- The tech giant has announced the first use of the legendary supercomputer for dynamically generating the most effective ads from separate creative elements.
- IBM says this exceeds traditional dynamic creative optimization (DCO) in the number of elements juggled and in the quality of intelligent predictions.
- The Accelerator can be used for any ads and any ad platform, not just Watson Ads.
In mid-2016, IBM launched Watson Ads. Powered by its Jeopardy-winning supercomputer, the ads conducted intelligent interactions or conversations with users via free-form text or voice, and the responses helped Watson tailor the ads for the particular user.
With Watson Ads, a consumer can type in ingredients on hand, for instance, and the Watson-powered ad can then show a recipe that employs, say, sandwich sauce and ad sponsor Manwich as well as some or all of those ingredients.
Now, the tech giant is adding another capability to Watson’s growing resume, with the announcement of Advertising Accelerator with Watson. This is the first use of Watson’s AI for ad creative optimization.
Any ad or advertising platform
It utilizes Watson’s massive intelligence toward delivering the most effective combination of creative elements for a given audience segment, such as male site visitors on the West Coast who drive a truck.
Currently, IBM Watson Advertising Head of Revenue Jeremy Hlavacek told ClickZ, the Accelerator controls five categories of elements in any kind of static or interactive ads: background colors, calls-to-action, key images, key text phases and audience targeting.
On the drawing board: adding the ability for the Advertising Accelerator to also control interactive variables, such as the responses to user inputs found in intelligent Watson Ads.
Although it is powered by Watson, the Accelerator can be connected to any demand-side platform (DSP) for any ad or ad platform. The advertiser will define the ad’s variables, such as how many key images.
Of course, dynamic creative optimization (DCO) platforms have been available for years to optimize creative elements based on the most effective results for specific combos for specific audience segments.
But, Hlavacek said, the new Accelerator differs from traditional DCO in several ways.
Compared to DCO
DCO typically is based around recombining a handful of creative elements, he said, while the Accelerator can be much more granular, juggling dozens of variables in the categories.
Second, he said, it utilizes Watson’s legendary ability to synthesize intelligence from vast reams of data, in this case including the effectiveness of certain combos for specific users, locations, weather, seasons and other conditions.
DCO is generally rule-based, he said, not driven by an elastic intelligence like Watson’s.
Additionally, once Watson’s dynamic control of conversational elements in intelligent ads is in place, the Accelerator could become the ad equivalent of an expert salesperson writ large – someone who knows exactly what to say, when, for a specific kind of audience.
While the Accelerator is now being tested with beta clients like Lending Tree, Hlavacek said the only publicly available result is a 300 percent increase in installation of the app from IBM-owed Weather.com after its ads were Accelerator-ized.