Omnichannel Grocery Requires Rethinking Demand Forecasting
Consumers desire to shop more ways than ever. As a result, omnichannel is on the rise.
More than half of U.S. grocers now offer omnichannel services, according to Progressive Grocer’s 84th Annual Report of the Grocery Industry, published this past April. But only 12.2 percent of grocers have a “fully integrated strategy” for omnichannel retail, which is necessary for today’s shoppers.
“Consumers are migrating to new selling channels as fast as digital players can make them, and it’s a challenge for traditional grocers to keep up and build out their capabilities quickly enough,” says Brian Baker, managing director in the products practice at Accenture, a Chicago-based technology services company.
Times are changing, though. Some 34.1 percent of grocers said that they have a strategy they’re executing, and slightly more (36.6 percent) said that they’re just getting started.
One of the most daunting parts of starting and executing an omnichannel strategy, however, is maintaining proper supply, no matter what the channel. Grocers previously had only their brick-and-mortar stores to take into account, but now many more channels must be considered.
Where Demand Lies
Johanna Småros, co-founder and CMO of RELEX Solutions, a Helsinki, Finland-based retail-planning company with U.S. headquarters in Atlanta, notes how demand forecasting in grocery, especially fresh food, is tricky due to the need to examine at a granular level — store-SKU-day — sometimes even within the day.
“This level of detail is needed to accurately match available stock with demand that fluctuates daily throughout the week to ensure good product availability and minimize costly waste,” she says, noting that shelf life needs to be added to the equation for the freshest products, such as meat and fish, and the impact of promotions and price changes must also be factored in.
She continues: “Omnichannel accentuates these challenges, as the consequences of poor product availability are more immediately visible to the online consumer in the form of the ordered products having to be replaced with substitutions. Online consumers are also highly sensitive to ‘best-before’ dates and freshness of products delivered.”
Before grocers can start understanding demand forecasting in an omnichannel environment, they first must tackle the greatest challenge: understanding the new, omnichannel consumer.
“In order to build the seamless omnichannel experiences that today’s consumers demand, retailers and FMCG marketers must first understand how and why shoppers engage across all channels and platforms,” says Sneha Uppal, VP of advanced analytics with Schaumburg, Ill.-based Nielsen. “Without this core understanding, a retailer will never gain a clear view into who their true competitive set may be. In a world where a one-size-fits-all approach will not work, this is a must-solve issue.”
Understand Categories, Inventory
When forecasting demand for omnichannel grocery, retailers must first determine which categories have disproportionately high online share. For instance, nonfoods naturally perform far better online than foods do, according to Dr. Kurt Jetta, CEO and founder of Shelton, Conn.-based TABS Analytics. However, certain foods, such as protein powders, are a major exception, getting almost 20 percent of their sales online.
Additionally, real-time inventory visibility at the store level is mandatory for van delivery and click-and-collect, offers Danny Silverman, SVP of customer success at Boston-based Clavis Insight. Without it, substitutions can result in an underwhelming shopper experience.
“The biggest challenge we’ve seen relates to time of day for restocking shelves,” he says. “If this is primarily a morning activity, by the afternoon high demand and fresh items may be low or out of stock, increasing the likelihood of disappointing impulse-driven shopping trips.”
Moreover, substitutions must have a clear and consistent strategy that follows the shopper decision tree for a given category, Silverman adds. In some categories, substituting with the same brand but a different size will be better received by shoppers than the same size but a different brand.
Another area critical for proper omnichannel demand forecasting lies in linking online sales to the right fulfillment channel. For instance, online orders picked in a regular store must be included in the demand forecast driving replenishment to that store, even though actual sales transactions belong to the online channel. However, simply lumping regular store sales together with online orders to form the basis for forecasting doesn’t suffice, RELEX’s Småros cautions.
“On many occasions, online orders follow a different sales pattern compared to regular store sales,” she explains. “We have, for example, noticed that easy price comparison online often drives even more pickup for promotions. Also, for major holidays, the timing of purchases may differ between physical stores and online. … Separate forecasting of online sales is needed to accurately account for the sales channels’ varying demand patterns.”
Dealing With Data
It’s true that grocers have to deal with mountains of data due to the vast number of SKUs they sell and their large number of daily transactions. However, forecasting in the grocery channel actually is often hampered by a lack of data.
For instance, estimating the effect of price changes can be surprisingly complicated, thanks to a limited amount of relevant price changes in the past, Småros explains.
“In brick-and-mortar retail, it is quite onerous to test different approaches to, for example, pricing, promotions and assortments, as it requires a lot of manual work such as updating shelf labels, producing promotional material and rebuilding shelf displays,” she says. “This means that the business is more static than one would think. Electronic shelf labels make changes easier, but online stores still offer a much better opportunity to run lean experiments on a limited scale.”
Systematic testing allows omnichannel retailers to better understand, for example, consumer response to assortment width or price sensitivity in different product categories and for specific items, Småros says. This kind of testing and analysis enables continuous optimization of the online offering, but is also highly likely to provide insights that can be used in brick-and-mortar stores.
In the end, demand forecasting in an omnichannel environment isn’t just about product stock — it’s also about workforce. In traditional brick-and-mortar retail, store personnel forms 14 percent of sales, the largest operational cost, according to Småros. In this case, shoppers do a lot of work picking their products. Online, retailers need to perform order picking, which is labor-intensive, whether in regular stores, distribution centers or dark stores. Multichannel grocers need to forecast picking volumes and their timing accurately to fulfill lead-time promises made to customers, without excess labor cost.
For instance, if online orders are picked in a regular store, a grocer needs three separate forecasts to optimize rosters: the in-store sales forecast taken down to the 15-minutes level, which is the driver for checkout work and customer service; the forecast of daily incoming deliveries, which is the driver for in-store goods and handling work; and online order lines to be fulfilled within a certain timeframe, which is the driver for order-picking work.
“Through smart optimization of the timing of work tasks, retailers can, for example, move personnel from the checkouts to order picking when business is slower in the store,” Småros asserts. “Planning for this in advance means that excellent service is possible while keeping costs in check.”