All articles
Card-Linked OffersFintech ArchitecturePaymentsBanking Loyalty

Behind the Swipe: Scheme Tokenisation & CLO Tech Infrastructure

The Reward Collection

A deep-dive architectural guide into card-linked offers: scheme tokenisation, webhook pipelines, transaction reconciliation, and sub-second attribution.

The Architecture of Frictionless Loyalty

To the modern banking consumer, the experience of a card-linked offer (CLO) appears deceptively simple. A cardholder opens their banking application, sees an offer for 10% cashback at an apparel retailer, taps "Activate", and pays in-store or online using their existing debit card or mobile wallet. Moments later, a push notification confirms their reward.

Beneath this seamless interface lies one of the most sophisticated, high-throughput distributed pipelines in contemporary financial technology. Delivering instant gratification without requiring coupon codes, physical loyalty cards, or point-of-sale (POS) modifications demands synchronisation across global card networks, banking core databases, real-time message queues, and automated reconciliation ledgers.

For enterprise technical leads, product managers, and fintech architects, understanding the plumbing behind card-linked offers is essential when designing scalable, resilient banking loyalty ecosystems. Here is an architectural breakdown of what actually occurs across the payment rails between the physical swipe and the cashback deposit.

1. Card Enrolment and Network Scheme Tokenisation

The fundamental challenge of any enterprise CLO network is privacy and compliance: how can an external loyalty engine track spend across thousands of merchants without storing, handling, or exposing sensitive 16-digit Primary Account Numbers (PANs)?

The answer lies in network scheme tokenisation.

When a bank integrates with an enterprise CLO provider like The Reward Collection, cardholder cards are enrolled directly via card network APIs (such as Visa Offers Platform or Mastercard Transaction Services). Instead of transmitting unencrypted PANs, the system leverages payment scheme tokens:

  • Payment Account Reference (PAR): A non-financial unique identifier generated by the card scheme that links disparate PANs, device tokens (such as Apple Pay or Google Wallet DPANs), and replacement plastic cards to a single underlying account.
  • CLO Scheme Token: An opaque surrogate string provisioned by the card scheme specifically for the loyalty network.

By leveraging PAR and token-based identification, the entire loyalty ecosystem achieves robust PCI-DSS de-scoping. Sensitive card data never touches the loyalty network's application servers, while ensuring that whether the customer pays via physical plastic, an Apple Watch, or an online virtual card, the spend attributes back to the single registered identity.

2. Dual-Messaging Rails: Auth vs. Clearing Webhooks

Card transactions do not settle in a single step. The infrastructure behind high-performance CLO engines must elegantly manage the distinction between the Authorisation (Auth) stream and the Clearing & Settlement stream.

The Real-Time Auth Stream (The Notification Layer)

When a consumer taps their card at a merchant terminal, the acquirer submits an ISO 8583/20022 authorisation request through the payment scheme to the card issuer.

During this sub-second roundtrip, the card scheme copies the transaction metadata—Merchant Category Code (MCC), Acquirer Reference Number (ARN), merchant identifier, card token, timestamp, and spend amount—and emits an asynchronous event webhook to the CLO ingestion engine.

This pipeline requires low-latency event-driven architecture, typically built on distributed message brokers (such as Apache Kafka or AWS Kinesis). Within 200 to 500 milliseconds, the ingestion service evaluates the transaction against activated campaign parameters. If eligible, it dispatches an event to the bank's push notification service, delighting the user with an instant confirmation: "You've earned £8.50 cashback at Merchant X!"

The Clearing Stream (The Financial Ledger)

However, authorisation is not final settlement. Authorisations can expire, tip amounts can be added later, or the customer might cancel the purchase before leaving the store.

The actual financial commitment relies on the secondary Clearing Record (typically delivered via batch processing 24 to 48 hours later). The CLO engine must maintain stateful idempotency, matching the incoming clearing file against the original pending authorisation. Only when the clearing record confirms successful debiting does the offer transition from "Pending" to "Settled" in the loyalty database.

3. The Rules and Attribution Engine

Filtering millions of daily transactions across a 500+ merchant network requires an ultra-efficient deterministic matching engine. When transaction payloads arrive, the system runs a sequential validation pipeline:

Incoming Webhook ──> Token Identity Match ──> Merchant Location Verification 
                                                        │
Cashback Dispatched <── Minimum Spend Met <── Exclusions & Caps Checked
  1. Merchant Entity Matching: Validating spend at specific outlets requires matching raw transaction strings (which are notoriously messy, e.g., MCHNT*UK STORE #402 LONDON) against normalized merchant IDs, MID tables, and terminal geofences.
  2. Spend Threshold Logic: The engine evaluates whether the net qualifying basket size meets minimum requirements (e.g., £50 minimum spend to trigger 10% cashback).
  3. Frequency and Velocity Rules: Offers frequently contain cadence constraints, such as "First purchase only", "Once per 30-day cycle", or "Maximum £25 cashback per redemption".
  4. Exclusions and SKU Caveats: Advanced implementations incorporate Open Banking transaction categorization or direct electronic receipt integration to filter out ineligible purchases, such as gift cards or prescription items.

4. Automated Settlement and Multi-Party Reconciliation

The final architectural pillar is financial reconciliation. Card-linked offers typically operate on a merchant-funded rewards model, meaning the merchant funds the cashback and network processing fees, while the bank acts as the disbursement channel.

To manage this seamlessly, the CLO infrastructure operates an automated ledgering engine:

  • Merchant Invoicing & Escrow: Periodically, the network aggregates verified, non-reversible settlements and generates programmatic invoicing for participating brands.
  • Automated Payouts: Concurrently, batch credit instructions are routed to the participating bank's payment gateway or Open Banking Payment Initiation rails (PISP) to credit cardholder accounts via direct ledger transfers or statement rebates.
  • Chargeback and Return Amortisation: The engine maintains an audit window (usually 14 to 30 days) to handle refunds. If a cardholder returns an item, the clearing file registers a reversal, and the settlement engine automatically recovers or decrements the pending loyalty balance prior to disbursement.

Enterprise Integration Without the Overhead

Building an in-house CLO infrastructure requires years of direct scheme certification, high-throughput message streaming architecture, complex merchant entity resolution, and continuous PCI-DSS auditing. For retail banks, neobanks, and reward platforms, building this stack from scratch represents a massive diversion from core engineering priorities.

By partnering with established infrastructure providers like The Reward Collection, institutions bypass technical complexity. With turnkey API integration, comprehensive card scheme connectivity, and access to over 500+ top-tier merchant brands, banks can launch a feature-complete loyalty engine in a fraction of the time.

Explore our technical documentation and FAQs or speak with our solutions engineering team to discover how easily our scalable card-linked architecture integrates into your existing digital banking stack.

Ready to power your loyalty programme?

Connect with the UK's leading Card-Linked Offers and Open Banking network.

Get Connected →