How Metadata Powers the Shows We Stream
By Brendan Healy
Within the past 50 years, the medium of film and television has transformed drastically, from exclusively movie theater showings and linear TV programming to a plethora of streaming content options available from the comfort of home. With the count of streaming services in North America reaching almost 50, there is an option for everyone, from household names like Netflix and HBO Max to niche platforms such as anime-hub Crunchyroll. But as the offerings grow, streaming services will have to compete for viewers' attention while keeping them satisfied. According to Nielsen, 46% of viewers today feel that it is harder to find content they want to watch because there are too many services available, while the total number of unique programs across US traditional TV and streaming has increased roughly 26% since 2019. As the streaming landscape continues to intensify, a key tool streaming services will leverage to satisfy and retain customers is the ability to precisely search and suggest content.
The search engines that viewers use to find the movies and shows they want are powered by algorithms which utilize rich metadata. Beyond just titles, movie metadata fields can include genre, cast, synopsis, rating, images, trailers, subtitles, and even moods. Another key metadata field is the Entertainment IDentifier Registry (EIDR), which is a unique ID for every single film and TV episode ever made. This ensures there are no duplicate entries when dealing with different shows or movies of the same name. For instance, the UK version of the series “The Office” has the same name, format and genre as the US version, but they each have unique EIDR IDs that are 34 characters long. It is typically the role of the producer to provide the metadata for their content. Platforms like Netflix and Prime Video have standardized metadata templates in order to streamline the delivery process and ensure uniformity across their entire library.
Netflix is well known for its recommendation engine which provides viewers with personalized recommendations in order to prevent them from getting lost in a vast library of content. The recommendation engine utilizes data on viewing habits, searches, scrolling, ratings, time of day, and devices used to find similar users on the platform and recommend the types of content that they have enjoyed. This is where the metadata that each piece of content is tagged with is utilized to find similarities amongst titles. Not only is metadata utilized to suggest what to watch, but also what cover image to display based on a viewer's preferences. As seen in the image below, a single title can possess multiple images with unique metadata for themes, scenarios, moods, celebrities and more to determine which image is most appealing to a certain viewer.
Nielsen’s Gracenote Personalized Imagery service offers sets of differentiated images that convey various key aspects of a movie or show
Metadata is so crucial to Netflix that they employ a team of metadata taggers who are tasked with watching every piece of Netflix content and tagging it with the relevant metadata. These film buffs log everything from the level of profanity or sex in a film to storyline tags like “doomed love” or “technology gone wrong.” So, the next time you are scrolling the ‘Users like you also like’ page on your favorite streaming service, think about the thousands of metadata tags fueling those recommendations and what tags you might give your favorite show.
Sources
