Payments — Article 6 of 12

BNPL, Installments, and Embedded Lending in Payment Flows

Buy now, pay later has evolved from a niche checkout option to a $309 billion global market processing 2.1 billion transactions annually. This deep dive examines the technical architecture, risk models, regulatory requirements, and operational challenges of embedding lending directly into payment flows.

8 min read
Payments

Klarna processed 2.2 billion transactions worth $95 billion in 2023, while Affirm originated $20.2 billion in gross merchandise value across 18.7 million active consumers. These numbers represent a 312% compound annual growth rate in BNPL transaction volume since 2020, driven by consumer preference for interest-free installments over traditional credit cards. The technical infrastructure supporting this growth requires real-time decisioning engines capable of underwriting credit in under 300 milliseconds while maintaining loss rates below 2.1%.

The integration of lending directly into payment flows creates unique technical and operational challenges. Unlike traditional credit applications that occur separately from purchase transactions, BNPL providers must perform identity verification, fraud screening, affordability assessments, and credit decisioning within the checkout flow without adding friction. This requires sophisticated orchestration between payment processors, credit bureaus, banking partners, and merchant systems.

Technical Architecture and Integration Models

BNPL providers typically offer three integration models for merchants. Direct API integration provides the deepest control but requires 160-200 engineering hours to implement. SDK-based integration reduces implementation to 40-60 hours by abstracting authentication, tokenization, and callback handling. Platform partnerships through payment processors like Stripe, Adyen, or Checkout.com enable activation in under 4 hours but limit customization options.

BNPL Integration Models
Integration TypeImplementation TimeCustomizationTypical Use Case
Direct API160-200 hoursFull control over UX/UIEnterprise retailers >$100M GMV
Provider SDK40-60 hoursModerate flexibilityMid-market merchants $10-100M GMV
Payment Gateway<4 hoursLimited to prebuilt widgetsSMBs <$10M GMV
Platform Plugin<1 hourNoneShopify/WooCommerce stores

The technical flow begins when a consumer selects BNPL at checkout. The merchant's payment orchestration layer routes the request to the BNPL provider's decisioning API, passing basket contents, purchase amount, customer email, and device fingerprint. The provider performs real-time checks against internal fraud models, third-party data sources like Experian or TransUnion, and bank account verification services. This entire process must complete within 300-500 milliseconds to meet checkout conversion requirements.

Affirm's architecture processes these decisions through a microservices platform running on AWS, with separate services for identity verification (using Socure and Jumio), bank account verification (through Plaid and Finicity), fraud scoring (internal models plus Sift and Forter), and credit decisioning. The system handles 14,000 requests per second during peak shopping periods, with 99.95% uptime SLAs.

💡Did You Know?
Klarna's decision engine analyzes over 200 data points per transaction, including device fingerprints, email age, social media presence, and shopping patterns, achieving approval rates of 68% for new customers and 89% for returning users.

Risk Scoring and Decisioning at Point of Sale

BNPL providers face unique underwriting challenges compared to traditional lenders. Without access to comprehensive credit bureau data in many markets, they rely on alternative data sources and machine learning models. Klarna's risk models incorporate purchase history across their network of 500,000 merchants, identifying patterns like frequent returns, suspicious shipping addresses, or velocity violations.

The underwriting models segment customers into risk tiers. Prime customers with FICO scores above 720 receive automatic approval up to $2,500. Near-prime customers (FICO 660-719) face lower limits of $500-1,000 and may require bank account verification. Subprime customers typically receive offers only for purchases under $200 with shortened repayment terms. This segmentation allows providers to maintain portfolio-wide default rates between 1.8% and 2.4%, compared to credit card charge-off rates of 3.4%.

We process 42 million credit decisions daily with an average response time of 287 milliseconds. The key is pre-computing risk scores for returning customers and caching decision rules at the edge.
Head of Risk Engineering, Major BNPL Provider

Fraud detection represents another critical component. BNPL providers lose an estimated $480 million annually to first-party fraud (customers with no intention to repay) and $320 million to account takeovers. Machine learning models analyze behavioral patterns like typing cadence, mouse movements, and browsing patterns to identify suspicious activity. Afterpay's fraud detection system processes 1.2 billion events daily, flagging 0.3% of transactions for manual review.

