Download Case Study

DAS42 gave us a lot more confidence in our approach. They suggested a lot of great optimizations, and they’ve brought together Snowflake with our BI layer.

– ZAC SHEPHERD, DIRECTOR OF ENGINEERING FOR DATA, BUSINESS INSIDER

News media giant Business Insider publishes original journalism and aggregated content across multiple verticals such as business, finance, technology, and lifestyle. Founded in 2007, they operate with an online delivery model using ad-supported and subscription-based content models.

The challenge

Business Insider’s users found it challenging to access a comprehensive view of the business, often limited to examining data from one data source at a time. They also needed more trust in the data due to conflicting numbers, sources of truth, and definitions of metrics. End users struggled to determine where to find an answer because there were too many options in data warehouses and business intelligence (BI) tools, with duplicate content residing within the same tool created by different developers. All of this contributed to a fragmented view of data, resulting in the technical teams spending a lot of time chasing down bugs instead of building out new workloads or optimizing existing ones.

The solution

To enhance trust in the data, DAS42 created alerts in Looker to identify late data or unusual variances, which helped serve as an early identification system for data and website issues that caused dramatically reduced traffic. We modeled a single source of truth for website traffic data. This approach allowed multiple teams to use it, ultimately leading to the deprecation of more than 65 use-case-specific Explores that had been created for individual teams. Consequently, visibility into the ways different teams accessed the data increased.

DAS42 assessed each of the core tools in Business Insider’s data stack (Fivetran, Airflow, dbt, Snowflake, and Looker) and presented plans for optimizing for cost, performance, reduction of duplication of effort, user experience, and ease of maintenance. Key to this optimization was a custom, long-term Looker roadmap. We then cleaned up the Looker environment beyond the consolidation mentioned above. We created custom rules for a code linting/compliance enforcement tool (Look at Me Sideways) and implemented a customized version of DAS42’s LookML style guide. We enabled cross-explore drill functionality and aggregate awareness, which helped optimize costs.

How did it turn out?

With fewer Explores, Looker end users found the platform more accessible to navigate, locate necessary information, and determine which Explore to use to answer specific questions. The cleanup facilitated removing more than 2,000 duplicate fields, minimizing conflicting reporting. DAS42’s roadmap enabled Business Insider’s in-house team to reduce from seven internal Looker projects to one (an 86% reduction) and from 85 models to 11 (an 87% reduction). We also added a cross-Explore drill functionality to Permutive Explore, which reduced drill query runtime from more than three minutes to two to five seconds.

DAS42 also provided consistent guidelines for development, including automated code compliance testing (“linting”) with eight default rules and ten custom rules, which serve as preventative maintenance to help keep organization and best practices in place. Finally, we held training and enablement sessions for Looker developers, Explorers, and viewers so that users better understood the tool’s capabilities and had additional learning resources available when they wanted to take their skills to the next level.