Travelpass’ Journey to Deliver More Curated Traveler Experiences — at Half the Cost

Have you ever considered how the act of getting away can actually bring people together? How it can strengthen our closest relationships and expand our understanding of others? 

Travelpass, a Utah-based travel-experience platform, was built on that very idea of discovering common ground through exploration. The company helps travelers satisfy their wanderlust by connecting them with the best hotels and travel experiences to meet their needs, while inspiring them with well-informed, curated content from its growing community of globetrotters. 

To achieve this, Travelpass manages a host of constantly variable information — from hotel inventory and pricing to restaurant recommendations and sightseeing reservations. “Data is the lifeblood of our business,” Data Architect David Webb says. “I don’t think we could build the platform that we want to build, where the community can share with each other, if data was not part of our DNA — and Snowflake is vital to that.”

After contending with high maintenance and a crash-prone system, the company looked to Snowflake for a platform that was robust, yet easy to use; offered peace of mind when it came to data governance and cost controls; and could help tap into the vast power of AI.

One of the reasons Travelpass moved to Snowflake was, as Travelpass Data Architect David Webb puts it, “we were spending a lot of time focusing on how to build, rather than what to build.” From their previous labor-intensive platform  — with a variety of complexities — Webb’s team was debugging errors and navigating job failures more often than they would have liked, often finding that one small change would take down the entire system. 

He recalls an instance where a data set containing insights into some of the company’s ads was suddenly missing entries. “I remember one day waking up, and the marketing team said, ‘Hey, what is going on?’” he explains. 

Webb’s team spent several days investigating, only to realize that the issue lay in how the files were named. “They had to be named in a specific lexical order, otherwise the system wasn’t going to see it,” he says. “We were constantly having to make changes to small things like that. It was a horrible problem that took days to figure out.”

The team would also routinely run into memory issues or be forced to restart the entire system after a simple tweak, like adding a new column to a data set, caused it to crash.

Almost immediately, Travelpass saw the benefits of true cooperation, enlisting Snowflake Professional Services to help streamline the migration process under a tight deadline of around 45 days. “Snowflake’s Professional Services team was amazing to work with,” Webb says. “Talking to them, you understand that they know what they’re doing, so you can trust them to put a job forward and get it done.”

Seamlessly transitioning 30 TB of data and 134 pipelines/processes took just six weeks. “And we didn’t have to disrupt the business to do so,” Webb says. “That’s a testament to the Snowflake system.”

Now, Travelpass has clear visibility into its spend — a challenge under their previous platform’s opaque system. After swiftly completing the migration, Travelpass found that Snowflake cut costs in half compared to its previous spend.

And because of Snowflake’s ease of use and approachability, Webb notes that “even non-data engineers are contributing to our Snowflake data in a really meaningful, quick way.” For instance, features like Dynamic Tables, which can be declaratively created with a simple SQL statement, greatly reduced the complexity of building pipelines. Unlike Databricks’ Delta Live Tables, which require additional orchestration, management and significant time writing code, Dynamic Tables allowed Travelpass to operationalize more of its workers — backend engineers, analysts, anyone with some SQL experience. 

This feature has been so effective that the company had more than 130 Dynamic Tables in its production environment within a week of completing the migration to Snowflake. Travelpass estimates the feature made delivering data to business units around 350% more efficient and helped quadruple the number of code optimizations the team could churn through in a week.

As for just what to build, Travelpass is eager to explore the boundless opportunities around giving customers bespoke travel experiences, now backed by the power of AI. That vision, however, goes beyond simply prompting an LLM: “Give me a three-day itinerary to see Tokyo.” The results, while adequate, won’t reflect the unique traveler — their interests, how they like to get around, what types of properties they prefer, their desire to taste dumplings from around the world or their penchant for street art. 

With a growing catalog of Cortex AI tools and capabilities built directly into Snowflake, Travelpass can easily see ways to realize its vision. “The team isn’t sitting down and saying, ‘How am I going to build AI?’ They can now answer the question, ‘How can I use Cortex AI to build something else?’” Webb says. “And enabling that on the team changes everything.”

As the company grows its community of explorers and deepens its data stores, it ultimately hopes to become everyone’s cosmopolitan, jet-setting friend — the one who’s been there, seen that, dined at the finest restaurants and discovered the hole-in-the-wall hidden gems. It aims to give travelers the kind of unique advice and curated recommendations that take the stress out of planning and make for unforgettable getaways — not only to get more out of their trips, Webb says, “but to just get more out of life.”