July 8, 2013 – Personalization of content discovery is moving to the big screen in the living room thanks to some clever approaches to figuring out who may be watching on the part of navigation tool suppliers.
While the increasing sophistication of recommendation engines has made personalized content discovery a big part of pay TV providers’ efforts to raise the bar on consumer experience, it’s not been clear whether recommendations delivered directly through the set-top to the user interface on the TV screen would be much of a factor. Instead, the focus has been on creating navigation apps that transform smartphones and tablets into remote controls, allowing the personalized functionalities to play out on the personal devices.
Digitalsmiths is one supplier of advanced discovery solutions that’s in the process of figuring out how to deliver recommendations that accurately reflect the tastes and viewing behavior of who’s watching directly to the big screen. “It’s going to be about the patterns and the behavior to determine who’s using the device,” says Matthew Berry, co-founder and CTO of Digitalsmiths.
ThinkAnalytics is another system supplier working on this issue. “We recognize the set-top box is a multi-personal device, but we see opportunities to enhance discovery to where navigation on the set-top isn’t just another low-end user experience,” says ThinkAnalytics founder and CTO Pete Docherty.
“Rather than just household-labeled recommendations, we separate viewing patterns according to what happens at different times of the day and different days of the week so that we can more closely tune the recommendations to whoever is watching at a given time without requiring them to log in,” Docherty says. “What you watch before going to work is not the same thing as you watch in the evening or on Saturday night or Sunday afternoon.”
As previously reported, Cox Communications recently took the log-in approach to enabling targeted recommendations on its Trio Gide for subscribers to its Advantage TV package. The guide delivers personalized recommendations for up to eight individual profiles per household, encompassing viewing histories tied to live TV, VOD and DVR.
Other attempts at personalizing recommendations on the TV set are in play as well, notes Digitalsmith’s Berry. “You’re seeing television manufacturers come out with things like facial recognition or voice fingerprinting, but I think there’s the issue of privacy that consumers are worried about when we try to identify who’s using that particular device. I’m not sure today whether we clearly understand what’s going to work.”
But Digitalsmiths is taking a shot at figuring it out. While he’s reluctant to go into details, Berry suggests some of the ideas under consideration. “There are definitely tell-tale characteristics on maybe how you might navigate versus myself in terms of the speed of navigation, where we navigate, the search terms we use,” he says. “We can use that information to essentially model out that user to better predict who they are when they’re sitting down to watch content.”
In general, whether on the living room TV or personal devices, the user interfaces and advanced discovery modes of recommendations and search that will define pay TV navigation are very much works in progress, Berry suggests. “Everybody has their idea of what they think is going to work, and my guess is 90 percent of what we think is going to work is simply not going to work in the user experience,” he says.
But figuring it out is definitely a top priority, he adds. “There’s the nice to haves and the pains, and I think without a doubt this is a real pain point for the service providers,” he says. “The guide has been what you might call the home screen in the living room for a long time. Because it’s so central to the core of their business and can affect ultimately the top-line revenue of their business, they’re selecting their partners very carefully.”
Digitsmiths, with a broad base of clients ranging from film studios and television networks to consumer electronics manufacturers and service providers has been working closely with cable operators for some time, mostly under the radar, although the company did raise its profile with an exhibit at the Cable Show in Washington, D.C. last month. There it introduced its Unified Data Service, which is designed to overcome the silo barriers to aggregating metadata for search and recommendations by normalizing access across a wide range of metadata sources.
“Data is the foundational layer of what’s driving the discovery experience,” Berry notes. “If your data is noisy, you’re going to have noisy recommendations. So you really have to figure out ways to collect and normalize this data across many different data sets.”
The potential for noise is growing exponentially, given the multiplying number of sources service providers have at the disposal to help define on-demand and linear content for discovery applications. For example, Berry says, “now all of a sudden service providers are reaching out to a lot of different online providers like Metacritic, Rotten Tomatoes or Common Sense Media. Those are just a few, but there are literally dozens of these, and they all have unique ID sets that they go by.”
This is where Digitalsmiths comes in, he adds. “We ingest all this data, aggregate it and normalize it to a single ID that allows the service providers to use that data.
“There’s no real natural way to reconcile that data together,” he says. “We approached the problem probably a little uniquely. We have an automated process that knocks out a good portion of it, but then we have a manual curating process that will map all that data to a common ID. And that common ID can be used by the service provider to reach out and access a number of third-party services.”
Everything service providers want to include for use in the discovery process from these non-traditional sources must be mapped with the metadata repositories of the two incumbent providers of metadata for TV shows, Tribune Media Services and Rovi. “It becomes very difficult, because there’s an awful lot of linear content out there,” he says, noting the rate of expansion comes to about 20,000 new linear assets every 30 days.
Accessing results of social media tracking mechanisms has also become important to discovery as well as to determining how shows are performing in near real time. “For example, if service providers want to track Twitter Buzz, if they want to understand what people are tweeting about relevant to what’s on screen right now, they can take that ID, pass it into the systems and get the current buzz,” Berry says.
The ability to aggregate metadata for discovery extends to the proprietary time-based system that Digitalsmiths originally developed for the studios to provide a frame-by-frame accounting of the content. “For about eight years we’ve been working with the studios on helping them develop time-based metadata that describes what’s going on inside the video, who’s in the frame, where they’re at, what they’re talking about, any actions, objects. So we have all that information about all the content, and we join that data with all the other third parties that are out there. And that gives us a much clearer picture as to what’s going on to make recommendations.”
As for the discovery process itself, along with search that leverages all the metadata sources, Digitalsmith’s Seamless Discovery includes a recommendations engine which coordinates the content selection process across several parameters, including group ranking similarities, which associates a title highly ranked by a defined consumer segment with other items highly ranked by the same group as well as user rankings and viewing tapping into individual viewing and ratings history. Other parameters include content similarity, purchase history and popularity based on traditional ratings.
Increasingly, as the metadata and algorithmic infrastructure supporting next-generation navigation takes shape, service providers will be able to leverage this foundation to help shape their content lineups to maximum effect. “Service providers today are very quickly trying to figure out how they can get their time to market reduced in their next-gen discovery experiences, and that’s ultimately going to drive better planning on how they roll out and target content to particular users,” Berry says. “We’re just at the beginning stages, but I think that will start happening sometime next year.”