When Spotify was still fairly new, the assumption was that people would want to listen to their own playlists and maybe share them with their friends. But developers of the app soon realized that people were quickly growing tired of their own music and were looking for someone or something else to decide for them. Hence the rapid growth and immense popularity of the Daily Mix selections, personalized radio stations based on your past listening history and personal interests.
IRIS.TV, a five-year-old Los Angeles-based company is doing something very similar for video. The company already has an impressive track record, using its AI-based personalization platform to promote viewership of ad-supported videos for publishers including Fox, Gannett, CBS and Time Inc.
“The idea behind what we do is to help publishers generate more advertising revenue by reclaiming their audiences on their owned-and-operated sites and apps, getting them off social media, and understanding their consumption behavior and preferences to better engage them in the future,” CEO Field Garthwaite told me during a recent phone call. “Our system is a brain and if we can improve the user’s experience by predicting the sorts of videos they’ll want to watch, then they will spend more time consuming on the publisher’s site and they’ll come back for more.”
IRIS.TV does this, Garthwaite explains, by anonymously creating first-party video consumption data on a publishers audience, captured via an API and plugin that the company has developed for every video player and device on the market. Using a series of AI-based algorithms, IRIS.TV creates consumption profiles on consumers and extrapolates viewing patterns from the individual’s own behavior and those of other viewers with similar interests. The result is a continuous learning system that understands which videos have the highest statistical likelihood to be completed. The approach is strikingly similar to the targeting capabilities utilized by companies like Facebook and Netflix to engage and understand their audiences.
The most recent showcase for IRIS.TV’s technology was the 2018 Winter Olympics where they helped Gannett’s USA Today target the right videos to the right viewers.
So doesn’t that just mean that if I like skiing, I’ll see a lot of skiing videos, I ask Garthwaite. “Not necessarily,” he tells me. “Say you’ve been watching videos about how different skiers train for their event. Our system may then begin to understand that you like ‘Skiing’ but also ‘DIY’, videos about how Olympians train in general. We make it easy to explore more of the publishers archive without having to search. Similarly, if we see an IP address from Detroit and there’s an athlete who is from the Detroit area, we might show you videos about her. The idea is to sprinkle in some serendipity-based logic, so that your playlists continue to surprise and delight you.”
If not surprised, Gannett was certainly delighted by the results IRIS.TV achieved, boosting video views and engagement by over 50% during the Winter Olympics as discussed in the companies’ presentation at Streaming Media East.
“What we learned is that context matters,” Garthwaite says. “With our insight into the context of every video in a client’s library and their audiences’ real-time consumption habits, we enable publishers to target specific types of video that are popular at certain times of day and to specific audience cohorts. The publisher also gets unique data insights that enables them to customize the machine’s decisioning when necessary. Combining these factors, we create video streams that are extremely relevant to the viewer. During the Olympics and the NBA Finals, we were also able to bring in breaking stories as well. So that if someone just won a medal and the video was trending, we could ensure that it also wound up in the feed of viewers who might be interested in hearing about it. It provides that sense that your feed is constantly being curated and that new stories are being added in near real time. That sort of functionality really boosts engagement because it makes viewers feel like the publisher understands them.”
The IRIS-ization of Television?
So the Spotify reference at the start of the article wasn’t random. In my book, Over The Top: How The Internet Is (Slowly But Surely) Changing The Television Industry, I talk about how the shift to a largely on-demand, library-based interface is going to lead to the “Spotifyization” of television—a system where viewers have the ability to watch the same show over and over again and/or create their own playlists, but are more likely to turn to curated playlists, from networks or others, for the same reason Spotify users turn to their Daily Mixes—your own tastes grow boring after a while and there’s a strong desire to let someone else take the wheel.
Given that TV is the ultimate lean back experience, there’s an even stronger urge to remain passive … so long as what you’re being served remains relevant, interesting and above all, surprising.
IRIS.TV, with its AI-driven algorithms seems like the perfect vehicle for this new universe. By taking into account a wide variety of factors, including prior viewing preferences, what’s trending among audiences, their location and time of day, it can help broadcasters and publishers keep audiences engaged, creating the perfect experience for viewers so they keep watching and keep coming back.
From there, it’s not hard to see these same algorithms applied to advertising, so that viewers are served up ads that meet their interests, along with some that may also surprise them.
This latter point is far more critical than many in the industry may realize: one of the things we stand to lose in this new hypertargeted ecosystem is serendipity, the ability to discover something you might not have known you wanted.
That’s why it’s important for brands to get in front of consumers who are outside of their projected target. By spotting connections that go beyond the expected, IRIS.TV may have the power to restore some of that serendipity to both the programming and the advertising we see.
“Our end goal is to help publishers build profitable digital businesses by providing them with the tools to make smarter programming decisions that keep their audiences engaged,” Garthwaite tells me. “What we’re really doing is taking viewer’s natural inclinations and turbo charging them.”
More power to him. (Pun intended.)
Originally published at Forbes.com on June 21, 2018