alleviates perceived financial constraint. This is a more precise claim than the usual one about impulsivity. Customers paying in instalments feel less constrained than customers paying an identical amount as a lump sum, whether that lump sum is now or later. The budget feels roomier, and a large body of prior work links lower perceived financial constraint to higher spending.
The heterogeneity in their results is the part worth taking into a trading meeting. The effect was stronger among shoppers who had historically used credit cards, and stronger among shoppers who had historically bought smaller baskets. In other words, the lift concentrates in the segment where you previously had the least revenue at stake, which is also the segment most likely to be marginal on contribution once you have picked, packed, delivered and processed a return.
There is a display effect layered on top of the budgeting one. Rhys Ashby, Shahin Sharifi, Jun Yao and Lawrence Ang published work in the Journal of Retailing in 2025 showing that presenting an instalment amount rather than the full price lowers perceived cost, and attributing it to numerosity: we take magnitude from the raw number in front of us rather than from the total obligation. Four payments of £30 is read as cheaper than £120. The two are the same money.
What the new rules do to that mechanism
Set the two mechanisms next to the two headline requirements, and the interaction is obvious. Segregation works because the total is not the salient number. Clear pre-contract information about amounts, dates and missed payment consequences puts the total, and the schedule of obligations, back in front of the customer at the point of decision. It re-aggregates precisely what the instalment display had separated.
Proportionate affordability checks have a different effect but are just as direct. They introduce an assessment into the checkout process which earlier did not have one, and as a result of this assessment two outcomes occur instead of just one. Some customers will be turned down during the process. This represents a much worse experience than simply never having been given the option, since the customer had already formed an expectation and, in most cases, had already mentally accepted the items in their basket. For the customer, being withdrawn from after having felt an expectation is experienced as a loss rather than as a mere absence, and the session does not end in a positive state.
One cannot conclude that adoption will decline. According to research by Clearpay, 48% of UK adults stated that they would be more likely to use BNPL once it is regulated, and there is a reasonable trust effect working in the opposite direction, especially among the older, more cautious shoppers who have so far avoided it. The realistic view is that the overall outcome is now truly uncertain in a way that it wasn't in June, and this varies from business to business. Anyone giving you a confident figure based on a vendor case study from before July is giving you a figure relating to a different product.
The margin question
Average order value is the wrong metric for this decision, and it has always been the wrong metric. It is simply the one BNPL providers report most enthusiastically.
Start with acceptance cost. Stripe puts Afterpay and Clearpay merchant fees in the region of 3% to 6% per transaction, against card acceptance that is typically a fraction of that. If instalment payments lift order value by around 10% but shift a meaningful share of your existing card volume onto a rail costing three or four points more, the arithmetic gets tight quickly, and in some categories it inverts. You can pay several points of margin for the privilege of changing how an order you were already going to receive is settled.
Then look downstream, because the customer relationship outlives the transaction. Research by Di Maggio, Williams and Katz, using transaction-level data matched to BNPL integration dates across a large set of US retailers, found that BNPL use increased the likelihood of incurring overdraft fees by around 20% and low balance fees by around 17% in relative terms. None of that is the retailer's liability. All of it is the retailer's customer. A customer under financial strain is not a repeat customer, and lifetime value is where the money actually is.
How I would measure it now
The first thing to fix is the counterfactual. Attributed uplift, meaning the average order value of orders that used BNPL compared with orders that did not, tells you almost nothing, because customers self-select into the payment method. Higher spenders choose instalments; instalments do not create higher spenders on their own. A geo split or a holdout at the point where the option is displayed gives you an incrementality read. It is more work, and it produces a smaller, truer number.
The next step is to change the reporting unit from order value to contribution per order, after having deducted acceptance fees, delivery costs, and the cost of processing returns. Returns should be given their own separate line instead of being included in a footnote. Since separating the payment reduces the sense of constraint, it also lowers the psychological barrier to ordering three sizes to test them out, and it is you who bear the cost of this action rather than the lender. I would like the return rate for BNPL orders to be reported separately from the first month and not mixed into the overall average.
Since July, a new point has been introduced. You should monitor the decline rate as a result of affordability checks and, more importantly, what happens to customers during those sessions—do they move on to using a card or do they simply leave? This one ratio shows whether the check is successfully excluding customers to whom you should not have been giving credit, which is acceptable, or instead causing checkouts to fail when they would have otherwise been completed on a different path, which is a design problem that can be solved. If the fallback is managed well, the customer is given an alternative within the same step rather than being brought back to an error state.
None of this is an argument against offering BNPL. In several categories, at several price points, it does exactly what it says. It is an argument for holding it to the same standard as any other trading decision, which means measuring the incremental contribution rather than the headline lift. That was always the right test. From 15 July, the mechanism producing the lift is materially different from the one those old case studies were measuring, so the answer you arrived at two years ago is not the answer any more.