Settlement Mechanics and Merchant Funding

BNPL settlement differs fundamentally from credit card processing. While card networks settle transactions within 1-3 business days minus interchange fees of 1.5-3.5%, BNPL providers typically fund merchants immediately but charge higher merchant discount rates (MDRs) of 2.9-6.9%. This premium reflects the credit risk assumption and operational costs of collection.

BNPL Merchant Net Settlement
Net = Transaction Amount × (1 - MDR) - Fixed Fee
Example: $100 purchase with 4.5% MDR + $0.30 fee = $95.20 merchant payout

The settlement process involves multiple parties. When Affirm approves a $1,000 purchase, they immediately transfer $954 (assuming 4.6% MDR) to the merchant through ACH or wire transfer. Affirm funds this payment through a combination of warehouse credit facilities ($3.5 billion available), forward flow agreements with banks purchasing the loans, and their own balance sheet. Cross River Bank and Celtic Bank originate the underlying loans in states where lending licenses are required.

Collection mechanics vary by provider and product. Klarna's Pay in 4 product attempts automatic ACH debits every 14 days, with failed payments triggering SMS and email campaigns. After 60 days, delinquent accounts transfer to third-party collection agencies who receive 25-35% of recovered amounts. Longer-term installment products (6-36 months) typically report to credit bureaus and follow traditional consumer lending collection practices.

$14.2BOutstanding BNPL balances held by U.S. consumers as of December 2023

Regulatory Landscape and Compliance Requirements

The regulatory environment for BNPL has evolved rapidly since 2021. The Consumer Financial Protection Bureau (CFPB) issued interpretive guidance requiring BNPL providers to comply with Regulation Z dispute and refund provisions, affecting providers processing over 600 million transactions annually in the United States. The UK Financial Conduct Authority mandates affordability assessments for all BNPL transactions by February 2024, requiring providers to verify income and existing debt obligations.

BNPL Regulatory Evolution
1
2021 - ASIC Guidelines

Australia requires BNPL providers to verify customer income for purchases over AUD 2,000

2
2022 - CFPB Inquiry

U.S. regulator demands data from major BNPL providers on lending practices and consumer risks

3
2023 - EU Consumer Credit Directive

European Parliament votes to include BNPL in consumer credit regulations

4
2024 - UK FCA Rules

Mandatory affordability checks and cooling-off periods for UK BNPL transactions

5
2025 - Singapore MAS Framework

Licensing requirements for BNPL providers with debt-to-income ratio caps

Compliance costs have increased substantially. Klarna spent $47 million on regulatory compliance in 2023, up from $12 million in 2020. This includes investments in affordability assessment APIs that check credit bureau data, open banking feeds, and employer verification services. The technical implementation requires real-time integration with services like Plaid for bank account verification and Equifax for credit checks, adding 200-400 milliseconds to decision times.

Data protection represents another compliance challenge. GDPR in Europe and CCPA in California require BNPL providers to implement robust consent management, data portability, and deletion capabilities. Affirm processes 14,000 data subject requests monthly, with automated systems handling 92% without human intervention. The technical architecture requires event-sourced data stores that can reconstruct and purge customer data across dozens of microservices.

Alternative Installment Models Beyond Classic BNPL

The market has evolved beyond the standard Pay-in-4 model. Splitit offers installment payments using existing credit card limits, eliminating underwriting requirements while earning interchange revenue. Their model processes $470 million in annual GMV by placing holds on customer credit cards for the full purchase amount, then capturing installments monthly. This approach achieves 94% approval rates but limits average transaction sizes to $780.

Emerging BNPL Models
B2B BNPL
Providers like Resolve and Billie offer net terms for business purchases, with average transaction sizes of $15,000 and 30-90 day payment terms
Lease-to-Own
Progressive Leasing and Acima process $7 billion annually in rent-to-own transactions for furniture and electronics, with 12-24 month terms
Embedded Cards
Affirm Card and Klarna Card combine virtual card issuance with installment options, processing $2.3 billion in 2023 across 4.2 million active cards
Crypto BNPL
Platforms like Credix enable stablecoin-based installment lending in emerging markets, with $120 million in originated loans

White-label BNPL platforms enable banks and fintechs to launch their own installment products. Marqeta's BNPL accelerator provides APIs for virtual card issuance, installment scheduling, and payment processing. ChargeAfter's embedded lending platform connects merchants with multiple lenders through a single API, enabling real-time comparison of offers from 20+ providers. These platforms handle the technical complexity of multi-lender integration while merchants maintain a single technical interface.

