AppBuilder
AI-Powered Fraud Detection for Secure E-Commerce Transactions
A leading global commerce platform integrated real-time fraud scoring into its order flow using Adobe App Builder and Kount. The result: faster, data-driven order reviews, reduced chargebacks, and zero impact on customer experience.
The Challenge
Merchants faced rising e-commerce fraud risks with no native automated risk assessment in place. Fraudulent orders could slip through undetected, leading to chargebacks, lost revenue, and operational inefficiencies.
The Scenario
The Solution
Embedding Real-Time Risk Scoring Without Disruption
We developed a serverless microservice on Adobe App Builder to integrate Kount’s AI-driven fraud detection into the Adobe Commerce order pipeline. The solution listens for order placement events, securely sends relevant data to Kount’s API, and retrieves risk scores asynchronously via webhook.
Risk scores are mapped to the relevant order in near real time and displayed directly in the Adobe Commerce admin interface. Admins can see scores at a glance, refresh them on demand, and flag high-risk orders for manual review. All of this was achieved without altering the commerce platform’s core, ensuring platform stability and seamless upgrades.
Key capabilities delivered:
- Event-driven fraud scoring triggered for 100% of orders.
- Asynchronous architecture with zero impact on checkout performance.
- Admin UI integration for risk score display and manual refresh.
- Secure credential storage and GDPR-compliant data handling.
- On-time delivery for a high-visibility industry event.
How we did it
Agile Execution on a Tight Timeline
The integration followed an accelerated 8-week timeline. We began with Adobe I/O event subscription setup, then implemented secure API connectivity to Kount’s risk scoring service. Webhook listeners were built to capture asynchronous results and persist scores in Adobe’s secure state storage.
We extended Adobe Commerce’s admin UI with a lightweight module to fetch and display scores on demand. Rigorous testing simulated both high-volume order loads and diverse fraud scenarios, ensuring scalability and accuracy. The final deployment was executed without downtime, leveraging Adobe’s safe deployment mechanisms.
The Results
The project delivered a fully functional, real-time fraud prevention capability that has already improved operational efficiency. Merchants now instantly identify high-risk orders, focusing resources only where needed.
Checkout speed remains unaffected, and early adoption feedback highlights the ease of use and increased confidence in approving legitimate transactions.
About the Client
A leading global e-commerce platform provider serving enterprise merchants worldwide. The platform supports complex, high-volume retail operations with extensible, API-driven architecture and a rich ecosystem of integrations.
Protect Every Transaction in Real Time
Talk to us about integrating AI-driven fraud scoring into your commerce platform.