Charmer Sunbelt Group CIO Paul Fipps doesn't usually make IT investment decisions on instinct alone. Except, that is, when it came to implementing SAP's new in-memory analytics engine.
The $4.9 billion wine and spirits distributor had been an SAP shop since 2003, but never saw a need for the vendor's business intelligence tools. IT exported ERP data into Microsoft analytics services for finance and sales analysis and ran custom reports in the core SAP system for operations. "SAP had a traditional business intelligence tool," Fipps says, "but there was no real business case for it."
Normally, every IT investment goes to a strategic governing body for cost-benefit vetting, says Fipps. But the executive vice presidents of all functional areas green lit the in-memory analytics project without any articulated hard benefits or even a pilot. "We jumped into this way ahead of what we would normally do," Fipps says.
It was the promise of near real-time analytics that attracted Fipps to SAP's high-performance analytic appliance (HANA), which places data in main memory rather than reading it off disks. The 15-plus-year IT veteran envisioned HANA streamlining the company's most labor-intensive processes: the pick-pack-and-ship work that takes place in the wholesaler's warehouses.
Warehouse managers do their best based on experience and intuition, but operational performance varies wildly from location to location. More timely analysis could help them better manage inventory and staff, maximizing productivity. "Most of our labor costs happen at night," Fipps says. "If we could give these guys the information to make better decisions [then], that would give us a competitive advantage."
There will be next-day benefits too, says Fipps. The company will use HANA to create a daily, visual analysis of operational performance for company executives so they can make decisions that lower costs or increase customer satisfaction. Previously, this time-intensive analysis wasn't feasible for daily reporting.
The implementation was halfway done in June with plans to go live later this month. "It's going to be bumpy," he says. "Right out of the gate, we had problems getting the [Dell] hardware and [SAP] software here. Just getting started is difficult because it is so new."
He wouldn't have taken the chance with any other functional group. "Operations is very agile and all about continuous improvement," Fipps says. "They want that next advantage to drive productivity."
The biggest question for Fipps has been, When does fast data crunching make sense? "We're making decisions about what needs to be done within 30 seconds and what needs to be done within fifteen minutes," Fipps says. Time card data won't need to be updated every half minute; pallet replenishment data will. One pleasant surprise has been how easy it is to model new data views once the architecture is in place, says Fipps.
Once the system is live, all eyes will be on operational metrics such as man-hours per case to determine whether the system is delivering as hoped. "The true measure of success will be [whether] we can we move the needle in real time," he says. "How many trucks can we receive per day? How fast can we get that Grey Goose off the truck?"
Fipps pictures monitors throughout the warehouse letting everyone know how they're doing. Eventually he hopes to use in-memory analytics for pricing and financials. He also has a backup plan to his SAP contract.
"We rarely use prayer as a corporate strategy," says Fipps. "But this is a unique situation."