Purple started with a materials problem. Brothers Tony and Terry Pearce spent years engineering a hyper-elastic polymer that could support a body without the pressure points of foam or springs, work that began in 1989 and eventually became the Purple Grid. In September 2015, the No-Compromise Purple Mattress launched on Kickstarter, and the company has since grown into a public manufacturer and retailer headquartered in Lehi, Utah, with roughly 1,100 employees and a product line spanning mattresses, bases, bedding and seat cushions.

Purple sells through nearly every channel available to a comfort brand. Direct-to-consumer ecommerce, more than 50 owned showrooms, national wholesale and club retail partners, and a phone sales team all move the same products to very different buyers. That mix generates a large and varied volume of sales data.

Purple’s data sits in the Snowflake Cloud Data Platform, and the reporting that runs on top of it is owned by a small but mighty data group led by Hyrum Ward, Director of Data Enablement. That team is deep in engineering and SQL. For years, it also absorbed most of the visual and reporting work for the rest of the company. Purple wanted that to change by moving to a different model that took demand off its data experts and enabled business users to answer questions on their own.

Self-Service Reporting at the Pace of Business

Purple’s existing reporting layer ran on Looker, and it had gotten slow. Some of the most-used reports took several minutes to load, which hurt adoption. This prompted Purple to move off Looker. After thoroughly evaluating the visual analytics market, they selected Sigma to replace Looker on top of the Snowflake layer. Beyond reporting, Purple was drawn to Sigma’s support for building AI-powered data apps and its native integration with Snowflake.

What Purple wanted was not a lift-and-shift. Hyrum’s team set three goals for the work. First, any business logic that had accumulated inside individual Looker reports needed to move upstream into Snowflake, where it could be governed and reused. Second, the tables underneath needed to be organized, documented and labeled well enough to serve as a single source of truth. Finally, someone needed to look honestly at the inventory of more than 1,000 reports and say which ones were actually driving decisions.

Behind those three goals sat a broader one, and it mattered most: governed self-service. If business teams in ecommerce, retail and phone sales could build and adjust their own reporting inside guardrails, the data enablement group could get back to engineering the foundation instead of servicing report requests.

Purple made one more decision that shaped everything that followed: it rebuilt its database at the same time. Running both efforts in parallel meant more coordination in the moment, but it also meant the new reporting layer would be built directly on the new foundation rather than migrated onto the old one and then rebuilt a second time.

Purple does not often bring outside help into its data estate, so the team was specific about what it needed from a partner. Delivery had to be onshore. It also had to cover the full report list, not just a comfortable subset of it. Purple came to InterWorks through Sigma, and InterWorks Account Executive Robby White and Solutions Director Anna Hadzic worked with Hyrum to shape a fixed-fee engagement around both requirements.

“The pricing was extremely competitive, and having onshore consultants we could work with every day made a real difference.”

– Hyrum Ward, Director of Data Enablement, Purple

Rebuilding on a Dynamic Data Foundation

Most of the original 1,000-plus reports had been retired before InterWorks arrived, leaving a much shorter list of reports still in active use. Hyrum had already sorted every report on that list by difficulty, which gave the scoping conversation a solid starting point. InterWorks turned that classification into an hours-by-tier estimate and built the fee around it. The estimate held up through delivery.

Two InterWorks consultants delivered the work over roughly 12 weeks. InterWorks Analytics Consultant Sean Spencer led the engagement and handled exploration, mapping and the daily conversations with Purple’s stakeholders. Fellow InterWorks Analytics Consultant Bretten Farrell handled the build in Sigma.

Because the database rebuild was underway, the tables and columns in the new environment did not line up with what the Looker reports had been written against. Sean and Bretten mapped the logic by hand, working from Purple’s reference documentation and verifying each measure against the new model rather than assuming a match. The team published one or two reports a week for the first few weeks, then three, then four and eventually six a week once it had learned the new data model.

