The Challenge
Priceline has a plethora of customers using its website and mobile apps at any given moment. To maintain leadership in a competitive market, Priceline needs to offer a compelling product that gives customers great flexibility in planning trips and making purchases across multiple travel categories. “Part of our ethos is continually innovating in the travel space with ongoing, accelerated testing for new product ideas, harvesting insights, and quickly rolling out new services and features our customers will like,” says Marty Brodbeck, CTO of Priceline.
The company wanted to improve the productivity of its product development and engineering teams and modernize its data and application platforms. Priceline ran an open-source Apache Cassandra database on-premises to support the massive throughput needed for its operations, but managing the environment became too taxing to handle internally.
The Solution
The pandemic afforded Priceline the chance to accelerate its business transformation initiatives, centered around migrating to the cloud and transitioning to a microservices-based architecture. “We fast-tracked our technology transformation to enable us to improve our product more effectively and make it easy to ship features for our customers,” says Brodbeck.
Coinciding with its cloud migration, Priceline shifted to a managed services approach for its Cassandra database using DataStax Astra DB on Google Cloud. Astra DB is a vector database for Generative AI that provides the high performance, availability, and scalability Priceline needs for real-time data, supported by the global reliability of Google Cloud.
Astra DB is a powerful component of the Priceline data infrastructure. Its machine learning algorithms use massive data troves to provide valuable customer insights, greater personalization, and better travel recommendations, fueling the company’s larger customer ecosystem.
“Real-time data infrastructure is important to e-commerce companies like ours because it’s the backbone for searching, pricing, booking, and enabling us to customize our products to customers,” says Brodbeck. “We can capture data events as they occur, instantly generate insights and recommendations, and provide better experiences for travelers.”
The Results
The new data platform has lightened the infrastructure management load on Priceline’s engineering organization. “We’ve definitely enhanced our ability to launch product improvements,” Brodbeck says. “Companies often exhaust a lot of data engineering equity because it can be so time-consuming to normalize data, create a pipeline that can capture Kafka events, and manage database nodes. With the streamlined and synergistic capabilities of DataStax and Google Cloud, we can do a lot more with less from a data engineering perspective.”
Priceline has hundreds of A/B tests for new product features running on any given day. One of the main responsibilities of its data engineers is to manage the testing platform, which is data-intensive, triggering thousands of events that the engineers need to evaluate. “We measure the value of new features via metric that we call NIBPD, which is net incremental bookings per day. The more we can free up our engineers to be focused on boosting NIBPD by delivering value for customers, the better it is for the company,” says Brodbeck.
During the pandemic, Brodbeck observed how modernizing the company’s technology stack helped Priceline quickly react to changing market dynamics. “We picked up signals for a new trend around road trips, so we were able to pivot rapidly to debut new features around renting cars and recommending hotels within driving distance,” Brodbeck says. “I don’t think we could have reacted so dynamically without the flexible, real-time data infrastructure we have now.”
Brodbeck credits investments in technology innovation and co-development initiatives with partners such as DataStax and Google Cloud with helping Priceline hire and retain top talent. “We’ve scaled up our engineering capabilities over the last four years, nearly tripling the size of our team. Providing best-in-class tools that help our people do their jobs effectively and opportunities to participate in leading-edge innovations around AI and real-time data helps us attract great people and provide them with great experiences. It makes a huge impact on our company culture and value proposition for our staff,” he says.
In his plans for next year, Brodbeck is exploring DataStax Astra Streaming, a managed service powered by Apache Pulsar, to fuel its machine learning capabilities with streaming, real-time event data feeding seamlessly into its Google Cloud environment. “With DataStax’s acquisition of Kaskada, we anticipate we can further enhance our ML search and sort algorithms and infuse greater personalization on the fly,” says Brodbeck.
Brodbeck was also excited that DataStax added Vector Search to Astra DB on Google Cloud. “Vector search gives us the ability to semantically query the billions of real-time signals we receive as part of our checkout experience that flow back to Astra DB. We plan to use Google Cloud’s generative AI capabilities alongside Astra DB’s vector search to power our real-time data infrastructure and generative AI experiences."
“We are fortunate to have strong partnerships with both DataStax and Google Cloud,” Brodbeck says, “making it possible for Priceline to ‘get travelers to their happy places at happier prices,’ as our motto goes.
(1.) Source: https://press.priceline.com/priceline-celebrates-25-years-with-25-days-of-deals/
