Establishing one source of governed truth

Wayfair implemented its unified analytics platform in phases. One key phase, driven by its data engineering and analytics teams, used Looker's modeling language, LookML, to build a semantic data layer. The layer sits between users and Wayfair's global data in more than 500K+ tables and defines how data can be consumed. So instead of configuring business logic, access controls, data models, and KPI calculations for individual business systems and data marts, engineers define all these parameters and controls once in the semantic data layer. And because all employees access their data through this layer, their insights always align. Additionally, IT teams establish and maintain consistent data structures and governance policies. “I'd add that restricting the number of teams that define and update models and metric calculations is another way to improve governance in a semantic layer,” says Ganzfried.

Simplifying and accelerating BI creation

To help employees feel confident in creating their own BI with Looker and the company's semantic data layer, Wayfair's engineers used LookML to provide employees with a library of Explores. Not only do Explores let teams drill into business metrics and drivers of those metrics across the organization but they are also fully aligned with the same data whether accessible via Looker, Looker Studio, or Connected Sheets for Looker.

From Looker, engineers used Explores to quickly build some initial dashboards for Wayfair's business and IT teams that provide them with instant multidimensional insights that they can drill down into. “Once we built dashboards and showed users Explores, our solution took off,” says Ganzfried. “Teams were like, ‘Okay, what's that?’ And after seeing what they could do with Looker, they wanted their engineering teams upskilled. So I'd say starting your implementation with the savviest data teams is a good strategy because they can help lead business teams on the journey in self-service analytics.”

To ensure the rapid multidimensional exploration of data from reports and dashboards, engineers built aggregate tables with LookML to support the most common queries. For the same reasons, they also used LookML to build persistent derived tables that contain the most popular datasets. Additionally, to continue delivering the reports that employees expect, engineers used the Looker API to automatically send PDF versions of the dashboards they created that show a unified view of critical daily and weekly metrics.

Expedites time to insights with the right self-service BI experience

Today, in just seconds, users can answer many questions that previously would have required IT help by just drilling down into the information that's supporting the surface-level insights in Looker-generated BI. When they need to see data differently or pull in more information to answer questions, employees can do it themselves — often in just minutes — instead of waiting weeks for IT teams to create or edit reports and dashboards. That's because users have the tools and data they need to quickly create the BI they're looking for in a way that's comfortable for them. “We can move faster with Connected Sheets for Looker,” says Ganzfried. “It provides another avenue for self-service. With it, we can make sure everyone can answer the questions that they have—at any time.”

Increases user confidence in creating BI

By equipping its global teams with one governed semantic data layer as well as Looker Explores, Wayfair is seeing a cultural shift, where business users are confidently answering their own questions and creating reports and dashboards that involve other business domains. For example, a marketer who has an operational question like slow delivery times can now access and analyze operational data on their own. Regional executives can also compare how they're doing compared with other regions. “Being able to easily see the data you need to answer your questions and know that the numbers you're seeing are accurate gives our users the confidence they need to work with and report on data that's outside of their standard domain,” says Ganzfried. “As a result, strategic decisions are being made significantly faster and more confidently.”