The Weather Company is sitting on a trove of data. Acquired by IBM in late 2015, Weather has been in a prime position to take advantage of IBM’s data, artificial intelligence (AI), and machine learning initiatives.
How so? IBM’s Watson supercomputer recently created ads for Toyota. And while the copy was all over the place at first, Watson’s AI-based system improved over time and eventually offered insight to Saatchi LA, Toyota’s creative agency. Weather last year teamed up with Watson on ad initiatives with Campbell's and Unilever, helping to create ad units that consumers could interact with. For example, a consumer could ask the Campbell's ad: “What should I make for dinner tonight?” Watson delivered AI-generated responses.
When it comes to programmatic or automated media, take the case of bid optimization. Weather is looking at moving Watson into bidding algorithms for IBM on the buy side. Demand-side platforms (DSPs) look at millions of ad calls to see which ones fit certain campaigns best. The DSP looks at all the available ad impressions in real time and decides which ones will be most successful for the ad campaign.
“Watson APIs are helping with that process in terms of decision-making. As it relates to advertising, Watson is constantly evaluating the choices and optimizing them to make a better decision next time,” said Jeremy Hlavacek, VP of global automated monetization at The Weather Company.
Hlavacek’s group sells inventory in multiple exchanges. He said there’s potential to answer even more questions and inform processes using Watson. “As a publisher, we have a very large pool of programmatic inventory. How do we know that we’re setting our price floors at an optimal rate? How could we apply Watson to the sell side of our business when selling ad impressions?”
Hlavacek emphasized that Watson and AI tools can help media buyers make better decisions by looking at inventory pricing and optimizing demand sources. Watson could also potentially offer pricing recommendations. Going even further, he said Watson could help Weather buy media.
Hlavacek said Weather is transitioning from being a publisher to a platform. “For us, as a publisher, our revenue is determined by how many impressions we have and what price we can get for them.” But Weather’s data could be valuable for other publishers and exchanges—across social campaigns, Facebook campaigns and search campaigns.
In fact, Weather’s partnership with LiveRamp in April enables it to distribute data to 500 ad-tech partners. The goal, Hlavacek said, is for Weather to take all the data it’s kept in-house and move it out into the “ecosystem” into out-of-home, search, social, and even TV.
The key is to use Watson to pull all the datasets together and make a super segment. Learning how all the data segments work together, then ingesting, and processing the data, fascinates Hlavacek: “When we add Watson and AI on top of the data, we look more like an AI company. Having AI to process all of the data is different.”