There’s no denying the opportunities analytics has opened up for retailers over the past few years. In a business where information is power, retailers are racing to understand customers better every second. It began with historical data mining: what brands has Birmingham-based Mary, 35, mom of two stuck to buying over the past three years? Then the focus moved to real-time analytics: which items does Mary consider as she navigates the aisles my grocery store right now? So it’s only natural for retailers to be interested in the next step in the flow of analytics, and perhaps the most powerful one: how can I scientifically predict what Mary is going to buy the next time she visits my store?
So why is this important? Customers today are creating their own shopping experiences at their own pace, and often have more information that’s relevant to their interests than the in-store employees. Popular collaboration-enabling technologies, such as social media, social networking, and context-aware and location-based technologies contribute to the high service expectations that shoppers have and this in turn places immense pressure on the retail supply chain. To meet these growing expectations, retailers must effectively manage the intersection of demand, supply and product.
The real power of predictive analytics can be realized when retailers collaborate with suppliers and use this information to improve supply chain capabilities. According to Gartner, supply chain success and beyond hinges on retailers’ abilities to decipher insights gleaned from their shoppers, develop scalable growth strategies and acquire supply chain talent. Retailers who are then able to deal with the impacts of these challenges will increase customer satisfaction, supply chain productivity and revenue.
Predicting downstream customer behavior
Segmentation often occurs in the retail space when customer data is used to create clusters which organize customers or stores that have common (or unique) patterns or behaviors. For example, clustering can be used to identify groups of customers like high spend, medium spend and low spend customer segments. At the store level, it can also be used to identify product purchasing trends. This would then not only have an impact on how stores are furnished with merchandise, or how marketers speak to customers, but it can also influence product options at the supplier level and pricing.
In recent years, the Dollar stores began taking on Wal-Mart head on by stocking cheaper basic supplies like toilet paper and medicine. Their strategy was successful enough to get Wal-Mart to sit up and take notice – it even started to lose market share. So what did the retail giant do? It went back to its data for a solution. Wal-Mart’s predictive analytics solution revealed that many Wal-Mart customers start tightening their purse strings at the end of each month, but still needed a few basic items to tide them over till payday. Wal-Mart then began stocking shelves with thousands of items that were prices under $1 at the end of each month, so customers who the Dollar stores lured for such items went back to Wal-Mart. This is important information for suppliers and vendors who work with Wal-Mart as it also affects their supply schedules.
Demand sensing in the supplier domain
Kimberly-Clark is a $19.7 billion manufacturer that operates within four global businesses – Personal Care, Consumer Tissue, K-C Professional and Health Care. The company has produced some of the most recognized consumer brands in the world, including Huggies, Kleenex, Pull-Ups, Depend, Kotex and Scott. Like many consumer products companies that rely on high promotional activities, K-C was challenged with managing trade promotion dynamics. The problem occurred because the actual orders received were different from the forecast, and inventory was often in the wrong place to effectively manage promotional orders. K-C ultimately improved its short-term statistical forecast using predictive analytics – specifically demand sensing. Demand sensing uses a combination of transactional data (orders and shipments) as well as well as promotional data, point of sale, daily order and shipments to produce a daily statistical forecast. This forecast is released to K-C’s ERP planning system every night and presents K-C with short-term forecasts, ranging from zero to five weeks forward. In this scenario, demand sensing steers the demand-driven journey at K-C, improving inventory management, lowering supply chain costs and improving service to customers.
Demand planning at the store level
Just like segmentation helps retailers better understand the spending patterns, communication preferences, and merchandising preferences of their customers, the same principles can be used to segment stores, partners, and vendors. Retailers cannot afford to treat each store or channel as the same so by creating segments, the retailer has a better perspective on consumer behavior and can strive for greater loyalty by speaking appropriately to the defined segment.
Predictive analytics can be used at the store level to make demand planning more accurate. This has an immense impact on customer satisfaction, store operating costs, relationships with suppliers, and much more. Stores often have difficulty in predicting situations like a stock out(which occurs when a store is out of stock on a particular item) or over stocking (when a store has too much of a particular item suggesting that the retail store does not understand its customer base) . Both this scenarios are unpleasant for retailers because segmenting based solely on past behavior to steer a future marketing strategy doesn’t work — it’s almost like driving a car by only looking in the rearview mirror. In the stock out situation, retailers not only lose sales but can also face a potential decline in customer satisfaction. On the other hand, the over stocking situation is also costly because items sit on shelves and are not moving. More importantly, the retailer has paid the supplier for unnecessary merchandise. Predictive analytics at the store level can integrate demand planning into the merchandising, operational, and store processes to make the store operations more efficient. Since retailers cannot assume that all stores and all customers have the same behavior, using demand planning enables stores to run much more efficiently.
Predictive analytics: putting retailer-supplier collaboration on the fast-track
When it comes to data sharing between retailers and suppliers for collaboration, retailers must look beyond just sharing data to develop the right analytical capabilities and viewpoints to make high impact decisions. A retailer who wants to pursue true customer-centric retailing must consider the use of predictive analytics which not only reveal the behaviors of the past, but also address the likely behaviors in the future by incorporating new customer data sources. A large part of this model depends on retailers sharing data with their suppliers so that both can enable the appropriate supply chain response that is required to sense, shape and profitably respond to these new demand signals.
Have you begun thinking about using predictive capabilities in your analytics and collaborative decision making?