Metadata, Hyperpersonalization, and Leveraging Longtail CTV Content
As streaming content providers increasingly realize not only the value of their back catalogs, but also how to optimize them through improved discovery, metadata and metadata standardization have become critical, as Gracenote’s Ashok Bania explains in conversation with Dataxis’ Ophélie Boucaud. Hub Entertainment Research’s Jon Giegengack chimes in on the role of scene-level analysis in maximizing the value of deep back catalogs and longtail legacy IP in this clip from Streaming Media Connect 2026.
Where the Industry Is Going
Ophélie Boucaud begins the conversation by saying, “One of the big revolutions of this era is really the natural language aspect of things. So it becomes way more … complex the way that [viewers] can interact with content. It’s not just using old taxonomies and very rigid categories.” She asks Ashok Bania, “as the metadata provider in the room,” how he sees Gracenote evolving and preparing for the future. “And also, is there any direction that the industry is taking that you need to look at?”
Bania shares a few things he’s been noticing. One, content owners are more willing than ever to use “AI to merchandise their old back catalog. So long-tail content is getting very important. Second is they’re also realizing that … the same piece of content can mean [one thing] for one person and something else for another person.” He describes Gracenote’s efforts to collate data and metadata and to standardize imagery. He’s seeing success with back catalogs being merchandized using video descriptors, “a service layer on top of the data.” An example “is personalized imagery. This is becoming a much more talked-about topic right now among content owners, where the same movie—something probably you all have also observed in Netflix as well—the same movie will have a different poster for a different query,” Bania notes, adding that AI can really help with hyperpersonalization, making content more interesting.
The New Trend of Clipping
“As [fellow panelist] Alok [Ranjan] mentioned, we also are looking at this trend of clipping. Clipping is a new industry term, which is nothing but basically getting aspects of certain content and highlighting those aspects,” Bania explains. “And what we are seeing is … there’s no standardization on clips. We have TMS IDs for movies, for episodes, for seasons, but we don’t have one for clips. So we are investing in that area.” This involves scene-level, moment-level analysis and surfacing that data appropriately to Gracenote’s content owners, publishers, and partners. That’s where Bania sees AI having the biggest impact.
Not Brand-New, but New-to-Me
Boucaud turns to Jon Giegengack. “Jon, do you have any takes on the clipping and the scene-level analysis?”
Giegengack replies, “We find that the most common way that people, especially young people, now discover a new show is from a clip, and usually it’s on TikTok or YouTube or somewhere like that before they actually go into a TV app. But I think that long tail of content, all of that IP that for many of these studios goes back a hundred years, I think is a hugely underestimated and definitely unfully monetized asset. Consumers today, and again, more especially young ones, they don’t care if a show is brand new, they just care, is it new to me? And sometimes the catalog ones are more attractive because they have lots and lots more episodes stacked up.”
AI Can Find the Diamonds in the Rough
He adds that Netflix is probably the only studio to be able to spend billions on new content, but the other studios have, in some cases, 100 years’ worth of IP that people can discover. “And I think that what AI can do—if it can do this, which I think it can—is figure out where are the diamonds in the rough?” he notes, pointing to the recent popularity of Gunsmoke, a 65-year-old show.
He wonders what made it so popular all of sudden, concluding, “There’s something about shows like that or The Last Ship or Suits that when they get onto a platform with enough scale, they start to spread, and they spread on their own. And I think at the moment, the IP owners don’t really know. I mean, if they knew which ones those were, they wouldn’t sell those ones to Netflix. So we know that they don’t know which are going to be the most successful ones. But something that would enable you to figure out where are those diamonds in the rough for any particular users or for an audience I think would be a massive game changer, especially for the legacy media companies that we work with.”
Join us November 9–11, 2026 for more thought leadership, actionable insights, and lively debate at Streaming Media Connect 2026! Registration is open!
Related Articles
Over time, LLMs will enable transformational TV experiences for viewers. Consumer expectations for how they interact with technology continue to rise, and CTV platforms have an opportunity to get ahead of the curve. LLMs can deliver on that promise-but only when anchored in trusted, up-to-date data. With TV viewers constantly reevaluating where they spend their time and money, the cost of unvetted or unreliable information becomes impossible to ignore.
03 Mar 2026
Reimagining ingest: Cut costs without cutting corners by treating ingest as a strategic, metadata-driven platform.
17 Nov 2025
The future of streaming is about unification built on top of a single open standard of content metadata. Right now, it's a matter of seeing the forest for the trees.
04 Oct 2022