Ahead of broadcast industry trade show NAB next week, vendor Quickplay has joined with Google to use generative AI for streaming video discovery.
The new Quickplay Curator Assistant announced Wednesday, powered by Google Cloud and Vertex AI using Large Language Models (LLMs), empowers content programmers to create dynamic storefronts with personalized recommendations.
By analyzing various data points and historical performance alongside LLMs, Curator Assistant provides intelligence to support more seamless content discovery creation for programming teams, including by surfacing titles that might not otherwise be flagged by traditional metadata programming tools, ultimately enhancing viewer experience.
This builds on a multi-year collaboration with Google Cloud announced last September, which included a Media Companion app from Quickplay that uses GenAI to help viewers discover content they want to see via voice interaction with the TV.
Curator Assistant on the other hand is designed for the service provider’s in-house content programming team, where they can conversationally engage with the tool, to drive deeper discovery into their libraries and pass that onto more relevant content recommendations to viewers. That includes through auto-generated content rails based on results pulled from the catalog and further refining based on engagement metrics, as well as personalized content carousels based on individual preferences, content analysis and consumption patterns.
“OTT providers’ programming teams are the unsung heroes of content search,” said Paul Pastor, Quickplay CBO and co-founder in a statement. “Curator Assistant gives them increased power to put the greatest assortment of relevant titles front and center before viewers. Programmers will no longer be limited by their own or licensed metadata - they can literally leverage the entire internet, with conversational search, to program their service and drive more discovery opportunities that are personalized to the user or cohort.”
Per Quickplay’s messaging, the Curator Assistant uses Google’s Vertex AI to help programmers discover new opportunities to merchandise their content. The Google Vertex AI in turn leverages LLMs to return results that tap into a much broader database than traditional metadata tagging.
Query results are matched in the Quickplay CMS against available content rights and user profiles, resulting in content presentation that is “both surprising and relevant”.
Timed with the announcement is a blog by Google about the many third party integrations for media and entertainment being enabled on Google Cloud. This includes development of another conversational GenAI search and discovery tool, cineSearch, from Cineverse.
Albert Lai, global strategic industries director for Media & Entertainment at Google Cloud, said in the blog post: “Consumers continue to demand more personalized and diverse entertainment offerings. To keep pace, media and entertainment companies need powerful and flexible tools that accelerate time-to-market for these new offerings, while also optimizing costs. This is where Google Cloud and its partner ecosystem comes in.”
Efforts to improve streaming content search, recommendations, personalization and discovery are happening in the context of a highly saturated U.S streaming market where churn (subscription cancelations) remains at record highs.
In its Q1 2024 report, Antenna found that churn for premium streaming SVODs had tripled in the last four years, pressuring net additions and growth overall. It also identified a category of ‘Serial Churners’ — individuals who have three or more cancelations of a premium SVOD service in the past two years. That segment now comprises nearly a quarter of streaming consumers.
Deloitte reports that 40% of consumers have canceled a streaming service in the past six months. It also notes that 10% of cancelations resubscribe the next month, and one in three are back by six months after canceling. That may be small comfort for streamers looking for consistent revenue but it’s behavior that more sophisticated customer retention strategies could potentially help curb.
The streamer business model has shifted from a focus on subscriber acquisition to amass scale, to one of managing subscribers and seeking profitability.
“This will translate to much more sophisticated marketing and product strategies, new success KPIs, and a whole lot more reliance on data,” said Antenna in its report – pointing to its own market intelligence skills to deliver this.
Surfacing better and more personalized recommendations and making content discovery easier could also be part of the picture in helping streamers retain viewers.
Metadata helps FAST discovery gets more personal
Separately, TV measurement company Nielsen is pushing its metadata services unit Gracenote as a means of monetizing and personalizing free ad-supported streaming TV (FAST) channels.
In a blog post ‘FAST has made linear TV cool again, Personalization will make it cooler’, Nielsen argues that to make FAST channels distinctive in a saturated market (and therefore profitable) data enrichment strategies are required. These, it said, build on the basics of imagery and descriptions by providing a more complete picture of available content. Additions like mood, theme, scenario and setting available from Gracenote “provide a new layer of information that can elevate the appeal of programming among audiences looking for something to watch.”
“The importance of normalized and comprehensive metadata can’t be over-emphasized,” Nielsen stresses. “Enriched metadata can help ensure that content stands out as FAST experiences build out enhanced search and discovery algorithms for recommendation engines.”
Capturing viewer attention with enhanced discovery
Part of the problem with capturing consumer attention is that viewers’ time is being more and more fragmented, including streaming TV competing with social media sites and video games.
As Deloitte TMT leaders put in the article titled ‘Streaming video at a crossroads’: “The biggest challenge for SVOD providers and studios may be philosophical: They no longer address a mass culture, but rather a fragmented landscape of competing digital entertainment options. Trying to rebuild pay TV business models around streaming services could help reduce SVOD churn and slow attrition in the near term, but the long game for success will likely involve reinventing the medium to be more personalized, more shoppable, and more social.”
And Deloitte pointed specifically to consumers’ desire for enhanced content discovery, noting almost 50% of respondents to its Digital Media Trends study said “they would spend more time on streaming video services if it was easier to find content,” potentially serving as a way to help keep viewers around.