Supply Chain Leader - Issue 4

There's Power in POS Data - Not Just for Retailers, but for Suppliers Too
by Mohan Balachandran and Jim Morganstern

The old techniques for meeting customer and market demand—forecasting from historical data and holding inventory—are no longer effective for consumer goods manufacturers dealing with increasingly high customerservice expectations. Traditional forecasting processes not only tie up capital but are riddled with errors when applied to short-term demand and supply issues like shelf replenishment.

And the increasing retailer/supplier need for timely visibility into store-level demand drives the development of new technologies—systems that employ point-of-sale (POS) data and intelligent, exception-based scenarios for tuning retail supply networks to market demand.

Visibility is the issue du jour in this era of rapidly moving products, and the ability to understand shelf-level events as they take place is critical to rapid response. Historically, POS demand information has not been visible to consumer products manufacturers in real time. The data aspects of category management were typically contracted to third-party market-research firms like ACNielsen and IRI. These firms submitted reports including data on brand performance and issues such as out-of-stocks—usually two to four weeks after the fact, or longer, depending on the amount of replenishment inventory on hand.

POS data have played a part in category management since the 1980s, when Wal-Mart first put its Retail Link® supplier network in place. “Generally, POS data were somewhat dirty at first, but are a lot better now because retailers are using the data for their own replenishment,” says Larry Lapide, director of the recently launched Demand Management Solutions Group at the MIT Center for Transportation and Logistics. “But even cleaned up, the data are still not very usable as-is for consumer products manufacturers who need to harmonize that data with their own language and systems.”

The downstream challenge

The benefits of demand-driven, POS-based business intelligence do not stop at the shelf. Analysis of downstream demand data also yields long-term demand-sensing and demand-shaping strategies. These strategies enable suppliers to use consumer demand signals to quickly take action as new opportunities arise, improving category management and increasing brand equity as well as profits.

Leading consumer brands continue to be plagued with issues like stock-outs in one location while suffering excess inventory in another.While consumer products manufacturers have employed a variety of approaches to more proactively manage fast-moving retail channels, they are still falling far short. All errors—whether due to stockouts, inventory build-ups or other process breakdowns— are costly to retailers and suppliers alike.

Compounding consumer products manufacturers’ production and extended supply chain pressures, retailers’ new emphasis on inventory management at the shelf level adds a new level of customer service. Essentially, it is vendormanaged inventory (VMI)—though not at the distribution center, where the manufacturer’s retailer teams traditionally are required to manage inventory and replenishment. Instead, it’s at the store level.

Wal-Mart (among other retailers) monitors supplier performance on an ongoing basis through a weekly scorecard, grading suppliers on a wide range of metrics, from on-time delivery and in-stock levels to more recent initiatives, such as the environmental impact of the packaging used. Other big-box retailers, as well as grocery, consumer drug, home improvement and electronics centers, all contribute to increased velocity and complexity.

According to Lapide, “Using POS data to support VMI is another permutation of the retailer / consumer products manufacturer, co-managed inventories relationship at Wal-Mart and elsewhere. Historically, retailers and consumer products (CP) manufacturers have been peering into a black hole regarding day-to-day shelf performance. During the months-long lag times, any demand fluctuations— by day, location, context [weather anomalies, etc.]— are obscured by the bleeding down and building up of inventories, making it impossible to track potential shelf trends.

“Also, you tend to get a bullwhip affect in the distribution channel,” Lapide continues. “Small changes at the shelf level are exaggerated as they move upstream, so you have to look at your own shipments much longer to see through the noise.”

The pitfalls of shelf blindness are legion. “We just researched this problem with an apparel manufacturer of seasonal products,” Lapide adds. “They knew what they shipped—by size, color, style—to a West Coast distribution center, but they had not collected data on what sold and why at the store shelf. It turned out there was an issue with SKU sizes that weren’t appropriate for Asian women, and consequently the larger-size products only sold by markdowns. This is exactly the sort of disconnect that would be solved by POS data.”

Even with a 98 percent fill rate, the fine print can reveal costly fluctuations.Wal-Mart shelf fill rates are measured each Friday, revealing a weekly average across all store locations. Consequently, demand variability by day or by store is invisible. Most consumers shop on weekends, and if stock levels fall below 98.5 percent on Saturdays or Sundays, proportionally more sales are lost.

An analysis i2 performed shows that even with a 98 percent in-stock average over the weekend, 277 stores owned by one retailer would be out of stock of a particular SKU on Sunday. To take effective action, companies need daily analysis by SKU and by store, and also need to evaluate how important each store is in terms of SKU sales. To obtain the data manually, they would have to look at every SKU, at every store, every day.

The ultimate benefit

The beauty of new tools for handling POS data is that manufacturers now have the means to preemptively sense demand at the shelf level—shifting the whole fulfillment paradigm from what has happened to what is happening. “In redefining processes to better serve shoppers at the shelf, manufacturers must focus on demand sensing by key account,” according to Cecere, in her recent AMR Research Report “Shouldn’t You Be Minding the Store?” (May 2007).

Cecere cites recent manufacturer pilots that illustrate the sort of benefits achieved using downstream POS data and focused account teams. A major distributor changed pre-determined routes and fixed delivery frequencies to dynamic routing based on daily POS data. The three-month pilot resulted in an 8.2 percent increase in sales. In another pilot, a consumer electronics company took over shelf replenishment for a major electronics retailer, reducing 18 weeks of inventory by more than half and improving instock positions by 4.3 percent, with a reduction in markdowns due to overstocks.

Daily demand sensing is most critical to the success of new-product introductions and promotions. Cecere notes that 50 percent of new-product introductions fail because of poor execution. “Out-of-stocks double or sometimes even quadruple in heavily promoted categories at peak shopping,” she says. “The synchronization of getting product to shelves is absolutely critical.”

For Lapide, a big issue is early indicators. “When you first put the product out in the channel, it typically doesn’t sell immediately at the shelf. You need real-time information from the retailer to know the point at which it takes off in order to keep product in the pipeline. POS data become the leading real-time indicator for introductions as well as promotions, allowing retailers and others in the supply chain to react to changes daily.”

Finally, well-planned and coordinated shelf-level execution based on real-time downstream demand data is only the beginning in realizing the full potential of customer- centered merchandizing. POS data represent one stream of a number of possible cross-channel interactions with assortment, allocation, space, pricing and promotional data that, when integrated together, can bring unprecedented granularity, timeliness and responsiveness to categorymanagement decision making.