AI Makes Forecasting Easy

The forecast is pivotal to every function of the retail business, but 34 percent of supply chain leaders admit that their top challenge is a lack of forecast accuracy.

Patty Mc Donald

An accurate demand forecast is a perpetual moving target for grocery retailers, with the bullseye being successfully delivering goods and services to customers. The forecast is pivotal to every function of the retail business, but 34 percent of supply chain leaders admit that their top challenge is a lack of forecast accuracy.

When it comes to demand forecasting and replenishment for fresh categories, having an inaccurate forecast is costly, from many perspectives. Out-of-stocks mean losing potential sales and overstocks translate to waste on both a physical and financial level.

What if it was possible to forecast perishable products with the same degree of precision as you would with other longer-shelf-life items? It’s very possible to build your confidence in managing the forecasting of fresh items, as long as you’re intentional about it.

A growing demand for fresh impacts today’s grocery store

The fastest-growing area of grocery retail today is fresh foods. Nielsen research shows that across the FMCG brick-and-mortar landscape, fresh categories in the U.S. drive 49 percent of all dollar growth. Maybe more surprising is that in 2018, fresh and perishable foods generated sales nearly 14 times as high as that of all online food and beverage sales. Retailers ought to see this trend as an opportunity to up their in-store fresh game, differentiating themselves through new options made available to shoppers. Ignoring the opportunity entirely means falling behind in the competitive grocery space.  

While demand for fresh grows, so does consumers’ prioritization of quality products, and both contribute to the changing store format. The store of the future – what we call Supermarket 2020 – will have fewer aisles and a more farmers’-market feel to the produce department. Research shows that 76 percent of consumers indicate they’re increasingly buying more prepared foods than cooking dinner, so curating fresh items that are ready-made for convenience-seeking shoppers is another opportunity for grocery retailers.  

Retailers can stay competitive through their fresh and ultra-fresh assortment, but the ability to meet consumer demand through that assortment is only as good as the originating forecast. 

Imminent threats to forecast accuracy

Your forecast is influenced by all of the activity that happens around a given product or category. Unpredictable weather causing supply chain interruptions. The ticking clock of expiration dates and perishability. Produce shortages and food safety recalls. All of these dynamics play a part in impacting a retailer’s fresh demand forecast. Retailers need visibility and full track-and-trace capabilities to know exactly where inventory is, and the technology to afford them the flexibility to make smart daily and intra-daily forecasting and replenishment decisions based on weather and price fluctuations. Furthermore, how do you forecast the impact of holidays? Are you replenishing goods that have the potential to go bad before they sell? Where can you make changes to shorten the supply chain, getting fresh and ultra-fresh items to into a shopper’s basket with greater efficiency? Do your systems enable you to manage multiple vendors per fresh item? The factors that influence a demand forecast for fresh can be tremendously cumbersome, especially if you don’t have an established system to manage it and react quickly.

Today, most retailers have one tool in place for forecasting the rest of the store, and something separate and manual to forecast fresh. Fresh requires different forecasting strategies, which is why retailers need sophisticated, grocery-specific logic to do so. It’s important for a holistic system to be in place to successfully forecast all categories, understanding the nuances of the grocery landscape and also adjusting the forecast in response to complex consumer behavior patterns.

Forecasting more accurately through AI and machine learning

If your demand forecasting system has artificial intelligence (AI) and machine learning powering it, you as a retailer will be a force to be reckoned with. These technologies can enable you to make more accurate predictions, more quickly and at scale, while bringing automation to retail operations.

A good forecast accounts for all types of seasonality – AI and machine learning work in harmony to recognize sales patterns and anomalies, understanding consumer buying behavior and incorporating all events that may have an impact on the forecast. Information gleaned from data guides the system, allowing for a more hands-off experience for users. AI can recognize that a brand-new item will perform like a similar one: for example, a new mushroom from a different suppler. No one has to worry about going in and adjusting the forecast accordingly – the system is one step ahead, saving valuable time for the user.

If your fresh forecast is accurate, you’ll find that surrounding business functions come with greater ease too. You’ll no longer be overcompensating because of poor inventory visibility. Food waste will be reduced, and you’ll alleviate inconsistent inventory buys, overstocks (and resulting markdowns), out-of-stocks, and margin erosion. By utilizing AI, forecasting corn and other fresh items will come just as easily as the more-predictable boxes of cornflakes, and at the end of the day, you’ll be better able to compete, too.