Supply Chain

Using weather data to predict demand is becoming more sophisticated

A growing number of retailers are using advanced AI applications to improve how they use weather data.
article cover

AdrianHancu/Getty Images

4 min read

Economists are looking at everything from the cost of living and inflation to interest rates and stock market performance to predict holiday spending this year, but missing from recent forecasts is a variable that many retailers have been anticipating for months: the weather.

On average, 3.4% of retail sales are directly impacted by the weather, according to a National Retail Federation analysis extrapolating data from the American Meteorological Society. This totals about $1 trillion of global retail sales per year, so the financial stakes are high for retailers trying to properly allocate products in line with weather-driven demand.

But until recently, retailers’ use of weather data wasn’t very precise, Don Coash, senior account executive and meteorologist at Accuweather, told Retail Brew. Many would simply look at what happened during the previous year and base their purchasing decisions off those sales trends. For example, if winter came early last year and a retailer didn’t have enough snow shovels to meet demand, this year it will make sure to buy more snow shovels earlier.

“The problem with weather is it’s almost never the same year over year,” he said. “So they’re always chasing their tail and pulling it forward this year because of what happened last year.”

Accuweather provides weather data and forecasting services to half of the Fortune 500 companies and multiple retailers, including Burlington and Lowe’s, and Coash said many of its clients are now using advanced AI and machine learning algorithms to measure more precisely how weather affects metrics such as foot traffic and sales.

Returning to the case of the missing snow shovels, he added, weather-based decision-making is evolving from “let’s send more snow shovels into that area” to a “more prescriptive” approach in which retailers figure out the exact number of shovels needed at a given location “based on the last five years of these types of weather events affecting that same area.”

Looking back to look ahead: So retailers are still looking at historical data, but now with more sophisticated technology designed to correlate a wide range of weather conditions, from temperature to raininess to iciness, with specific business metrics.

Scott Bernhardt, president of Planalytics, said its platform combines weather data with historical sales data to help companies come up with the exact number of units needed for a given product at a given location.

Retail news that keeps industry pros in the know

Retail Brew delivers the latest retail industry news and insights surrounding marketing, DTC, and e-commerce to keep leaders and decision-makers up to date.

He said the platform calculates a “derivative of the weather to show how much more or less of a product or service you’re going to sell in Philadelphia on Friday, September 20 when it’s 76 degrees and partly cloudy.”

The platform also tracks how demand for different products changes in response to the weather. Speaking from Philadelphia, Bernhardt cited the example of traffic at a King of Prussia store going up 3% but sales of boots going down because of the heat. “So you can get scenarios like that, which is why you do it across the different product categories,” he said.

Purchasing to positioning: This information is mainly used to inform purchasing decisions around volume, he said. The rest is up to merchants.

But once the purchasing is done, weather data can inform the next stage of its distribution, which is placement. Once shipments arrive in warehouses, Coash said, retailers “want to position the inventory in the right place. So that’s the next question. They’re like, ‘Okay, I know how much to buy. I bought this much for the company. Now I need some help on positioning.”

If a retailer is aware that snowfall is likely to lower in one region, for example, it can move that product to a snowier area that is likely to see more demand.

Timing is crucial to this endeavor. Coash said Accuweather is generally talking to retailers six months out to help them “get on the right side of normal.” In other words, it’s telling retailers if weather conditions are going to be broadly above or below normal. It then provides more granular forecasts closer to the season in question.

Indeed, some uncertainty is built into the enterprise, especially in light of rapidly changing weather conditions due to climate change. That’s why Bernhardt cautions retailers against working with anyone who overpromises on their predictions.

“If anyone tells you that they can forecast that it’s going to snow on Christmas Day in New York City on September 20…run, don’t walk, away from those people,” he said.

Retail news that keeps industry pros in the know

Retail Brew delivers the latest retail industry news and insights surrounding marketing, DTC, and e-commerce to keep leaders and decision-makers up to date.

R
B