ShopBack Cuts Infrastructure Costs by 60% with Amazon EC2 Spot Instances

ShopBack, the leading rewards and discovery platform in Asia Pacific, enables shoppers across the region to shop “The Smarter Way.” It provides users a one-stop rewards platform to earn cashback, while delivering performance-based marketing to merchants. ShopBack has experienced remarkable growth since its launch in Singapore in 2014, having expanded its reach to Malaysia, Indonesia, the Philippines, Thailand, Taiwan, Australia, Vietnam, and Korea. ShopBack partners with leading e-commerce merchants to reward its users with cashback across a wide range of categories including general merchandise, travel bookings, fashion, health and beauty, groceries, and food delivery.

The company relied on Amazon Web Services (AWS) from day one to build a flexible and highly secure infrastructure. But as ShopBack expanded, it ran into some issues with scalability. To meet growing demand, ShopBack needed to evolve its IT architecture from single instances of Amazon Elastic Compute Cloud (Amazon EC2) and Amazon Relational Database Service (Amazon RDS). “Each time we moved into a new market we had to deploy a new stack more or less manually,” says David Chua, engineering manager at ShopBack.

The challenge was magnified each time a large retail sale took place across Asian markets, which often resulted in a tenfold increase in ShopBack’s web traffic. “We had a lot of problems trying to scale up,” says Chua. “If you aren’t sure you can scale to meet a big bump in traffic, you either end up throwing money at the problem or accept that your data will process slowly. Neither is sustainable or conducive to business growth.” Large spikes in traffic could bring the entire ShopBack site down, resulting in disappointed customers and merchants.

In 2018, the company adopted Kubernetes and containerization using Amazon Elastic Kubernetes Service (Amazon EKS), which helped ShopBack scale more effectively and improve resilience, but the startup still required greater cost-efficiency. “As we gained traction, expenses followed the growth curve quite closely. We needed to bring costs back down,” says Chua.

Looking for further cost reductions as well as improved scalability and reliability, ShopBack discovered Amazon EC2 Spot Instances in 2019. Using Spot Instances for various stateless applications allowed the company to increase scalability and reduce its infrastructure costs by 60 percent. Being able to scale out at drastically reduced costs enabled ShopBack to process more customer transactions concurrently within budget, helping customers get their cashback faster. This translated to happier customers and busier merchants, improving the company’s profitability. ShopBack can now process customer cashback faster using fewer internal resources. “The amount of money we have to spend on server costs to generate cashback for our customers is significantly lower with Amazon EC2 Spot Instances,” says Chua.

Today, when ShopBack’s ecommerce partners launch flash deals or participate in large sales, ShopBack can scale up to meet spikes in demand while keeping costs in check. “Now I just buy more Spot Instances, and I can clear our queue much faster,” says Chua. And once a high-traffic event is over, ShopBack engineers can easily scale resources back down. That enables ShopBack to sustain strong growth while keeping infrastructure costs low, which in turn means ShopBack can free up time and budget for engineers to focus on more strategic tasks.

Spot Instances were also easy to launch and manage. “Using Spot Instances doesn't change the way we view applications, so moving from On-Demand Instances to Spot Instances was an easy switch to make,” says Chua.

During the transition, ShopBack leaned on AWS Enterprise Support to ensure the most efficient use of infrastructure as it grew. “AWS Enterprise Support helped us make further tweaks to our infrastructure and gave us a better sense of how much we should scale up so we didn’t overspend,” says Chua. “Once we got through the busy season, the support team helped us focus on sustainable long-term growth.”

Because ShopBack has shifted most of its workload to Spot Instances, the company no longer experiences site outages related to spikes in traffic. And when it wants to launch in new markets, expansion is now easier to accomplish. Since making the change, ShopBack has dramatically grown its active customers, resulting in about four times higher use of the app and the site. Today, ShopBack serves over 20 million users across its nine markets, supports over 3,000 merchants, and has distributed US$100 million in cashback to its users.

Going forward, ShopBack plans to use AWS to manage expansion cost-effectively when launching in new markets. “If you buy services from different vendors, it gets difficult to integrate everything. With AWS, the managed services all work together, and security is built into it all. That means we can focus less on the technology and more on growing the business,” says Chua.

Chua says ShopBack plans to add more AWS services. For example, the company is working to provide more personalized user experiences, and it will explore AWS machine learning capabilities in products like Amazon SageMaker and Amazon Personalize. “We want to provide a customized experience for every user, including content and recommendations about what to buy based on user preferences, browsing history, or location,” says Chua. “We want to offer our users the kind of personalized content that they will enjoy and keep them coming back.”

To learn more, visit aws.amazon.com/ec2/spot.