Demand forecasting for promotions: AI is the best way to manage the complexity

In the tough world of promotions - three strikes and you’re out.

Loyalty is hard to gain and easy to lose. If customers can’t avail themselves of a given promotion in-store, they switch stores after the third letdown. It’s harsh, but true, and with FMCG retailers struggling to accurately forecast demand, it’s almost impossible for them to get it right for promotions, consistently. Unfortunately, customers don’t care about how hard it is for a retailer to forecast for promotions. One, two, three strikes. They’re gone.

Overstock to solve out-of-stock

To deal with the challenge of demand forecasting for promotions (See related paper on Demand Forecasting and AI), retailers have typically overstocked to prevent out-of-stock, which solves one problem – keeping the customer happy, but it creates two more – cost of holding stock and shrink, with shrink being a significant issue for fresh items.

The complexity of promotion mechanics

Promotions are complex and legacy systems are not designed to process data from the many inputs that need to be considered. With so many components to the promotion including in-store, marketing and price mechanics, the complexity arises when you try to pinpoint why a promotion behaved the way that it did while also considering if the store executed the promotion in the right way for the right duration?

Of course, there are other factors to be considered alongside these – the weather, local events, history, other promotions running in-store and even when a promotion isn’t linked to a holiday, in some regions it can coincide with local sporting events or festivals which may drive demand for the product even if it wasn’t part of a promotion.

Artificial Intelligence and forecasting for promotions

Processing the data from the many inputs needed to accurately forecast demand for promotions is beyond the capability of legacy systems, not just because of the volume of data, but because of the insights and intelligence needed to provide a view of future demand. It’s key to have a forecasting system that can easily integrate the multitude of variables that culminate in an effective, impactful promotion. AI-enabled demand forecasting has the power to process all data and increase the accuracy of the demand forecast for all stages of a promotion – the start, the middle and the end, driving data downstream through the supply chain and into store operations. This is critical because promotions aren’t just about marketing and merchandising, they cut across all operations right down to compliance on the shop floor. At every stage, the priority is to keep you, the retailer, at a distance from the dreaded three strikes.

For more information, read the Viewpoint: Improving Forecasting Accuracy in an Age of Retail Disruption

Join us at NRF 2019 in booth #3036 to learn more about how retailers are utilizing AI-enabled solutions to meet and overcome their forecasting challenges.