Weather forecasters get a bad rap for getting things wrong, but what if your insurance company had a forecast so accurate that it could send you an alert to move your car before a hailstorm destroyed your windshield?

That's the promise IBM sees for the Internet of Things -- the tech industry's term for embedding Internet sensors into appliances, cars and other everyday objects -- and its potential for collecting weather data.

The company announced Tuesday that it's taking a $3 billion bet on the Internet of Things over the next four years, starting a new business unit based just on helping companies collect and make sense of massive sets of data. And to kick off the new unit, the company also announced that it will partner with the Weather Co., the owner of the Weather Channel, to show off how it can turn data into insights.

The Internet of Things has been a focus for IBM for quite a few years. Last year, the company announced that it would work with the mobile processing firm ARM to build a "starter kit" of sorts to let hobbyists make their own connected devices. It was also a founding member of the Industrial Internet Consortium -- along with AT&T, Cisco, General Electric and Intel -- which is focused on encouraging the development of Internet of Things projects for businesses and government.

The new partnership should have applications for both the public and private sectors, said Mark Gildersleeve, president of the Weather Co.'s global business to business unit, known as the WSI Corporation. Gildersleeve said his company uses data from more than 10,000 sensors across the globe to generate 10 billion forecasts per day.

Some of those sensors, he said, are placed by government entities such as the National Oceanic and Atmospheric Administration and the National Weather Service, while other data are collected by sensors embedded in vehicles, on buildings or through user-submitted information from smartphones. (That information is only collected from people who opt-in to Weather Channel programs, and is turned into anonymous data.)

The goal of all that collection is to generate useful information for businesses and governments which, in turn, can help those entities make better decisions about when to cancel flights, postpone school days or respond to bad weather.

"It's not just a weather forecast," Gildersleeve said. "It becomes a tool to say how many staff a retailer should schedule for based on the weather, or for telling an insurer how many policy holders were in the line of a storm when it hit -- so they can quickly mobilize their customer response unit to handle claims."

Glenn Finch, IBM's lead for Big Data & Analytics, said weather information can often fly in the face of business owners' intuition -- and reveal surprising results.

For example, Gildersleeve noted, it's no surprise that a rise in temperatures will prompt an increase in beer sales. But the Weather Co. has found that there's a stronger correlation between high temperatures and strong beer sales in Seattle than there is in Arizona.

Why? That's not for the data to say, but that information could help store owners better understand how to stock their stores to keep up with the whims of the weather -- and to keep consumers from facing empty shelves.

"As the weather becomes more volatile -- which we have seen -- this becomes more important," Finch said. "Where I live [in Fairfax, Va.], we've seen we can have seven inches of snow, and next day it can be 75 degrees. If you're a retailer now, what do you do besides hope?"