Things to consider when forecasting online demand

For the past 4 months, as we witness the emergence of new buying habits, the online grocery retail industry has experienced growth like never before. The average number of daily online orders has more than doubled, while the user base of e-grocery retailers has grown threefold for the past two months. To catch up, brick-and-mortar grocers that traditionally generate the majority of revenues offline are aggressively entering online space. Changing the business model brings a lot of challenges, and serving online demand that requires a vastly different forecasting approach is one of them. In this article, we'll share common issues grocers need to bear in mind when serving online demand in order to succeed.

Customers behave differently in real life and online. Shopping for groceries is no exception. Parameters influencing customer demand in offline and online also vary. For example, competitors' prices are a more critical parameter influencing the purchasing decision in the online sector. Opposed to offline shopping when visiting a store takes time and effort, online customers can easily switch between several competing stores, compare prices, and chose the best option.

Nothing is more disappointing on a Sunday morning than being left without your breakfast favorites because an online store couldn't fulfill your order and delivered only a part of it. Unfortunately, such out-of-stock situations are common in the online space. Trouble is, incoming orders are not synchronized with the number of goods readily available for delivery from a store or a warehouse. While orders are placed in real-time, availability data is usually updated once a day.

With the traditional offline format, errors in forecast of store traffic usually cost retailers a few man-hours of labor of extra cashier or several more minutes of the checkout line. Online is different though. The last mile of delivery - from grocery stores or warehouses to customers' doors - is expensive and can comprise up to 80% of the total transportation costs in the retail supply chain. Customers want their orders delivered quickly - in the span of 15-30 minutes -, anytime they want. It means retailers need not only to forecast demand but optimize courier availability.

For many retailers, the quickest way to go online is to launch a web store that offers delivery from their brick-and-mortar stores. However, it is crucial to bear in mind that store layouts were designed to enhance customer experience and spend more time inside the shop. The more time customers spend in the store, the higher the average purchase sum.

The less time employees spend searching for goods in a store and preparing them for delivery, the faster customers get their orders. For example, industry leaders like Samokat, the fastest-growing online grocery store in Russia, spend as little as two minutes to prepare orders for delivery.

From layout to promotion stands to sampling, there are plenty of tried-and-true methods to make goods stand out on a supermarket shelf. In online space, however, grocers have nothing but the size of a smartphone screen to catch customers' attention.

Online shopping has uplifted the competition in the industry. With thousands of prices available for customers to compare 24/7 and hassle-free online shopping experience, it's easy for customers to optimize their behavior and choose best offers switching between competing web stores. To discourage cherry-picking behavior, grocers need to enhance customer experience in the online space.

All said, going online poses a lot of challenges to grocery retailers. However, the biggest one may be synchronizing offline and online operations.

While building warehouses and expanding delivery fleet require extensive capital and time investment, retailers can adopt business processes for online now. Demand forecasting, safety stock optimization, personalized offers, and customization of online shopping experience are among the challenges that can be solved with machine learning technologies. Change is never easy but available technologies and modern forecasting tools are here to help.

To learn more about the potential benefit of machine learning for grocery retailers, read the article Three high-impact areas and easy wins in ML for grocery retailers