Gracenote content supercharger makes trending appear ancient

While Nielsen doesn’t generally receive much praise here at Faultline Online Reporter, we can at least appreciate it giving metadata expert Gracenote the freedom to innovate within the audience measurement behemoth since its takeover a couple of years ago. Hot off the press is Gracenote’s latest product launch called the Video Popularity Score – essentially a personalization turbo for video service providers with a few clever tricks up its sleeve.

Of course, this is worlds away from the automatic content recognition-pioneering Gracenote the world fell in love with and Nielsen is essentially using Gracenote as a front for its own push into video analytics to enhance its core data business – and an effective one at that.

Part of Gracenote’s wider Advanced Discovery suite of metadata products, the Video Popularity Score tool identifies and surfaces what it perceives to be “hot content” based on a number of data inputs including recency of TV show airings, viewership and awareness signals. A proprietary algorithm calculates consumption data covering TV, VoD and OTT from Nielsen Total Content Ratings and movie box office data from Gracenote, combined with some secretive third party data.

These metrics are used to calculate a numerical score which it says represents the general population’s recognition level of individual TV series or movies. But surely the entire premise of video ecosystems evolving into highly-personalized and efficiently-targeted beasts is about getting away from such generalizations? We’re sure you’ll agree, which is why Gracenote describing the general population’s viewing habits is a bit of red herring here – as the technology goes much deeper.

Gracenote goes on to explain that entertainment service providers can use its Video Popularity Score as an add-on tool for existing personalized recommendations and search results within the UI – implying it serves as a sort of supercharger for existing recommendation engines. But how does this differ from Trending segments now commonplace within most video UIs today?

“Many existing “Trending” displays are largely one-dimensional usually considering the service’s own viewership data, recency of content or social media chatter as indicators of popularity. But this can be limiting. Gracenote’s Video Popularity Score takes in multiple high-quality signals including consumption data across multiple sources and services, box office data and social data to calculate a weighted score for each piece of video content,” Gracenote’s VP of Product, Video, Kamran Lotfi, explained to Faultline Online Reporter.

He provided the example, “It’s possible that a given program can be “hot” across multiple services but still not rise to the level of “trending” on a specific service due to a delayed response from its viewers. Similarly, a title can be trending in the world of social media but not have traction yet on a single service. Both these scenarios could lead to a hot title not being presented prominently in a Trending display. By leveraging Video Popularity Score, which takes in multiple signals, the provider would have a more holistic understanding of what’s trending broadly across multiple services, platforms and catalogs – and have the ability to present that content to viewers accordingly.”

You would therefore expect Gracenote Video Popularity Score to be infinitely more accurate in terms of what content a consumer deems “hot” compared to your run of the mill “Trending” landing page. Yet such comparisons to rival systems were strangely absent from the announcement – missing a prime marketing opportunity to tell operators and pure play providers why their existing software is inferior.

“The supercharger analogy is spot on,” said Lotfi, referring to our earlier point, “to be clear, we are not looking to deliver consumer-facing UIs or features to replace those our customers are deploying. Rather, through our new Video Popularity Score and previously launched Video Descriptors offering, we aim to create the industry’s broadest, deepest, most descriptive entertainment metadata to help our customers deliver the ultimate user experiences.”

The secret sauce, however, which really differentiates it from your average Trending listing, is that Video Popularity Score takes into consideration the social content consumers are talking about watching, on top of the content consumers are broadly watching.

“Given the full range of signals it incorporates, and the proprietary weighting system used to process the signals, we believe it provides for the most accurate and timely measure of what consumers think is hot,” added Lotfi.

We have drummed up the value of plugging social media data into TV viewing to power personalized experiences on a number of occasions, and quite often the idea has elicited anxious faces – which is possibly why what we see as the Video Popularity Score’s pièce de résistance has not been given the emphasis it deserves.

Another key component is voice support, with Gracenote claiming improved natural language processing, giving the example of being able to differentiate between the broadcast TV show Chicago Fire and the Major League Soccer team of the same name when processing a voice command – although it doesn’t explain how.

This builds on Gracenote’s Video Descriptors product launched in January, which is all about bringing more contextually relevant and satisfying TV and movie discovery experiences. Again, this essentially serves as a booster for existing recommendation and personalization software – applying additional levels of granular metadata to content. Video Descriptors encompasses Mood, Theme, Scenario and Characters – featuring structured keyword sets for individual shows.

Another example cited by Lotfi helps bring the nitty gritty of Video Descriptors to life with Game of Thrones. “Gracenote pinpoints Themes such as “Greed” and “Betrayal,” describes Scenarios including “Power Struggle” and “Manipulation,” assigns Mood elements like “Dark” and “Gripping” and classifies the program’s Characters from “Royalty” to “Dragons.” Focusing on Scenarios, Gracenote data can be used to identify similar programs based on related descriptors such as “Sweet Revenge” or “Good vs Evil.” By understanding what GoT is about and relating attributes to other programs, Gracenote Video Descriptors enables nuanced discovery to surface content that will best resonate with individual viewers.”

Gracenote virtually came out of the blue and into a blaze of video analytics last year, although in fact this goes way back to a company called Telephia it acquired about 12 years ago – from which the fruits today are thriving with Nielsen’s backing.