Democratizing data access to deliver personalized travel recommendations

About

Priceline began in 1997 with ten people and a simple plan to use the internet to let customers name their own price for empty hotel rooms and airplane seats. Today, the organization is part of the world’s largest travel company, Booking.com. Though still a small company with only 1000 employees, Priceline makes a big impact, helping consumers save over $1B a year on travel.

The organization has structured datasets stored in multiple on-premises and cloud databases. While users can run ad hoc queries and create reports using Tableau, their data access is limited to structured data — they need a simple, fast way to access and explore raw, unstructured data.

Challenge

Priceline has data stored in multiple on-prem and cloud systems, including Google BigQuery, Google Cloud Storage data lake, Oracle RDBMS, and CloudSQL instances in Google Cloud Platform. Previously, users only had access to curated, transformed, or refined data stored in warehouses. They couldn’t query raw data stored in unstructured formats, such as historical data — a developer or data engineer would need to provide it.

In addition, this structured format increased compute and storage costs. Priceline wanted to create a decentralized data mesh infrastructure to provide users with a more comprehensive view of the data, enabling them to explore it in new ways.

Solution

Priceline chose Starburst Enterprise to serve as a secure, high-performance analytics engine and data mesh enabler. Starburst provides a federated query layer that lets users tap into multiple heterogeneous data sources, regardless of where or in what format it resides. Users get faster time-to-insight and no longer wait one or two days for data processing — improving both internal insights and customer-facing application performance.

“We can make decisions faster based on the analytics,” Menon says. “Instead of waiting for days, this happens in near real-time.”

Starburst also separates storage and compute, allowing Priceline to move a significant volume of data to an unstructured, raw format, lowering costs without compromising performance.

Results

Priceline wanted data access, security, performance, and cost savings. Menon notes that Starburst provided all of these, as well as four primary benefits:

• Tailored recommendations for shoppers based on their historical preferences.
• Freedom for data consumers to be curious and explore new patterns and trends. 
• Faster time-to-insight.
• Reduction in cloud storage costs by 5X to 10X.

Users no longer need to wait days for data processing so they can analyze unstructured data. With Starburst, Priceline can leverage and directly access vast amounts of streaming and historical datasets. This gives the organization the ability to make data-driven decisions in near real-time, while also providing detailed, accurate recommendations and historical trends to customers.

“One of the key reasons we are on this journey is to democratize data,” says Menon. “By using Starburst, we can make the data available for all users and, then, make more informed decisions with data sets that we never explored in the past. We can see the hidden data and leverage it to provide an even better customer experience.”

More resources: Sachin Menon’s Voyager profile