The Path to Trustworthy Ecommerce: Fraud Prevention and Customer Satisfaction

For many ecommerce retailers, the difference between profitability and going out of business is fraud prevention. And although you might need to add several layers of security to prevent fraud, you don’t want a cumbersome customer experience. That balancing act often leads ecommerce retailers to sacrifice fraud detection to keep customers happy and maximize revenues, but this short-term tradeoff isn’t sustainable.

I sat down with Kevin Cole, Operations Director of Omnyex, a global wholesaler and distributor of video games and related products, to discuss their decision to use Amazon Fraud Detector to help secure their ecommerce platform while maintaining an exceptional customer experience.

AWS: Who is Omnyex?

Kevin Cole: Omnyex is a wholesaler of digital products headquartered in Dubai, and we operate multiple ecommerce websites. One site in particular that is popular with gamers is CDKeys.com, where we resell game keys and top-ups to popular gaming titles across multiple platforms. We deliver a trustworthy, reliable, and fast purchasing experience that delivers the game keys to customers as quickly as possible so that they can make purchases with confidence and spend more time playing.

AWS: What is your fraud challenge?

Kevin Cole: Since 2013, we have witnessed an increase in attempted fraud during checkout. Sometimes stolen payment instruments, like credit cards, are used by fraudsters to make purchases on our site. Once we fulfill an order and send digital codes for the keys, we can’t recover the key data, so accurate fraud detection at checkout is essential to prevent losses.

The original fraud management system we developed had a lot of issues and manual processes. As an example, to limit our fraud, we sent about 10% of orders to a queue for manual review prior to approval. On any given day, one of our websites might process tens of thousands of transactions, with surges during the holidays or with new game releases. It’s difficult to plan for additional staffing during these surge events so we have enough people to manually review the potential fraud purchases. If we were understaffed, purchase approvals could take a long time, which negatively impacted the customer experience—and our customers could turn to our competitors. If we were overstaffed and had idle employees, that labor cost went down the drain.

We started looking for ways to automate our fraud detection and prevention processes at every stage, as much as possible. We wanted to augment our approach with machine learning, and we looked for third-party solutions that could seamlessly fit into our system. Some solutions looked promising, but they didn’t get the job done well enough on their own. Other applications weren’t a fit either because they couldn’t scale, or they were too expensive.

AWS: Why did Omnyex choose Amazon Fraud Detector?

Kevin Cole: When we learned that Amazon offered a fraud detection service that leveraged its own fraud prevention experience across its ecommerce businesses, we were definitely intrigued. We liked that Amazon Fraud Detector is a managed AI service because we wouldn’t have to maintain the software. Because it is an AWS service, we also knew it would automatically scale to keep pace with the transaction spikes in our business, and it would seamlessly integrate with the existing AWS services across our business. Additionally, the solution had a cost model that made sense to us.

In a matter of hours, we were able to train a machine learning model based on our historical transaction information and deploy the solution as a Detector API. Pretty quickly, we integrated the model into our existing fraud detection system by following the code samples showing how to call the API and pass in checkouts for fraud evaluation.

AWS: What results have you seen so far from using Amazon Fraud Detector?

Kevin Cole: We sent the transaction details for a portion of our checkouts to the trained AI fraud detection model, and we were really pleased with the results. The model scores were helpful in a couple of ways:

• We could accurately predict which checkouts would be fraudulent.
• We could trust the automated decisions about which checkouts we could approve as non-fraudulent and which transactions we could block as fraudulent—without doing manual reviews.

With Amazon Fraud Detector, we reduced fraudulent transactions by 6%. At the same time, we’ve been able to automate checkout fulfillment on more than 90% of the transactions that would have previously been flagged for manual review. Now, we’re manually reviewing less than 1% of our transactions—down from 10%—and we are looking at ways to automate the process even more.

Since we implemented this service, we’ve seen a significant improvement in our Trustpilot score, and we know it’s a result of this checkout detection automation, as well as additional enhancements we are consistently making on the website. Trust is a big part of our value to customers, so that’s a huge win for our business.

AWS: Do you have future plans to expand your use of Amazon Fraud Detector?

Kevin Cole: As our confidence in the fraud detection scores have increased, we plan to automate more and more of our checkouts to reduce manual reviews. We also want to expand Amazon Fraud Detector to some of our other ecommerce sites around the globe, should the need arise. We’re excited about the positive impact this solution can have across our enterprise, and we very much look forward to continuing our strong partnership with AWS.

AWS: Thanks for chatting with us, Kevin. We appreciate your insights and expertise.

If you have questions or other feedback for Kevin, Omnyex, or AWS, please leave it in the Comments section. For more information about Amazon Fraud Detector, see our Catching Fraud Faster by Building a Proof of Concept in Amazon Fraud Detector blog post.