Rather than starting with quick wins, the team started with the hardest report Purple had. The Company Flash is Purple’s cross-departmental KPI summary and its highest-traffic piece of reporting, and in Looker it lived across roughly 10 separate reports. Bretten rebuilt it as a single Sigma workbook with an executive summary screen and a summary screen for each department. Consolidating it first also surfaced the logic that repeated throughout the rest of the inventory, which accelerated everything that followed.

That consolidation pattern held across the engagement. Reports were regrouped and reorganized to suit how Sigma works rather than copied one-for-one, and the team asked stakeholders to make the case for each item still on the list. Many conversations ended with a Purple stakeholder deciding a report had outlived its purpose.

The reporting that was built reflects how Purple actually sells. Unit and dollar volume by product line, mattress type mix, then the same view split across ecommerce, physical retail and wholesale. Unit mix, pillow attachment rates and conversion. Call-to-sale reporting for the phone teams. Shipped-on-time and SLA attainment by department.

Purple also asked for reporting that had never existed in Looker. The company had been relying on its wholesale partner’s own forecasts to understand ordering patterns, with no independent way to check them. The new reporting measures that partner’s forecast accuracy against what Purple actually sold, which turns a recurring source of friction into a number Purple can bring to the table itself.

Every report was reviewed and signed off on by the stakeholder who owned it, with permissions configured and verified before handoff. The team finished the committed scope ahead of schedule, then used the remaining time to build additional reporting Purple asked for.

Handing Over Capability, Not Just Dashboards

Finishing early created room for the part of the engagement Purple got the most out of. Midway through, Bretten attended Sigma’s AI app training in San Francisco and brought it back as a lunch-and-learn for Purple’s team on building Sigma data apps, a capability Purple had become interested in as the migration progressed.

Then, in the final week, Purple made another key decision. When offered written documentation for the new dashboards, the team asked for AI skills instead. Purple had recently rolled out Claude across the organization and preferred something that generates documentation on demand over documentation that would go stale the day it was delivered.

Bretten built two skills. Sigma Atlas documents any Sigma workbook from a supplied link and flags development patterns worth a second look, such as oversized case statements. Sigma Report Builder lets a user describe a report in plain language and have AI build it, with InterWorks development standards written into the skill so new reports come out consistent with the ones already in the environment. Both were handed over at close, and both are now used with other InterWorks clients doing similar work.

“The AI skills Bretten built for us were excellent. They didn’t just hand us reports; they facilitated easier report building for us, and that has kept us moving forward long after the engagement ended.”

– Hyrum Ward, Director of Data Enablement, Purple

Well-Positioned for Whatever Comes Next

Purple set out to push its business logic somewhere governable, turn its tables into a documented single source of truth, get honest about which reports it was actually using and move the company toward self-service. Purple achieved all four.

More than 40 Sigma workbooks are in production, covering roughly 55 of the original reports once consolidation is accounted for, and about 15 more were retired. Company KPIs live in one place instead of 10. The tables underneath are clean, documented and labeled, which is what makes the self-service model workable rather than risky. Business teams can build against a governed layer, and the data enablement group is no longer the bottleneck between a question and an answer. Purple can measure its wholesale partner’s forecasting with its own numbers, and documentation for the environment is generated from the environment itself.

InterWorks’ recommendations point in the same direction as Purple’s original goals: train business stakeholders to build properly in Sigma, and put governance around the environment as more of them come on board. Both are the natural next moves for an organization that has decided self-service is where it wants to be.

Hyrum and Purple’s data enablement team deserve much of the credit. In moving from a familiar Looker environment to something bolder in Sigma, they made it clear throughout the engagement that “good enough” was not a standard they would accept.

For a company that got here by engineering its own materials rather than buying someone else’s, this is a familiar position. Purple now has a reporting foundation it understands, controls and can extend without waiting on anyone, which is exactly where it wants to be when the next round of questions arrives.

“InterWorks was easy to work with, and they delivered. Their work was top-notch, and they went above and beyond, delivering everything in scope and then helping us build additional reports we hadn’t originally planned for.”

– Hyrum Ward, Director of Data Enablement, Purple