U.S. BNPL Market Share by Provider (2023)

Implementation Challenges and Operational Considerations

Merchants implementing BNPL face several technical challenges. Cart abandonment increases by 12-18% when BNPL decisioning takes longer than 500 milliseconds. Payment orchestration platforms must implement intelligent routing to failover between multiple BNPL providers when primary options decline transactions. Modern payment hubs maintain real-time performance metrics for each provider, automatically routing transactions based on approval rates, response times, and merchant fees.

⚠️Integration Complexity
BNPL providers average 2.3 breaking API changes annually, requiring merchants to maintain versioning strategies and regression testing. Webhook reliability varies significantly, with some providers achieving only 94% delivery rates for critical events like payment confirmations.

Reconciliation presents ongoing operational challenges. Unlike card payments that settle in predictable batches, BNPL settlements occur throughout the day across multiple rails. Affirm settles via ACH, wire, and virtual card networks depending on transaction size and merchant preferences. Finance teams report spending 40-60 hours monthly reconciling BNPL transactions, compared to 8-12 hours for traditional card payments. Automated reconciliation tools from providers like Modern Treasury and Stripe reduce this overhead by 70%.

Return processing adds another layer of complexity. When customers return BNPL purchases, merchants must coordinate refunds across partially collected installments. The provider must stop future collections, reverse any reported credit bureau data, and handle partial refunds. This process fails in 3-5% of cases, requiring manual intervention. Target reported processing 2.1 million BNPL returns in 2023, with 67,000 requiring escalation to resolve payment discrepancies.

BNPL Implementation Requirements

Customer service implications often surprise merchants new to BNPL. Support tickets increase by 2.3x for BNPL transactions compared to standard payments, driven by questions about payment schedules, failed collection attempts, and return policies. Best Buy expanded their customer service team by 120 representatives specifically for BNPL-related inquiries after implementing Klarna and Affirm options. The complexity requires specialized training on each provider's policies and systems access for payment lookup and modification.

The embedded lending market continues to evolve rapidly. Apple's entry with Apple Pay Later, processing $1.3 billion in loans before shutting down in 2024, demonstrated both the opportunity and challenges in the space. Traditional card issuers have responded with their own installment products — Chase My Plan and Citi Flex Pay now cover $34 billion in converted purchases annually. As real-time payment rails mature and open banking adoption increases, the technical barriers to launching BNPL products continue to decrease, suggesting further market fragmentation ahead.

Frequently Asked Questions

What are typical BNPL merchant discount rates and how do they compare to credit cards?

BNPL merchant discount rates range from 2.9% to 6.9%, compared to 1.5% to 3.5% for credit cards. The premium reflects credit risk assumption and immediate merchant funding. Rates vary by industry, with electronics averaging 3.2% and fashion averaging 5.4%.

How do BNPL providers make money with 0% interest offerings?

BNPL providers generate revenue primarily through merchant fees (70-80% of revenue), late payment fees (10-15%), and merchant marketing services (5-10%). Affirm earned $1.67 billion in 2023, with $1.2 billion from merchant network revenue and $312 million from late fees.

What happens when a BNPL payment fails or a customer defaults?

Failed payments trigger retry attempts over 3-7 days, followed by email/SMS campaigns. After 30-60 days, accounts transfer to collections. Default rates average 1.8-2.4%, with providers typically selling charged-off debt for 5-8 cents on the dollar to debt buyers.

Can merchants offer multiple BNPL options at checkout?

Yes, 42% of enterprise merchants offer 2-4 BNPL options. Payment orchestration platforms like Primer and Checkout.com enable dynamic BNPL selection based on cart value, customer geography, and real-time approval rates, increasing overall conversion by 8-12%.

How do BNPL providers verify customer identity and prevent fraud?

Providers use multi-layered verification including device fingerprinting, email/phone verification, social media checks, and bureau data. Real-time fraud models analyze 200+ signals per transaction. Identity verification vendors like Jumio and Socure provide sub-second verification with 97% accuracy rates.