pay.2nth.ai Tree modes bnpl
modes · BNPL · Leaf

Free for the shopper. Someone still pays.

BNPL felt like magic: interest-free, instant, no card needed. Underneath it is short-term lending with merchant-funded economics and thin affordability checks — and in 2026 the free pass from regulators is ending.

Deferred credit Merchant-funded Split-pay Affordability FCA / CCD2

A point-of-sale loan dressed as a payment method

BNPL is consumer credit with a UX team. Treat it as lending, not as a wallet.

Buy now, pay later lets a shopper take the goods today and pay in instalments — classically four payments over six weeks, interest-free to the consumer. The merchant gets paid in full, up front, minus a fee. The BNPL provider carries the credit risk and the float.

Strip away the branding and BNPL is unsecured short-term consumer credit originated at the checkout. The genius was never the lending — it was the distribution: a one-tap decision embedded in the buy button, with no card, no application form, and (until recently) no visible regulator. That frictionlessness is exactly what is now under scrutiny.

Big players: Klarna, Afterpay (Block), Affirm, PayPal Pay in 4, Zip. In South Africa: PayJustNow, Payflex (Zip-owned), Float, Happy Pay. The models differ but the spine is the same — defer the consumer’s pain, charge the merchant for conversion.

The money flow and the two product shapes

Two products wear one brand. Split-pay is a conversion tool; instalment is real lending. Conflating them is where merchants and regulators both get tripped.

The unit economics are merchant-funded. The merchant pays a merchant discount rate (MDR) — typically 2–8%, well above card interchange — in exchange for higher conversion and bigger baskets. The consumer pays nothing if they pay on time; the provider earns on the MDR, on late fees, and (for longer products) on interest.

ModelShapeConsumer costWho it suits
Split-pay (Pay-in-N)3–4 instalments, ~6 weeks, 0% interestFree if on time; late fees if notFashion, low-ticket retail, impulse baskets
Instalment6–36 months, interest-bearingAPR applies, often 0% promo then standardElectronics, furniture, higher-ticket
Pay-in-30Single deferred paymentFree if on timeTry-before-you-buy returns flows
Merchant pays, not the shopper

The MDR is the engine. BNPL sells conversion uplift and average-order-value growth; the merchant funds the interest-free window.

The float is the balance sheet

Providers front the full purchase to the merchant and recover from the consumer over weeks. That funding cost rose sharply as rates climbed 2022–2024.

Late fees and interest are the upside

Split-pay margins are thin; profitability leans on longer interest-bearing products, late fees, and cross-sell. Pure pay-in-4 has struggled to make money.

Soft-check origination

Approval is near-instant and often uses a soft credit check or none at all — the speed that drove adoption is the same gap regulators are closing.

Why retailers pay 2–8% to offer it

The merchant trade is blunt: pay a higher fee than a card costs, get more sales. Providers cite double-digit lifts in conversion and basket size. For high-margin or high-AOV categories that math works; for thin-margin grocery or commodity retail it rarely does.

The provider’s P&L is harder than it looks. Revenue is MDR + late fees + interest. Costs are funding (the float), credit losses, fraud, and acquisition. When rates rose and loss rates normalised after the 2021 boom, several pure-play BNPLs swung from growth-at-all-costs to a brutal focus on credit quality — Klarna’s path to profitability and its 2025 US listing being the headline example.

For PSPs and acquirers, BNPL is increasingly a checkout option to be orchestrated, not a business to build from scratch. The strategic question is whether you integrate third-party BNPL, white-label it, or stay out — weighed against the regulatory burden now attaching to the lender of record.

The part the UX hides

BNPL’s growth came from removing friction at exactly the moment a credit decision should add some. That tension is the whole regulatory story.

The core risk is the one BNPL was designed to obscure: can the customer actually afford this? Frictionless approval, stacking across multiple providers (a shopper can run four BNPL plans at four providers simultaneously, invisible to each), and thin or absent affordability checks combine into real consumer-harm exposure — especially for younger and lower-income users.

