Everquote: Delivering Insurance Policies Online Using Real-Time Data Insights

Everquote logo
View PDF Button Arrow

EverQuote — one of the largest online marketplaces for insurance — empowers customers to better protect their most important assets, whether it’s their family, property, or future. Through the use of data and technology, EverQuote aims to be a trusted source for simple, affordable, and personalized insurance policies.

While EverQuote was data-driven from the start, the company couldn’t scale its home-grown technologies across the organization to business users. Once EverQuote shifted away from its in-house online analytical processing (OLAP) solution to Snowflake using AtScale, the insurance platform was able to dramatically accelerate its data-driven insights.

Snowflake logo Microsoft Excel logo Tableau logo


Limitations with scaling in-house custom OLAP solution to business users

Since EverQuote was a spinoff from Cogo Labs, a technology-driven venture accelerator, the new company already had a data architecture consisting of a MySQL cluster, Python services, a direct connection to Excel, and a custom OLAP interface. But this custom legacy approach, which was over ten years old, had a number of bottlenecks that prevented many use cases and suffered from poor query performance.

The other challenge EverQuote faced was the limited technical knowledge throughout the organization. While EverQuote started as a group of engineers and analysts, as the company grew, it proved difficult to scale self-service analytics to non-technical employees. The company needed a modern data architecture that could democratize data analytics for all.

“Our custom OLAP tool has served us very well for over a decade, but it’s owned by engineering,” stated Jonathan Spencer, Data Engineering Lead at EverQuote. “Over time, the tool has required a lot of maintenance and governance by the engineering team. It also wasn’t flexible enough to serve all analytical needs.”



Migrating from legacy OLAP to Snowflake’s cloud data platform using AtScale

Using AtScale’s Semantic Layer, EverQuote was able to seamlessly transition its analytics workloads to Snowflake without impacting the existing business user experience. The semantic layer now enables the business team to access data stored on Snowflake within TableauExcel, and many other business analytics and visualization tools.

Through AtScale’s semantic modeling capabilities, EverQuote is able to flexibly add new metrics and data definitions that provide consistency across consumption tools. Data virtualization also makes it easier to onboard new data quickly so that it can be queried from BI tools almost immediately.

“We found AtScale, and it became the right tool at the right time to revamp our custom OLAP tool,” Kwan Lee, EVP of Engineering explained. “In 2020, we adopted Tableau, and this year we started allowing business users to create data cubes in AtScale, which has been very useful for our organization.”



Democratizing data analytics for both business and data science teams

By modernizing its data architecture with Snowflake and AtScale, EverQuote have been able to dramatically reduce the time-to-insight for business users. In fact, the ability to perform both incremental and full cube refreshes has enabled EverQuote to provide more real-time data to all downstream lines of business.

With AtScale, EverQuote has also been able to more easily leverage its data for machine learning use cases. Data teams can now use Jupyter and other Python-based approaches to generate data science insights against Snowflake data. This has broken down the traditional silo between business intelligence and data science teams so that both teams can produce better insights for the organization.

Now that users across the organization can leverage data using the tools of their choice, EverQuote has shifted its culture to embrace data as a strategic asset. By promoting data literacy through training and offering the right business intelligence tools, EverQuote has also made self-service data analytics a reality.

“The challenge for data engineering teams should be building self-service components that can be used by other teams,” concluded Spencer. “Using technology and the principles of self-service, we can reduce the cycle time for those looking to do analytics and give them more autonomy in their own business.”