Content-Based Data Analysis and Streaming Ad Optimization
How are studios leveraging scene-by-scene content analysis to improve ad specificity and targeting, and how effective are current implementations? Byron Saltysiak, Former VP, WarnerMedia, Turner Broadcasting discusses this topic with Pete Scott, Chief Strategy Advisor, Play Anywhere, Jesse Redniss, CEO & Co-Founder, Qonsent, and James Lauzun, VP of Product, MagellanTV, in this clip from Streaming Media East 2023.
Saltysiak begins by mentioning that classifying content by simple genre specifications does not fully capture all types of action and, therefore, may not be the best means to target advertising. “Let's say you have a movie, it's a comedy, well, a comedy is going to have a car chase, it's going to have a sad part, it's going to have obviously funny parts and love parts. So simple genre isn't going to represent this two-hour piece of content very well…is anybody really pushing the needle forward on that? Like, ‘I know exactly what's on this scene, and I'm targeting the advertisement to that?’”
“Didn’t we do that at one time?” Scott says. “Where we were meta targeting all the shows? And then some psychologists that we would roll out in front of the upfronts would basically say, ‘I know that if you show a pair of shoes at the end of the break, someone will buy it. Payless will basically buy that inventory. So that was the correlation, right? The last thing you saw was going to be the first thing you saw in the commercial break.”
Redniss says, “There's some cool companies out there. There's a company called Fabric…the guy that started the company was deep into the content metadata tagging structure across all of Warner Media. And you're ingesting everything from TMS and Gracenote signals. Now there are small companies out there like Vionlabs that are doing actual video-based ingestion and doing scene-by-scene analysis of color, frame rate, music, the actors, and then taking the actual color wave and the speed of the video to start categorizing more content metadata on top of that so that you can make recommendations on the fact that, ‘Oh, Jesse loves Tom Cruise-faced action thrillers.’ And then you can actually drill directly in and deep link to those specific moments within those movies or shows. And obviously you can identify ad breaks and identify the mood…to say, ‘This advertiser needs to be aligned with this ad break coming out of this type of an experience within this show.’ That is really exciting. It's also really scary. But it's really exciting in regard to really understanding behavioral science of people's attention to the video itself.”
James Lauzun highlights the services that MagellanTV uses for its own deep analytics and emphasizes that this kind of detailed data is becoming more accessible to smaller platforms. “There are now startups creating this technology at a more affordable price point,” he says. “We're working with a startup called Lucid Insights that's doing a similar service. They're analyzing all of our content scene-by-scene. From that, we're creating dynamic video previews that are self-reinforcing and self-learning. And we're working on an audio-only deal right now. And sure enough, we run that back through the knowledge graph and have it identify titles that would be well suited for an audio-only experience versus titles that really rely on visuals…something that would've obviously been pretty tedious to have someone actually do, but now I can have the team focus on vetting those recommendations rather than trying to find them in the first place.”
Pete Scott talks about how this type of deep targeted analysis also helps with promotions for the content itself. He says that in his previous experiences, only one promo would be made for a particular piece of content. This was ineffective, he says, “Because you have to use different scenes and different shots for different types of people [who] may want a particular character they want to see first.” He mentions the ways that enhanced data can improve “controllable costs.” “More and more,” he says, “rights or content will be expensive to create. So the controllable costs that we manage in this room are going to be reliant on data to create more cost-effective shows or more cost-effective targeting and personalization. And I think it's going to get there just because everything else becomes so expensive and the risk is too high.”
Redniss argues that risk remains high when it comes to leveraging data if it is not leveraged correctly. “When you don't have it governed in the right way, you end up in a position like Facebook Meta Pixel and the NFL and 40 other media companies are now getting sued for Video Privacy Protection Act (VPPA) non-compliance because they were tracking every [viewing] on nfl.com, and then you went back to Facebook, and they're instantly making recommendations on you without your consent,” he says. “So it's understanding the data and the consumer experience to say, ‘Oh, great, well we picked up [the] IP address [for] IP delivered streaming video. That's actually protected in a way that you can't just use it unless the consumer actually knows and understands how that information is getting used…you can't just run rampant with that. You could do really cool innovation, but then all of a sudden, you're in an issue where you're handing out 10, 15, 25 million dollar fines. They just announced a Facebook fine that's going to exceed the $780 million fine for their data transfers from the EU to the US. That's an expensive risk. So those two things need to go together, where you can actually balance it out, do right by the consumer, and innovate at the same time.”
Learn more about a wide range of streaming industry topics at Streaming Media Connect 2023.
Personalized data is critical to video content monetization, whether in a subscription-based environment or an ad-driven FAST channel. Coming at the topic from both angles, Starz' Rob Collins and Revry's LaShawn McGhee discuss their organizations' strategies for leveraging personalized data in this clip from their panel at Streaming Media Connect 2023.
While recent slumps in subscription-driven OTT have convinced many observers that great content and plenty of it simply aren't enough, what can content providers do to reverse that trend by following the numbers and understanding their audiences better? Hub Entertainment Research's Jon Giegengack has the numbers and some strong recommendations in this clip from his presentation at Streaming Media Connect 2023.
Quickplay's Naveen Narayan and Reality Software's Nadine Krefetz discuss how gathering the correct data can improve a streaming platform's understanding and knowledge of its audience, yielding enhanced user engagement and content monetization, in this talk from Streaming Media West 2022.
Revry's Damian Pelliccione and Legion M's Jeff Annison discuss how they gather and apply useful streaming analytics for ad-driven OTT and FAST in this clip from Streaming Media West 2022.
How does better data impact the way that users react to ad-supported streaming tiers? Will having the impression of more options for personalized ad preferences lead to more acceptance of ad-supported streaming? Three industry experts discuss these questions and the future of heavily data-supported advertising and its impact on both the streaming industry and the web as a whole.