AB Tasty: Optimizing customer experience with experimentation and machine learning
About AB Tasty
AB Tasty is a leading customer experience optimization SaaS provider, helping enterprise companies to understand their customers and act on that knowledge.
Tell us your challenge. We're here to help.
Contact us
About SFEIR
SFEIR is an industry recognized leader in advanced applications development and in delivering customized new technology and IT training programs, serving 100+ clients every day.
AB Tasty redesigned its architecture with Google Kubernetes Engine, Bigtable, Dataflow, BigQuery, and Pub/Sub to deliver enhanced customer insight at extreme scale.
Google Cloud results
- Enables clients to run their own optimization experiments and receive customized recommendations in near real-time with BigQuery
- Empowers developers to create and test new features in Google Cloud Data stack instances, without worrying about IT resources
- Provides solid stability with a 99.99% service level agreement
Analyzes approximately 2 billion events per day
Whether you’re a small business that’s just starting out or a well-established brand, understanding how your customers interact with your website can provide invaluable insights. You could discover that your customers spend more time on particular pages or your bounce rates, and these insights can help you optimize your business opportunities. AB Tasty is a customer experience optimization company that helps brands build better user experiences and unlock new possibilities through the power of experimentation.
"Our job is to optimize our clients’ web traffic for their unique needs," says Sergey Chernyakov, VP Data and Infrastructure, AB Tasty. "If they’re an ecommerce company, for instance, we focus on conversions. If they’re a media company, we focus on retention and engagement. We give them the tools and insights to better understand what their customers want from their online experience."
AB Tasty distinguishes itself from its competitors with both the quantity and quality of the data it offers to clients. "Our platform can handle massive amounts of web traffic, but we’ve also built quality assurance pipelines to monitor the data we ingest," explains Chernyakov. "We designed our architecture to handle this kind of load and provide the best quality data to our clients."
In 2017, AB Tasty was reaching the limits of its technical infrastructure. Bigger clients meant more operations management, and Chernyakov and his team realized that they needed not just a bigger infrastructure, but a new kind of architecture. They designed and built it with Google Cloud.
Delivering optimization through experimentation
The AB Tasty platform allows its clients to experiment with new website designs, apps, and other features with an intuitive, drag-and-drop interface. With access to a variety of A/B, split and multivariate tests, clients can see the effect their experiments have on their users and what approaches they respond best to. AB Tasty also provides a personalization service, which uses AI and powerful analytics to help clients create the most resonant campaigns for their chosen demographics.
In its early years, AB Tasty relied on a cloud-based infrastructure designed around virtual machines. By 2017, however, this approach was showing its limits. "It was complex to manage with a small DevOps team," says Sergey Chernyakov. "When we took on bigger clients, managing all that traffic was very painful for us."
Sergey Chernyakov and his team realized that a VM-based infrastructure was not going to be able to handle the scale at which the business was operating. What AB Tasty needed was a new architecture built around microservices. After evaluating its options, AB Tasty ran a proof of concept with two major cloud providers. The winner was Google Cloud.
"We wanted to migrate to a microservices-based architecture for scalability but we had limited DevOps resources," says Chernyakov. "With the automations and managed services available with Google Cloud, it was the natural choice for us."
"We wanted to migrate to a microservices-based architecture for scalability but we had limited DevOps resources. With the automations and managed services available with Google Cloud, it was the natural choice for us."
—Sergey Chernyakov, VP Data and Infrastructure, AB TastyBuilding a platform for scale
With the help of Google Cloud Partner SFEIR, AB Tasty perfected its new architecture during the proof of concept stage, working out the right combination of products to achieve what it needed. When the design was finalized, the company began migrating its clients over to Google Cloud.
AB Tasty’s platform is event-driven. In this case, an event is created every time a user interacts with a client's website or app, for example by clicking on a link. When a client takes on AB Tasty, they integrate the platform's Javascript code into their products, allowing AB Tasty to register each event. These events are streamed with Google Kubernetes Engine and Pub/Sub into Google Cloud Storage buckets and BigQuery. "Pub/Sub makes streaming events data incredibly easy and it is very cost effective for the amount of traffic we put through it," says Sergey Chernyakov. "You don’t have to worry about scalability or replication across regions, you just push your traffic and pull your traffic."
With Dataflow, the company streams the events data into BigQuery, which serves as AB Tasty’s serverless, highly scalable data warehouse. Throughout the pipeline, the data is monitored for errors, ensuring that customers receive the highest q data possible. With BigQuery, Sergey Chernyakov and his team can run analyses and deploy the machine learning tools that feed into the platform’s personalization features. "Typically, we’re trying to predict whether users will buy a product or not, or stay on a page," he says. "The machine learning models we need are available to implement out of the box with BigQuery. It’s very simple to create the model, train it, and deploy it."
AB Tasty creates a dashboard for its customers where they can get an overview of the experiments they run and their tailored recommendations. Thanks to BigQuery, clients with more technical resources can access an export of raw data, giving them the freedom to investigate and explore their results in whatever way they see fit.
"Pub/Sub makes streaming events data incredibly easy and it is very cost effective for the amount of traffic we put through it. You don’t have to worry about scalability or replication across regions, you just push your traffic and pull your traffic."
—Sergey Chernyakov, VP Data and Infrastructure, AB TastyBigger, faster, more customized
Since migrating to Google Cloud, AB Tasty has had the infrastructure designed for scale without having to compromise on its quality of service. "I’m really proud of the architecture we built with Google Cloud," says Sergey Chernyakov. "It means we can handle massive amounts of traffic and it doesn’t affect our service at all. In fact, we’ve achieved a 99.99% service-level agreement."
The previous infrastructure could handle a maximum of around 5,000 events per second. The new solution usually handles around 50,000 events per second, with up to 200,000 events per second at peak times, without affecting performance. Ingesting, processing, and analyzing around 2 billion events per day, AB Tasty has been able to expand its range of clients and provide near real-time service.
AB Tasty has also used the analytics and machine learning capabilities of BigQuery to enhance its personalization product. The company achieved this with the same size DevOps team that worked with the previous infrastructure. This is due to smart automations and managed services with Google Cloud. Developers are now able to create and test new features and ideas by simply spinning up an instance with Google Kubernetes Engine without having to worry about provisioning or draining IT resources, resulting in richer, more mature products.
This has all contributed to AB Tasty’s continued growth over the last few years. The company has over 1,000 enterprise clients, including some of the biggest brands in the world, such as Disney, L’Oreal, and USA Today. Even as the company grows, it continues to evolve and experiment with new ways of working. "What’s great about Google Cloud is how well it integrates with open-source technology. That’s something we’re really keen on exploring in the future," says Sergey Chernyakov.
"I’m really proud of the architecture we built with Google Cloud. It means we can handle massive amounts of traffic and it doesn’t affect our service at all. In fact, we’ve achieved a 99.99% service level agreement."
—Sergey Chernyakov, VP Data and Infrastructure, AB TastyTell us your challenge. We're here to help.
Contact us
About AB Tasty
AB Tasty is a leading customer experience optimization SaaS provider, helping enterprise companies to understand their customers and act on that knowledge.
About SFEIR
SFEIR is an industry recognized leader in advanced applications development and in delivering customized new technology and IT training programs, serving 100+ clients every day.