The freshly redesigned eighth floor of the Weather Channel's headquarters feels like some sort of breach in the time-space continuum. This is a 30-year-old TV network in Atlanta, yet the atmosphere is pure Silicon Valley: an open floor plan with no cubicles in sight; employees in their twenties huddled around whiteboards, gripping green markers; even a well-used Foosball table. The young vibe "helps with recruiting and morale," says Cameron Clayton, president of the company's digital division. "The goal is to be more collaborative, like a startup."
TWC's office overhaul comes at a crucial time for the Weather Channel brand, which is itself in the middle of a reboot. According to Nielsen, the network averaged a five-year low of 211,000 daily viewers in 2013, which is down from 273,000 the previous year. Partly to blame was the year's unusual dearth of catastrophic U.S. weather events, which typically translate into major ratings. But it isn't just a lack of hurricanes: TWC is facing a major migration to mobile devices, which are fast replacing TV as the primary source of weather information. Over the past year, according to analytics firm Distimo, the total number of weather apps for iPhone and Android doubled to nearly 10,000.
Thanks largely to its brand recognition, TWC's mobile app is by far the category's most popular, boasting around 100 million users. But an eager crop of startups, including apps such as Dark Sky, SkyMotion, and WeatherSphere, are beginning to eat away at TWC's dominance. They're also changing users' expectations about what a weather forecast should be. Nowcasting—highly accurate short-term weather prediction for a specific location—is finding favor among weather watchers who've grown weary of semi-reliable five-day forecasts. While TWC's app can present conditions in 15-minute intervals, Dark Sky, for example, predicts to the minute, often with startling precision, when it's going to start raining or snowing within the next hour. SkyMotion does the same for the upcoming two hours. And WeatherSphere's product, RadarCast, can navigate drivers around oncoming storms.
Now TWC finds itself looking over its shoulder at these growing competitors. "We do pay attention," says Clayton. "The Weather Channel is perceived as the 800-pound gorilla in the [room]. We've allowed some of those apps to occur. We haven't defended our space as well as I would like."
Adam Grossman started Dark Sky for a simple reason: "I just got tired of getting wet." The 32-year-old web developer, who has a physics degree and no background in meteorology, was sure there had to be a better way to predict imminent weather. Companies like TWC and AccuWeather have typically relied on methods that can delay forecasts, such as having meteorologists interpret data for longer-term predictions. But Grossman envisioned a method that made instant use of available information in order to create something more immediately useful than the seven-day, or even hourly, forecast. He thought he could tap National Weather Service radar, satellites, personal weather stations, and his own statistical "secret sauce" (a program that virtually erases distortions such as birds and fog from weather radar) to improve short-term forecasting. After raising $40,000 on Kickstarter, Grossman and a partner, Jack Turner (who has since left the company), debuted Dark Sky in early 2012. The $3.99 app has since been downloaded about 400,000 times.
Large companies are starting to notice the success of the Dark Sky model. Last fall, agriculture giant Monsanto spent $1 billion to acquire the Climate Corp., which provides hyperlocal weather data to farmers. That was followed by AccuWeather's November purchase of the Dark Sky–like SkyMotion. "Most of the good weather apps aren't made by the big guys anymore," says Grossman.
At one point, I ask David Kenny, CEO of TWC parent company the Weather Company, what he thinks about the new crop of nowcasting apps. His response is decidedly chilly. "That's radar data," he says dismissively. "We've done that for years. But they've found a better way to present it for people to understand it." When I tell Grossman about Kenny's reaction to his company, he's just thrilled that the head of Big Weather's largest player even knows who he is. "Oh, yeah?!" he exclaims. "Well, that's kind of flattering. If they don't crush us."
To stanch viewer migration, TWC revamped its TV channel in November with major changes to its appearance and its coverage (in December, it even hired weatherman Sam Champion away from Good Morning America). The channel now sports a sleek, almost app-like graphic display on the bottom third of the screen that shows local hourly weather conditions 24/7, even during commercials. The national network, which currently generates a reported $350 million in annual income, is also rolling out technology that allows it to televise location-specific information to a defined geographical area. Powering these changes are hefty upgrades to its network of content-delivery servers, which involved having its 220 meteorologists slice up the country into 4,000 different zones. No longer will St. Louis residents be forced to sit through coverage of thunderstorms in Orlando. "It's capable of doing so much more than just putting up a local forecast every 10 minutes," says David Clark, president of the network.
Though TWC, which is jointly owned by NBCUniversal and investment firms Blackstone and Bain Capital, has long had the resources and infrastructure to deliver powerful and relatively accurate forecasts, it must now channel that expertise into expediency and timeliness in order to maintain its top position. While visiting the company's headquarters, I get an early look at some mobile projects that are designed to combat upstart forecasters such as Dark Sky. Radius, for example, is a (very) early-stage app designed to let weather watchers share user-generated photos with other people in their area.
But it's not until I sit down with TWC's principal scientist, Bruce Rose, that I get a peek at something with truly transformative potential. Forecast on Demand is a new technology that incorporates elements of nowcasting but is able to create a detailed forecast—at the request of a user—for more than 2 billion points around the globe. Rose demonstrates on a raw-looking website featuring a world map. As he clicks around various random locations, Forecast on Demand instantly generates real-time forecasts for that specific geographical point, using the freshest information available from its more than 75,000 data sources. This upends traditional forecasting, which relies on pregenerated predictions. The plan is for Forecast on Demand to go live during TV broadcasts in March. It should also be integrated into the mobile app shortly thereafter, finally allowing an up-to-the-minute forecasting feature. If it works as intended, the technology could represent a paradigm shift for prediction techniques.
But TWC's rising competitors are fast thinkers, too. Last year, Dark Sky created an API service that makes short- and long-term forecasts available to other up-and-coming companies, edging further into the territory of TWC, whose data fuel apps from the likes of Yahoo. And Grossman is in talks with car companies to install Dark Sky technology in vehicles, which could dramatically expand its user base.
Will TWC's efforts be enough? It does have one big advantage: its current market dominance. "My hope," says Clayton, "is that in six months every VC says [about proposed weather apps], 'It's a great idea, but it's just not worth going up against the Weather Channel.'" Whether or not that will happen, however, is tough to forecast.
A version of this article appeared in the February 2014 issue of Fast Company magazine.