July 15, 2020
Navigating unpredictable changes in customer shopping behaviors
An accurate demand prediction is one of the most critical elements of a resilient supply chain. When catastrophic events occur that can cause a massive shift in consumer behavior, retailers of all shapes and sizes are suddenly scrambling to adjust their forecasts. With the pandemic potentially forcing new or extended “shelter in place” orders, people will continue turning to new methods of receiving groceries, clothing, and other consumer products.
As we see different parts of the country and globe beginning to reopen, it’s possible many of these shopping behaviors may stick as people opt for less dining out due to new grocery shopping trends: pantry loading, working remotely, willingness to try private label, and a shift to new fulfillment channels.
As inventory levels drastically fluctuate, many shoppers have settled for what they (or their personalized shopper) can find, ultimately throwing brand loyalty out the window for cheaper brands, private labels, or store brands. Consumers have opted for new buying channels as well, ordering online and having groceries delivered. If the ease of these channels stick, it’s possible to see a long-term shift.
Unfortunately, the footwear and apparel industry has experienced the complete opposite. With many retailers being closed for quite some time, demand vanished, resulting in stagnant inventory. Priority apparel purchases have plummeted, and most shopping has been online, not in stores.
With many consumers now working remotely, they’ve shifted to a more comfortable wardrobe that also supports increased outdoor activity. While formal and business attire remain significantly down, purchases in active wear have increased. Footwear has also seen a major decline in sales except for performance shoes, which have seen a strong increase in demand.
But will these new perspectives on life and focus on personal health last? Once everything fully reopens, will demand shift back to previous behaviors or will remote working opportunities and active lifestyles remain permanent trends? The same question applies for home improvement and home goods stores. With people adopting new activities such as decorating, gardening, repairs, and remodels, home projects have emerged as a huge focus of time and money.
Addressing the demand challenges
The challenges with demand prediction involve new shopping behaviors, unanticipated store closures, and the uncertainty around developing trends. Innovative AI-based software vendors partner with retailers by using a team of data scientists to tune machine learning models for their unique business categories. By using a “pandemic” demand feature, the scientist can train the machine learning (ML) algorithm to automatically account for newly witnessed selling behaviors—eliminating the irrelevant time-periods (panic buying, store closures, etc.) for models to accurately predict future needs.
In traditional models, demand planners must manually update time-series parameters, aggregate levels, assign like items, manually update forecast for promotional lift, and more. But machine learning extends beyond recent impacts to other demand features such as new product forecasting, which uses product and location attribute trends, promotional details, halo/cannibalization, external data like weather, social, local events, and more.
Machine learning models provide the automation and flexible demand modelling required for accurate prediction during and after these uncertain times. Next generation artificial intelligence in retail will utilize neural networks, deep learning, and other modelling techniques to further drive automation and accuracy.
Having a team of data scientists behind the scenes ensures models are tuned to produce accurate results is a far superior approach to training retail demand planners on models, parameters and hoping they get it right. Demand planners should focus their time on managing exceptions, analyzing the business, looking for opportunities, and during these uncertain times, strategizing what the retail looks like moving forward.
For additional strategies for building the resilient supply chain needed for the Retail Revolution, download the complete Supply Chain Resiliency Best Practices Guide.