But data from Weather Trends International that called for a cold snap this September in the Midwest is paying off handsomely for the $3.5 billion, 278-department store chain that includes such brands as Carson Pirie Scott and Elder-Beerman. Without the data, Mr. Carroll would have planned based on last year and "we would have been dead wrong because it was excruciatingly warm and there wasn't a natural buying frenzy because people were walking around in shorts."
While some marketers and retailers still depend on a Farmer's Almanac to integrate the variable of weather into everything from merchandising, advertising to replenishment planning, others overlay sophisticated and increasingly digestible weather data that predict weather patterns up to 12-months out. When combined with historical sales data, the forecast models can determine, say, the impact on soup sales across time when there's 3 inches of snow in Buffalo compared to 3 inches of snow in Atlanta.
"It's about building business decisions objectively and looking at how the weather is driving consumer behavior," said Paul Walsh, senior VP-client services and senior business meteorologist at Planalytics, whose clients include the Gap, Home Depot and JC Penney. Wall Street retail analysts also use it to better predict retailer results.
At Casual Male, Planalytics data have begun to affect everything from how many sweaters to ship to each of its 510 stores to the timing of direct-mail campaigns to the 6 million customers in its database.
"For men, it's 'I need a coat now; it's cold,"' said Dennis Hernreich, chief financial officer and chief operating officer of the Canton, Mass., retailer with $450 million in sales. "We get a lot of spikes associated with weather."
Now merchandising teams can increase or decrease sweater shipments based on forecasts. "It's helping us get our allocations a bit more precise," Mr. Hernreich said. "It's not a perfect system yet, but it beats the way we used to do it."
For Jonathan Swiskow, director-marketing at Tractor Supply Co., weather data from Planalytics used over the last five years have influenced when his company brings out riding mowers and makes markdowns, along with when it buys newspaper inserts.
But integrating the data into the planning process at the 664-store chain is no easy chore. "Knowing what the weather is going to be like only helps you if you are set up to take action on that information," he said, adding: "Their forecasts are not always perfect; there are times when they say it's going to be cold and it's not as cold as they said or warm and not as warm as they said."
The Nashville, Tenn., retailer with locations in 34 states and $2.1 billion in sales even tried to plan its radio buys based on weather forecasts but stopped when it "became very complicated and was complete data overload."
Forecasting is crucial for Chris Caron, VP-brand development for fire-log maker Duraflame, because nearly 80% of its more than $100 million in sales happen over just four months.
By integrating weather-trend data into planning in recent years, the Stockton, Calif.-based company has been able to better plan shipments, preventing markdowns and retailer returns. "We've seen dramatic results," Mr. Caron said.
The normal kickoff to the season is Halloween, but if data show it's going to be a mild October, Mr. Caron said he might hold off on advertising, figuring demand will peak later in the season.
Sharing weather data with retailers also helps Mr. Caron get more shelf space for Duraflame, which has seen sales jump 20% over base volume on weather factors alone.
Last year on the West Coast, where colder weather was expected to last well into spring, Mr. Caron persuaded some retailers to continue stocking fire logs instead of charcoal for barbecues no one would have in chilly temperatures. It ended up the biggest spring season in the history of the company.