Metadata is a word that shows up in many different contexts across a variety of industries. While the word can easily be defined as “information about data” (or “data about data”), its importance in the world of television cannot be overstated. This is especially true when it comes to content discovery and personalization. That’s why we asked Michael Kraskin, director of product management at TiVo, to join us to help clarify the role of metadata in the television industry today.
What is metadata in the context of TV content?
Michael: Metadata is basically all the information any modern video service needs to create a satisfying experience for an end user. This could be an electronic program guide (EPG) which has all the schedules for every channel, along with all the titles and the other relevant metadata that a customer would need to make a decision about which channel to turn to. Or it could be a video-on-demand service, an SVOD or TVOD, etc.
Metadata is going to include obvious information like the program title and description, but also things that will let you differentiate one program from another, even if they have the same title. That will include unique information like artwork, the year a title was released, runtime, age rating and the reasons for that age rating.
How does metadata improve the searchability of TV shows and movies?
Michael: Metadata is going to help in a couple of different ways. The most obvious and common use case is people searching for a program by name. Providing an accurate title is important, whether it be for a movie, a TV series or an episode.
Descriptors like genres also help improve searchability. For instance, let’s say you’re in the mood for a comedy. Comedy is a genre, which is one type of descriptor, probably the most familiar and most basic.
Genres — and everything we have discussed up until this point — are what we think of as basic, factual metadata. But, at TiVo, we take it to another level and provide deep descriptors, which describe the emotional and artistic qualities of a program. This will help searchability even further, as well as improve personalization.
In what ways does metadata enable personalized content recommendations for viewers?
Michael: This is where those deeper descriptors start to become interesting. A good personalization engine is going to take everything they know about a given program and start to build associations and profiles to understand what types of programming the user is interested in.
One example I like to give is when we identify someone who likes a romantic comedy. Again, comedy is the genre and romantic comedy is a broad subgenre, and you could just target them with one romantic comedy after another. But when you start to dive into the wide variety of romantic comedies that are out there you find that some are racier like “Sex and the City.” Or, from a specific time period like “Shakespeare in Love.” There are romantic comedies about heads of state, “The American President” or “Coming to America” for example, where the stakes are perceived to be higher because the characters are world leaders.
We need to understand what specifically is driving someone. It might not be as simple as just the genre. You need more depth. In this example, the person is not necessarily just into romantic comedies. Maybe they like content set on a global stage with heads of state. With good metadata, you can draw these interesting connections.
What role does metadata play in understanding and predicting viewer behavior and preferences?
Michael: Think of metadata as a two-way street. A platform owner (cable company, OTT service, etc.) will need to power that recommendation engine and put accurate information on the television.
That’s all about providing metadata and recommendations based on metadata to the customer. But the customer is also telling you something about themselves. With every click, every image they dwell on, every description they read and every decision to watch something or not, they are telling us about their preferences.
This again points to the importance of the depth of the metadata. What trends are being accepted by the user and getting them to click on it? What similarities in programming are getting that user to stay and watch? On the flipside, what similarities in the metadata are getting them to reject a piece of content? Users are teaching us about who they are. That can guide all sorts of business decisions, not just recommendations for this one customer, but in the aggregate what type of content should I be investing in? What aesthetic, artistic choices does my audience respond to?
Overall, there’s so much that a user is telling us about themselves. This information could also be used for advertising purposes. If you are a service that targets advertisers, being able to deliver information about your customers upstream to potential advertisers to aid in targeting and creative development, this is invaluable information. Hence the two-way street of metadata.
How can high-quality metadata drive increased viewer engagement and satisfaction?
Michael: With high-quality metadata, you can better target content. You can learn about your customer and give them the right experience at the right time.
A newer offering we have to address this is a popularity score. We can tell, based on proprietary data, what programs are the most watched and enjoyed right now.
The second new offering is an engagement score. Engagement is more behavioral than popularity. In other words, a program with real engagement is not just one people are watching or talking about because everyone else is. A program with engagement is one that interests the viewer, draws them in and has them on the edge of their seat. This is of particular importance to advertisers because studies demonstrate that when someone is actively engaged in watching a program, they are more likely to retain the information from the advertisements as well. Higher engagement equals more brand recognition.
For more information on TiVo’s metadata capabilities, click here.
