Walmart taps inventory management system to predict demand ahead of holidays

Every retailer is arguably part fortune teller, and every merchandising plan is a bet that future demand will justify a particular product mix or inventory level.

Indeed, major retailers often make bold predictions about the course of the US economy: “It’s going to be a big holiday season,” Walmart US president and CEO John Furner told investors during an earnings call in August.

But where does such a prediction come from? Does Walmart have a crystal ball hidden somewhere in its Bentonville, Arkansas, HQ or is this just a best guess?

The answer is obviously the latter, but the company’s economic predictions aren’t coming out of nowhere. Walmart uses an AI-powered inventory management system to predict demand, and this year it’s testing out a new and improved version designed to help meet demand without building up excess inventory like it did in 2021.

“The inventory management system is going to be key, and we are super excited to use it this upcoming holiday season and serve our customers in a much better and improved way,” Parvez Musani, SVP of Walmart Global Tech, told Retail Brew.

History buff: In 2019, the company started developing the software, which is known internally as the Walmart Inventory Management System, and initially it relied on historical sales data to come up with predictions that could inform an inventory plan.

Historical data doesn’t help prepare for major one-off disruptions such as the Covid-19 pandemic, but it can help account for patterns in the data that may repeat themselves.

Musani explained that a snowstorm in Arizona, for example, is something you probably want the data to forget, because it’s an anomaly. But a snowstorm in New York is far more likely and something to account for in forecasting.

Finger in the wind: Now Musani and his team are working on incorporating more predictive inputs into the system, such as weather reports, macroeconomic and industry data, and other metrics such as advertisement click throughs and website traffic.

“Now we are even looking at how people are searching,” he said. “Those types of things will influence [our models] as we get closer to the holiday season, and we will react to them and position inventory accordingly in our stores.”

In addition, broader economic indicators such as inflation and how it might impact buying patterns are also being factored into the model.

Data is as data does: But no matter how sophisticated the data, it’s not worth anything if Walmart doesn’t use it to inform its inventory placement. Musani said Walmart’s vast store footprint and logistics network allows the company to execute in response to data.

“After a point, you have to stop planning,” he said. “You cannot keep planning till the end. Then you get into execution, but we need nimbleness.” This means moving products in response to shifting consumer demand.

He provided the example of a particular toy selling better on the east coast than it is on the west coast, in which case Walmart could divert supply from one region to another.

Musani wrote in a blog post in 2021 that Walmart’s supply chain and inventory plans are based on a 52-week outlook. Going forward, he explained, the goal is to create a system that can adjust in real-time and seamlessly incorporate new data points.

He said the goal is to make it easier to “constantly tune” the data and still churn out “strong recommendations” to the rest of the company.