Every day, we make consumer choices based on the weather: Hot coffee or iced? Walk or take the bus? Go see a movie or watch Netflix at home? At a moment when advertisers use data in all sorts of ways to better target consumers, it’s only natural that they’d start paying attention to the weather. Vikram Somaya is the general manager of WeatherFX, the advertising arm of the Weather Company, responsible for crunching and analyzing huge amounts of data. Here, he talks about why weather—and how we react to it—is so valuable in today’s economy.
Alexis Madrigal: What exactly is WeatherFX trying to do? Why does the Weather Channel want to go into selling data?
Vikram Somaya: We have always sold data. We recently rebranded the Weather Channel as the Weather Company, and part of the reason is that the channel is just one piece of the company. We have consumer brands, like the Weather Channel and Weather.com; we have apps; we own a company called the Weather Underground, which serious weather junkies might know. It’s convinced about 25,000 weather junkies to buy weather stations [outdoor weather-measurement systems] on their own coin and then send the data to us every 1.5 seconds. In return, they get to tap into our system. Because these people get off on weather in a variety of ways, that return on investment is good for them.
We are also embedded with all the major airlines. We provide data feeds to GM and Toyota, and to energy farms that might be interested in knowing how much wind they’re going to get this year, and to derivatives traders. WeatherFX uses the information that we have, and correlates that with marketer data.
AM: At its simplest, how does selling weather data actually work?
VS: Retail is an easy example. We have a retailer who may have a couple hundred or even a couple thousand stores across the U.S. We take data from each of those locations for each of their products, then we look at the information over time. We are looking to see what products start jumping off shelves when the dew point is X, the temperature is Y, and the rainfall is Z. What we give them is essentially: “Here are the 15 products you should be selling right now.”
We also overlay certain basic parameters on the weather. If it’s Friday, mothers are looking for what they’re going to be doing on the weekend. If they are looking to stock groceries for the weekend, and we know that the weekend is going to be gorgeous, how do we make it clear to them that they should be buying items for a barbecue?
AM: How much does weather really matter to companies?
VS: One of the fascinating things about weather is that it is so fundamentally local, and relevant to people in a very localized context. It’s hard to find another data set that does that—usually we run into some kind of data scarcity, where you only know whether one in every 10,000 people is interested in this product or that service. We have to make sure that we have the data-gathering apparatus, and not just today’s data—ideally we have historical data too, and then we have to build in the fact that the climate is changing as a whole.
As a company, we are beginning to take a much stronger approach to climate change, which you’ll see—we are the only ones who covered President Obama’s climate speech in its entirety. More-extreme weather is interesting to businesses, to insurance companies.
AM: Can you walk me through an example of how weather influences consumer behavior?
VS: There is a correlation for bug spray that’s kind of bizarre. We found that a very small difference in dew point made a huge difference in bug-spray orders. When the dew point changed, insects popped up, and everybody ran for the bug spray. Now that we’re working with bigger and bigger data sets—say, across a couple hundred thousand products, or a couple thousand retail locations—we are beginning to find a lot of situations like this.
Retailers have a fundamental knowledge of weather because of their supply chains. They know that it is impacting trucks, and how you get stuff from production locations to retail locations. But they hadn’t really taken that to marketing. For someone who came from a background where I sold more-esoteric data products, it has been a very humanizing experience to sell weather, because it’s something that everybody has a story around—it makes for a very personal tale.
AM: So tell me about some of the more business-to-business efforts. Do you find things like, for some reason, corporate-insurance sales go up if it’s rainy for two weeks?
VS: We have a division, WSI, that does nothing but build applications for insurance companies. There’s one that interacts with a professional hail-zone data feed: essentially, 30 minutes before a hailstorm hits an area that an insurance company covers, an SMS text is sent to everyone in that area saying, “Pull your cars in because you are about to be hit by hail.” In a medium-size hailstorm, that saves the company anywhere between $1 million and $5 million. Insurance companies are willing to spend a ton on understanding how weather can impact their business.
AM: What have the most-popular applications of these data been so far?
VS: We did a recent campaign with Pantene on our mobile app. The background of our mobile app is essentially an ad template. Depending on the weather in the city you were looking at, Pantene would suggest a shampoo for you. And we saw localized sales being driven in these areas.
AM: Throughout the tech world, people are worried about the accumulation of data, and about how various data sources are being used to track people. Do you worry about provoking that kind of backlash? Or are weather data not likely to be subject to such scrutiny?
VS: The great thing about weather is that there are no feelings of privacy violation in telling people that weather conditions are going to be X, Y, or Z. The data are not something we’re taking from them. The data are allowing people to be safer, or be smarter, or dress better, or eat better.
AM: How do you see our relationship with weather changing in the future?
VS: Weather has always been fundamental to how we live; we want to make it more transparent in terms of what impact it actually has. Fifty years ago, a hurricane was like an act of God. We have come a long way. Where will we be in another 50 or 100 years?
We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.
