Grocery stores need to maintain the right balance of supply and demand to meet the needs of their customers on a daily basis, as well as during surges such as this month’s pre-Thanksgiving period. The goal is to prevent out-of-stocks due to brisk sales and targeted promotions.
The latest demand forecasting technologies can take place at both the store and at the distribution center. But are grocers taking advantage of these tools?
Some experts say retailers still rely too much on manual ordering and intuition to execute the replenishment process. It’s not common to employ more sophisticated methods to predict consumer demand and to use this knowledge to automate the required inventory estimates to satisfy demand.
“We see demand forecasting as a capability that most grocers need, but few extend this capability to the level where harnessing consumer demand is a competitive advantage in managing the business,” notes Tim JW Simmons, general sales manager, North America, demand chain solutions and services at Dayton, Ohio-based Teradata. “Not many grocery retailers have fully exploited demand forecasting capabilities to include store item-level forecasts that include baseline needs and support promotion events, or the planned launch and support of new products.”
According to Simmons, smaller operators have been slow to adopt these tools and gain experience to support decision-making. Most continue to rely on last year’s shipment-withdrawal data or rolled up t-log data in tools like Excel to help plan their activities for distribution or buying.
“While almost all retailers in grocery use or have a forecast, it is not typically regarded as accurate or used with confidence, particularly in the prediction of promotion demand,” he says.
And that’s when out-of-stocks typically occur.
“Hitting this problem with a single-forecast mentality doesn’t work,” asserts Alan Lipson, global retail/CPG industry marketing manager for Cary, N.C.-based SAS.
He says the key is to address the problem on two fronts: at the distribution center to support desired service levels with better demand forecasting, and at the store level with better overall forecasts.
“The best forecasting technologies generate forecasts for promoted and regular-price products. The models generating the forecasts should account for time series, life cycle and regression components. The technology used also should include functionality to support new-item forecasts that extend beyond the simple assignment of a ‘like item.’ You’ve got be granular to get there,” he observes.
According to Makarand Deshmuck, VP at Hoffman Estates, Ill.-based Sears Retail, the objectives of demand forecasting are twofold for a grocer: footfall and profit maximization.
“A lot of grocers are after one of these objectives,” he says. “For mass retailers such Target and Walmart, grocery is a subset of the total business. They use the grocery section as a footfall driver more than a profit center. Other specialized retailers pursue the profit maximization route. For pure-play grocery chains such as Jewel-Osco, it is always both, and it is a holistic model.”
Shalabh, the one-named director of U.S Midwest Operations for Princeton, Ν.J.-based LatentView Analytics Corp., lists two key factors that complicate and drive a need for robust demand forecasting: the emergence of online operations and digital ecosystems that have demand shapes that are the exact opposite of their traditional counterparts, and customer- and member-based models that generate more stable demands in large proportions.
“In grocery specifically, we have observed long-tail patterns in shelf life and seasonality causing variation and issues in achieving goals,” he notes. “The sophisticated systems that attack the problem statistically, stochastically and in hybrid are becoming the industry trend. The big chains also use stochastic demand forecasts and promotions for demand tailoring and shaping to objectives. Experimental analytics such as test campaigns in these areas are becoming a permanent arm to these demand-shaping initiatives.”
One grocery executive who has mastered these complexities is Abby Fox, procurement strategy manager at Commerce, Calif-based Unified Grocers. Progressive Grocer named her one of 2014’s Top Women in Grocery for her outstanding achievements in inventory replenishment buying.
She was recognized for leading a significant inventory reduction initiative, as well as for establishing a KPI-driven replenishment program designed to maximize service and profitability. Fox led Unified Grocers to an aggressive inventory reduction of $11 million while increasing customer service levels. United teamed with Marietta, Ga.-based Blue Ridge, a cloud supply-chain planning provider offering demand forecasting, planning and replenishment solutions.
The business context for demand forecasting should go beyond inventory replenishment to include planning and collaborative elements. Category managers, buyers, store managers and vendors all should be able to reconcile their forecasts using the same system, according to SAS’ Lipson.
“The technology should be flexible enough to provide insight to the forecast and the ability to manage the impact of changes made by any stakeholder in the system,” he explains. “Ideally, the forecasting technology would be linked to inventory optimization technology so the retailer and the manufacturer can work together to make sure consumers get their favorite holiday menus on the table without any hiccups.”
Teradata’s Simmons stresses that retailers don’t need to “boil the ocean” to get started. Instead, a small team and good detailed data acquisition practices, combined with some industry-leading tools, should be all that are necessary. “They can deliver impressive benefits,” he sums up, “in terms of reduced stock-outs, improved sell-through, inventory turns and customer service levels.”
“While almost all retailers in grocery use or have a forecast, it is not typically regarded as accurate or used with confidence, particularly in the prediction of promotion demand.”
—Tim JW Simmons, Teradata
“In grocery specifically, we have observed long-tail patterns in shelf life and seasonality causing variation and issues in achieving goals.”
—Shalabh, LatentView Analytics Corp.