Loan stacking is invisible

Without shared bureau reporting, no single provider sees the consumer’s total BNPL exposure. A shopper can be current with each lender and drowning overall. Bureau reporting is now being mandated in several markets.

Late fees hit the vulnerable hardest

The customers most likely to miss payments are the ones least able to absorb fees. Regulators read this as a fairness and treatment-of-vulnerable-customers problem, not just a credit one.

“Not credit” was a structuring choice, not a fact

Many providers structured products to fall outside consumer-credit law (e.g. exemptions for short, interest-free, low-instalment-count agreements). That arbitrage is precisely what the UK and EU are now closing.

Returns and disputes are messy

When goods are returned, the BNPL plan must unwind correctly. Reconciliation between merchant refunds and instalment schedules is a common operational failure that lands on the consumer’s credit file.

Funding-cost sensitivity

BNPL margins are rate-sensitive. A model that worked at near-zero rates can be loss-making at higher ones, pushing providers toward fees and longer interest-bearing products.

UK, EU and the South African gap

The free pass is closing — unevenly by region.

JurisdictionInstrumentStatus (May 2026)What changes
UKFCA regime for Deferred Payment Credit (DPC); PS26/1Final rules published Feb 2026; live 15 July 2026BNPL brought into the FSMA perimeter. Mandatory creditworthiness checks (incl. under £50), CONC arrears handling, FOS access, s.75-style protections.
EUConsumer Credit Directive 2 (CCD2)Transposition deadline passed (20 Nov 2025); rules apply from 20 Nov 2026BNPL and interest-free instalments explicitly in scope, including loans below €200. Stricter affordability assessment, pre-contractual disclosure, ban on bundled sales.
South AfricaNational Credit Act (NCA) — no BNPL-specific rule yetUnregulated in practice; classification unresolvedMost SA providers structure as “deferred billing” to sit outside the NCA, avoiding NCR registration and affordability checks. The Conduct of Financial Institutions (COFI) Bill is the expected vehicle to close the gap.

The direction of travel is one-way: BNPL is being reclassified as regulated credit. In the UK, the FCA’s Temporary Permissions Regime registration opens 15 May 2026 for firms already active before 15 July 2025. In the EU, CCD2 ends the small-loan and interest-free carve-outs that BNPL relied on.

South Africa is the laggard and the live debate. The NCA would apply if BNPL were classed as credit — triggering NCR registration, affordability assessment under s.81, and disclosure duties. Providers have so far avoided this by branding the product as a payment or deferred-billing arrangement. Enforcement has been minimal, but the arbitrage is fragile, and COFI is designed to remove it.

Offer it, build it, or skip it

Offer third-party BNPL (most merchants / PSPs)

If you want the conversion uplift without the lending burden, integrate an established provider and let them be lender of record. You pay the MDR; they carry credit, funding and — critically — the new regulatory obligations. Best for retailers and PSPs orchestrating checkout.

White-label or build (banks / large lenders)

Worth it only if BNPL is strategic to your credit franchise and you can fund the float cheaply, price risk, and absorb the compliance build. Post-2026 you are a regulated lender in the UK/EU — treat it as a lending line, not a marketing feature.

Cost of getting it wrong

Pre-2026 the risk was reputational and credit-loss. From mid-to-late 2026 it is regulatory: in the UK, conducting DPC without permission or skipping creditworthiness checks is a perimeter breach with FOS exposure; in the EU, CCD2 non-compliance is enforceable. In SA, the present arbitrage may not survive COFI — do not build a model that only works if you stay unregulated.

SA-specific read

For a South African merchant, BNPL is still cheap to offer and lightly regulated — but treat the regulatory gap as temporary. Favour providers with real affordability logic and bureau reporting; they will adapt to COFI with the least disruption to your checkout.

Where BNPL meets the rest of the stack

Primary sources