IBM’s Weather Company employs Watson to boost its updated ad targeting platform
Watson interprets the company's massive first-party data from its app users, plus powers intelligent Watson Ads, to better understand behavioral patterns.
The Weather Company has taken another step toward predicting much more than just weather.
Now owned by IBM, the company has unveiled an updated and rebranded ad targeting platform called Journeyfx. It uses the analytical skills of IBM’s Jeopardy-winning Watson supercomputer to better predict which users make the best targets for specific kinds of ads, and when is the best time to reach them.
And there’s now another major layer of data: how users react to Watson Ads, just launched this week by the Weather Company.
The company had been using location and weather for basic geo-targeting of its ads for the inventory in its own properties, Weather.com on the web and the Weather Channel mobile app. But now, backed by Watson, it is also learning more sophisticated behavioral patterns.
Users agree to send their GPS location data via the Weather Channel app, so that they can automatically receive weather forecasts without having to enter a ZIP code. Watson analyzes location/weather patterns for each user, who is tracked via an anonymized mobile ID.
If you’ve been to five dealerships in the last month, for instance, the Journeyfx platform infers that you are in the market for a car and thus might be receptive to car ads. The platform utilizes machine learning to recognize what kinds of hidden patterns indicate certain kinds of user interest.
Watson is helping to infer not only which customers should be included in specific segments, Weather Company Global Head of Sales Jeremy Steinberg told me, but also which moments are the “optimal conditions,” taking into account weather, individual patterns of movement and patterns by other users in the targeted segment.
Although the results currently target ads only for the Weather Company’s inventory, he said the data may become available at some point for other brands’ inventory.
Additionally, The Weather Company is launching this week its first Watson Ads, initially for Campbell’s Soups during cold weather days. Another campaign using the intelligently responsive ads, first announced in June, will be unveiled with Toyota in first quarter of next year.
The ads are designed to offer intelligent conversations with the brand via natural language conveyed through voice or text, with the result being highly personalized brand-related info.
For the Campbell’s ad, for instance, the user is invited to “name any dish, ingredient, or occasion.” When I suggested spinach and tomatoes, Watson came back with a Tomato Lasagna he claimed to have invented, using Campbell’s Condensed Cream of Mushroom Soup. Here’s a screen shot of the Web demo:
Watson has reportedly been trained to understand how food flavors work, although, of course, he never has to suffer through the results.
Steinberg noted that users’ ad responses — both individual ones and ones from segments of users — add another layer of data which is incorporated into the ad targeting.
Given the Weather Company’s large user base, Watson has a lot to work with. There have been two hundred million downloads of the app, and, on a monthly basis, there are about 87 million different, regular users.
On a daily basis, there are more than 20 million users of the app, of whom 80 percent have shared their location. On average, they check the weather in the app between two and five times a week. The company says this constitutes “the world’s largest continuous sets of first-party data.”