One might assume that a manufacturer that is vertically integrated — with direct control over delivery routes and the customer purchase experience — would have no trouble devising a workable demand plan. If only that were the case.

Sleep Number is a rising star in the world of bedding and mattress sales. Its innovative technology draws on billions of data points to deliver the optimal sleep experience to users. In addition to plants in Irmo, South Carolina and Salt Lake City, Utah, the company operates a network of more than 600 retail stores.

That integrated approach to operations extends to corporate responsibilities as well. Chris Meyer, who holds the title of senior finance manager, not only supports the company’s sales organization, real estate activities and “relationship center,” but also plays a lead role in demand planning.

That last role isn’t an easy one, notwithstanding Sleep Number’s insistence on maintaining direct control over its outbound supply chain. Failure to get the forecast right means scrambling to build more product, and relying on expedited delivery to prevent impatient customers from defecting to the competition. Yet the company was having problems predicting unit demand and tying that to its supply-chain and logistics requirements, Meyer said at Gartner’s recent Supply Chain Planning Summit in Denver, Colorado.

Sleep Number, ironically, was experiencing extreme variability in its numbers. The process of developing a forecast was heavily manual, spreadsheet-based, and time-intensive. Errors were common, impacting decision-making company-wide. The company describes its plight at the time as “data-rich and insight-poor.”

A new demand-planning tool was the answer. Sleep Number scrutinized four vendors, all of whose products “were solid and could do the work,” said Meyer. The problem was that the sophistication of most of those solutions didn’t match that of Sleep Number.

“We were early on the maturity curve in getting to where we wanted to be,” said Meyer. “If we were to go the path of others, it wouldn’t have allowed us to use [the software] for multiple business units. It would have forced us into a way we weren’t prepared for, and created a lot of inefficiency.”

Suffering from what Meyer described as “RFP overload,” Sleep Number entertained a proposal by Anaplan, a specialist in cloud-based “connected planning.” One day before attending a workshop with the vendor, Meyer provided it with data from an earlier planning event that had gone wrong.

Anaplan came through with flying colors, applying analytics to solve the planning problem presented by Sleep Number. “It was so quick and easy to use.” Meyer said. “We realized at that point that we were on to something.”

With Anaplan on board, Sleep Number embarked on a 13-week implementation of the demand-planning software, going live in February of 2019. The first order of business was setting up a data hub and learning to use the tool, but the company was already looking ahead. “We intentionally routed more information than we needed into our data hub so it was scalable for future use,” Meyer said.

One month later, the company rolled out sales-forecasting capability, completing the task in nine weeks instead of the projected 13. Meyer called the project to date “an absolute success, to the point where other people in the company are coming to use it.”

Sleep Number’s decision initially to focus on the business-unit level sped up its return on investment from the tool, Meyer said. The next step is to expand it to the delivery stage of operations.

Unity of effort across the company, which has previously struggled to conform conflicting forecasts, is a priority. Sleep Number sought out a solution that could be used harmoniously by finance, sales, manufacturing and supply chain.

“We’re all holding hands,” said Meyer. “All business units are presenting a unified front in terms of the numbers they submit. That gives a lot of power to the tool.”

Improved accuracy made possible by the tool ensures that the delivery schedule is correct, and eliminates the need to reschedule appointments caused by a lack of inventory at the hub.

Since implementing the software, Sleep Number’s forecast cycle time has been slashed from eight hours to just three minutes. Planning for an entire season, which used to take three and a half days, can now be done in eight hours.

What’s more, the company can “surgically” change unit forecasts, ensuring accuracy down to the SKU level. Meyer said it has seen a five-point improvement in SKU accuracy nationwide, and recently experienced 99% accuracy over a 25-day period during a recent sales event. “That’s nothing short of magical for us,” he said.

With demand planning up and running, the company is looking for opportunities to go deeper, particularly through application of a “one-stop shop” dashboard for sales. The goal, said Meyer, is “to drive forecast accuracy at our stores across the continental U.S.”

The elimination of manual effort has freed up individuals throughout the company to focus on boosting sales in an increasingly crowded business. Said Meyer: “Now people are doing what we pay them to do.”