AI, Advertising’s Next Big Step In Personalizing Branded Videos





  • by @lauriesullivan, October 5, 2016



    Artificial intelligence software remains the foundation of most services and products that advertisers will see this year.



    While search became the first to demonstrate the power of AI and machine learning, it now shares the glory with video and other media. Google CEO Sundar Pichai said during a recent product launch event that Knowledge Graph now holds 70 billion facts.


    On Wednesday, IRIS.TV released technology that runs on AI and serves in real-time branded video campaigns. It analyzes the publisher’s video archive and the behavior of the individual viewing the content.


    The goal is to create a continuous stream of videos that have the highest probability of being watched by that person.


    Rohan Castelino, IRIS.TV director of business development and marketing, says Campaign Manager creates additional pre-roll inventory downstream. Viewers of video players powered by IRIS.TV can watch up to eight videos per viewing session. AI will change the way brands create content.


    Consider findings from data company Beakon, which estimates just 5% of branded content garners 90% of total consumer engagements. The other 95% shares the remaining 10% of engagements. In other words, 19 out of 20 content pieces get little to no engagement, per Beakon’s findings.


    The average number of content — images and videos — that brands created and posted in the past 12 months, in both free social media channels and paid media channels, jumped three times compared the prior 12-month period, according to Beakon.


    AI is also the backbone for IRIS.TV’s platform, which creates custom video streams served one-on-one by pulling, creating and analyzing a variety of data in real-time. Titles, descriptions, keywords, categories are the main data points, but if the publisher using the platform doesn’t tag the video with metadata, IRIS.TV’s platform uses natural language processing.


    Without user login data, it tracks viewer behavior through device, browser, location, time of day, when they click on “skip the ad,” and on what type of content they typically apply a thumps up or down. Combining these analyses, the AI makes a decision of what to play next.


    Finally, the cloud-based system processes billions of data points daily to serve videos in real-time. To build a consumption history on users, when they go to a Web site for the first time and click play on a video, the system drops a session cookie that creates an anonymous user ID.


    As long as the user returns within a 30-day period, the cookie extends.



     


    MediaPost.com: Search Marketing Daily

    (5)