Accenture Interactive, Accenture’s agency arm, is pitching publishers and brands on a product that uses AI to automatically overlay product placements on streaming video content.
The way it works is, for example, if an online streaming service serves a video ad featuring a moving plane to two people, one in the U.S. and the other in France, the former will see the plane with, say, the Delta logo on it, while the latter will see the same plane with the Air France logo instead.
Alex Naressi, managing director of Accenture Interactive’s research and development unit, said that the software uses machine learning — a specific field of artificial intelligence that looks for patterns in large data sets to identify patterns and make decisions — to learn from and simulate human vision. In this way, he said, it can run simulated eye-tracking to create attention heat maps for any video automatically, then decide what the viewer is paying attention to and place a product in the corresponding area in real time.
Naressi said the platform also combines automatic semantic analysis with human insights and moderation to ensure brand safety and contextual relevance.
Accenture Interactive launched the ad format at Cannes Lions but hasn’t tested it with any brands or publishers or decided how it would charge clients for it, as patents are pending.
Naressi said he envisioned the video solution as an “interruption-free native inventory” that’s immune to ad blocking.
Brandon Purcell, senior analyst for Forrester Research, said he hasn’t seen the solution firsthand, but that based on Digiday’s description of it, he thinks it looks like a better implementation of AI than the chatbots that are most organizations’ first foray into AI.
“This solution is incredibly interesting to me, as it dynamically changes the content for an individual not just based on gender or geolocation but also on behavioral factors,” he said. “This approach can be applied to other marketing areas like email and even dynamic display advertising in-store. Lots of AI algorithms today are open-source, so the true differentiator is the data that an organization uses to train the machine. If Accenture has its own human eye movement data, it will be very powerful.”