Big Blue says Watson is saving it a lot of green.
For almost a year now the company has been trying out Watson, its cognitive computer that learns and makes snap decisions on the fly, on making programmatic ad buys for a portion of its online ad campaigns.
The results proved so effective that IBM now says it will use Watson for all of its programmatic campaigns by the end of the year, a big commitment for a company that spent nearly $53 million on digital display advertising in 2014, the most recent figures available, according to Kantar.
"Because of the volume and the dollars involved, trying to save those fractions of a dollar, or fractions of a cent, really matters to us," said Ari Sheinkin, VP of marketing analytics at IBM. "What makes this really exciting is the system learns. That's the essence of cognitive."
The company says Watson reduced its cost per click on average by 35%. At its best, using Watson reduced the cost per click for IBM by 71% when compared to its previous buying methods.
At the same time, IBM is working on making the technology available for agencies and exchanges so they, too, can eventually maximize ROI through what IBM calls "cognitive bid optimization."
But the trick, Mr. Sheinkin says, is to figure out what Watson did when it was at its most effective, so it can repeat those results across the board in the future.
"It's almost as if the system we've created right now is like a child prodigy who is already beating you at chess," Mr. Sheinkin said. "But it's actually only starting to learn. It's only had so many campaigns to learn from, so it's going to get better and better because every time it runs it learns from the last thing it did. And that is the most important, distinguishing feature."
A machine described as a child prodigy that has only begun to learn might sound terrifying to some, but IBM sees things in a different light, one where machines like Watson coexist with data experts in harmony. "My experience is first of all, data is a tool," Mr. Sheinkin said. "There is an art and a science to paid media and the way we are rolling this out is very much as a tool for the expert."
"And because we are becoming more effective in our ad spend, we are investing more in the organization around the technology, not just on the technology," he added.
So what sets Watson apart from the other technologies used to buy programmatic ads?
For starters, Watson is an advanced machine learning computer that can improve programmatic paid media by reducing costs per click. It uses advanced analytics to create efficiencies in the bidding process by ingesting massive amounts of data and assigning value to potential target consumers based on the time of day, what device they are using, what language they speak and what browser they are using.
Watson becomes even more mind boggling when it begins to factor in the size of ads, and whether a smaller size is more effective when shown at 3 a.m. for a $2 CPM, or cost per thousand impressions, than larger ads at noon at a $3 CPM, for example.
Eventually, Watson will be able to access the 80% of data that can't be read by most computers -- also known by experts as "dark data." Dark data includes making sense of what consumers mean when they say something in casual language or share an image, video or audio on social media.
"A human or even a great expert is never going to be able to ingest a velocity of volume of information in the moment and then react to it," Mr. Sheinkin said. "Watson can."
About $27 billion will be spent on programmatic display ads in 2017, a 24% increase from the previous year, according to a report released by eMarketer in June.