Areeba is an innovator in payments. It was using Hadoop for unstructured data and MariaDB as a relational database. Now, Areeba has added SingleStore to the mix for several projects, including fraud detection, anti-money laundering (AML), and product recommendations.
Areeba is a financial services company that was spun out of Bank Audi, a leading Middle Eastern bank, last year. Previously, Areeba was the payment cards and electronic services division for Bank Audi. Now, Areeba has more than half a dozen banks as clients, running credit card loyalty programs and acting as a payment gateway for e-commerce in countries across the Middle East and Africa.
For the first project, SingleStore and Apache Spark work together to detect potentially fraudulent transactions and implement anti-money laundering (AML) regulations. This is a critical function for Areeba. “SingleStore is helping Areeba cover all our operational analytical needs,” said Elie Soukayem, director of data and analytics at Areeba.
For the second project, SingleStore will store the insights generated by customer activity. Then, Areeba will use predictive analytics to recommend special offers, on a real-time basis, for each customer.
“I was first exposed to SingleStore when we were still part of Bank Audi,” said Soukayem. “A partner company used SingleStore to build a prototype for card processing, including suggesting offers. When I left Bank Audi and came to Areeba, to save time, we went straight to SingleStore.”
Areeba Moves Faster and Faster
“Now” is always the best time to detect – or prevent – fraud, money laundering, and other illicit transactions. Currently, fraud detection and AML checks are run against each day’s transactions, so response to fraud and money laundering attempts is reactive. Fraud and AML reporting needs to run as fast as possible so the response can begin immediately.
In the near future, it’s expected that fraud and AML detection will be enabled for transactions as they occur. That way, an illicit transaction can either be prevented or responded to very quickly after it occurs. Illicit activity can then be prevented for subsequent transactions on the same card, or from the same customer, shortly after the initial suspect use. When that change happens, real-time responsiveness will become critical.
For the recommendation engine, SingleStore will need to combine the latest activity for each customer with the latest special offers. Then, each time a customer makes a purchase, they can be presented with an offer most relevant to their updated purchasing history, in real time. Only with real-time processing can the best offer be determined.
In both cases, with stale data, the risks of a negative outcome for the vendor and a negative experience for the customer are high. Areeba has to move beyond traditional solutions as it works to create the best outcomes for its customers. That’s where SingleStore comes in.
“Translytical” to the Rescue
SingleStore is part of Areeba’s solution on both challenges – detecting illicit transactions and real-time recommendations. Because SingleStore combines transactional and analytical processing, each task gets full analytics power on the latest, continually updated data; not old data that’s been put through batch processes.
On the fraud and AML side, Areeba’s Soukayem said, “We’re using SingleStore with Apache Spark to dig for suspicious activities. Performance with SingleStore is impressive.”
“Our biggest table is around 100GB in size and holds hundreds of millions of authorizations, covering a period of three years,” continued Soukayem. “We have another similar table, and a smaller table for transactions, that holds the settlement of pending authorizations.”
All of this data has to be processed in real time for effective results. It will also be used to extract hints for the merchant to use to make recommendations for add-on purchases.
“I’ve been in this business since 2005, and performance has always been a challenge,” Soukayem said. “Storage in memory and on disk is a plus.” Which is great, because those are the default options for SingleStore’s rowstore and columnstore.
The Future is Faster
Areeba is solving problems today and busily preparing for the future. “We might leverage blockchain to make payments easier in the region,” said Soukayem. In addition, Areeba is increasing its usage of the cloud. Soon, it will establish an innovation garage to work on even more advanced technologies.
Areeba’s founding principles are to upgrade technology to higher levels; innovate on both products and services; and establish reliable, long-term relationships with customers. Areeba is looking to take SingleStore along as they move to the cutting edge.
For more information on SingleStoreDB Self-Managed 6.7, see the press release and all eight blog posts on the new version: Product launch; managing and monitoring SingleStoreDB Self-Managed 6.7 (includes SingleStore Studio); the new free tier; performance improvements; the Areeba case study (this post); the Wag! case study; the creation of the Visual Explain feature; and how we developed adaptive compression.
