Metadata is simply data about data. While it sounds really boring, metadata is the key to unlocking exponential growth in viewership, discovery and monetization opportunities for premium video content.
Like everything in digital media, there are different flavors of metadata, with no standard definitions and a variety of descriptions. I'm going to muddy the waters a bit and provide my take on this vital component of the video landscape.
Standard metadata (for example asset metadata, attribute-level metadata, static metadata) consists of basic tags that describe a piece of content. Common tags include title, time and date: surface-level data that you would see in a typical YouTube video description (for example, "dancing baby").
While standard metadata provides important information, in the era of digital media, we need to dive beneath the surface and analyze video content on a frame-by-frame basis.
Enter time-based metadata (for example temporal metadata). This class of metadata provides a frame-by-frame analysis of everything that is happening on-screen throughout the entire film or episode. Time-based metadata gives you deeper insight into the video because you're analyzing the content at DNA level and providing a digital blueprint of content. Examples of time-based metadata include categories for actor names, characters, locations, scenes, objects, product placement, genres, dialogue, and subject matter. It even covers rights, licensing issues (geography, music) and distribution deal terms. In short, it creates a wealth of intelligence about your premium content that covers thousands, if not millions, of video elements.
If harnessed properly, this intelligence can drive new monetization opportunities for video content such as discovery, content recommendations and contextual advertising.
Advanced Discovery (for example, search)
What's the value if you distribute your videos to multiple platforms, but users can't find your content?
Time-based metadata allows for users to find an exact match of whatever they are searching for, and get the exact clip they need.
By contrast, traditional metadata search results would return full episodes, not exact clips. And those episodes may or may not contain the content the user actually wants to access.
Here's a concrete example of what I'm describing, using one of the most popular television series of all time.
Scenario 1: Say you're searching for a particular scene from a particular episode of "Friends," and all you remember is that Monica and Joey are discussing Valentine's Day. If the episodes were tagged with traditional metadata, your search would likely return every episode of "Friends" that mentioned Valentine's Day and includes Monica and Joey. Entire episodes, not clips.
If your search results come back with 10 episodes, at 22 minutes per that's almost 4 hours of content you'd have to sift through to find the exact clip you're looking for.
Scenario 2: Now, let's say the "Friends" library is tagged with time-based metadata. The same search with the same terms would return results of the exact scenes where Monica and Joey discuss Valentine's Day.
You could then generate a customized clip of that exact scene, or choose to watch the whole episode - and it would take seconds, not hours.
"Smart" Content Recommendations
With so many videos and so many choices, how do you keep users engaged and "locked in" to your content?
Time-based metadata can power similarly "smart" content recommendations and clips that help build loyalty and increase dwell times and repeat visits.
If you're watching "House" on-demand, your cable system (which already knows you and your preferences) may suggest some of Hugh Laurie's other work, such as "Blackadder." If comedy is not your thing and you prefer medical dramas, you get steered towards another NBC Universal television series like "Law & Order: SVU."
Simply put, knowing what content is made of allows for better, "smarter," more accurate content recommendations. This suggestive selling model translates into more revenue for the content holder.
If you worked for an airline, would you run an ad that followed a plane crash sequence? Hopefully not.
When an ad spot is available for sale, it is always best for the buyer to understand what is happening during that content. Time-based metadata gives advertisers the ability to not only determine where to put an ad break, but also can provide insight on what is happening right before and right after the break.
For example, if a film features two men fishing on a river, that could be the perfect spot for an ad for The Bass Pro Shops.
I'm giving micro examples to illustrate my points, but the larger benefit of time-based metadata is that it can apply to entire libraries of content. With automated controls in place, it's suddenly much more feasible to generate ad dollars from library and other archival content that is otherwise gathering dust on a shelf somewhere.
As you can see, time-based metadata is clearly the more robust solution for premium content holders. I think you'll see a clear trend of media companies "going deep" and moving toward time-based metadata as multi-platform distribution continues to mature. As an added bonus, when you "go deep" with time-based metadata, you can more effectively analyze viewership trends and behavior patterns. Understanding these trends can drive innovation, new types of content, new delivery technologies and new monetization opportunities.