Media companies spend a lot of time and money studying and modeling consumer behaviors. It's big business and a critical component of today's media marketplace. Entire companies, platforms with specialized engineering teams, academic researchers, entrepreneurs, and the public attempt to find the Holy Grail of search algorithms that provide the best way to recommend titles, so you don't change the channel.
Algorithms are components of computer programs that analyze data to identify market and user trends, track inventories, improve network traffic efficiencies, and provide content recommendations to new or long-time system users. Artificial intelligence and machine learning are increasingly incorporated to augment and add insights into data analysis that would take years to build into the datasets these programs use. But when it comes to recommending what program to watch, one type of information is so important that without it, algorithms would fail miserably: metadata.
We've written about metadata before , so there's no need to revisit all of that here. In this post, we will focus on how metadata contributes to the effectiveness of search algorithms and why getting it right can lead to increased revenue.
Metadata Feeds Search Algorithms
Over 1.7 billion households are now searching for a movie or TV show to watch over the air, stream, buy or rent. That number is expected to increase to 1.8B by 2026. At some point, whoever controls the remote has to think about what to watch. If they're searching on a VOD or OTT platform, they're entering titles, names of actors, genre type, age-rating, words, or phrases that describe what they'd like to see. If the customer has been on the platform for a while, the service has kept track of what they've watched and searched their catalogs for titles with similar styles of content.
Think about it this way: if you're searching for a title to watch with young children, your search is likely going to include a "G" (or comparable) age rating. That's metadata. If you're looking for a romantic movie shot in Ireland, both the genre and the location are metadata. If your favorite musician is Andy Gibb, and you want to find which movies or TV shows he appeared in, guess what? You're going to find out using metadata.
That's the high-level stuff. Studios, distributors and platforms utilize metadata that consumers don't even think about when building their search engines and algorithms. Here are a few examples of those kinds of metadata:
Depending upon the platform, the number of metadata fields varies. Some platforms may request more descriptive details on character traits, such as whether the lead is "kind" or "obnoxious." Is the story originally written for the screen, or is it based on a novel or actual events? Does the film have a strong female lead, or is it a film about a group of friends? These data add context to the film record and enhances search, classification and matching capabilities.
Additional subscriber and profile details are drawn upon to further define possible interests. For example, is the subscriber male or female? What is their income level? Where do they live? Are there children or senior citizens in the home? Which movies do critics or people near them recommend? These are key factors in personalization.
There can be thousands of bits of information used to suggest something for you to watch. For example, Netflix has 222M subscribers , each having dozens of data points about their content preferences and watch history. The amount of data processed for each search means not only must the programs and network systems themselves be extremely robust, the algorithms doing the work are very complex.
Figure 1 is a search algorithm. This one won the Netflix Prize, which the company crowdsourced to see if their search model could be improved by more than 10%. The winning team was awarded $1 Million. You can find details about their formula here .
Why Content Creators and Platforms Take Metadata Seriously
The theatrical, linear, streaming, online, or retail video content market is enormous. According to IMDb Pro, over 235,000 TV and movie titles are available in the US alone, and over 5.7 million available worldwide. The question for content creators and distributors is how will you stand out in such a crowded market? How will you get to the top of the search results? Can you even get noticed?
Notwithstanding whether the title is any good or not, metadata alone isn't going to get it near the top of the search results page, but it can help. Providing as much of it as the platform or store requests and developing it is a good investment of time and resources. Search algorithms do not care whether your title has data for each of the fields it utilizes, but you can be sure that if nothing is in the key fields, your title may be harder for consumers to find. The worst films still have metadata that describe them, especially if they've won awards for being bad. You may not have watched them, but most movie junkies have heard of " Plan 9 from Outer Space " or " The Room ." When you search for them, you're going to find them and see detailed information about the title, its plot, actors and director, and why people think it's so bad that it's worth watching.
Obtaining high-quality and effective metadata is not a task left to the uninitiated. Studios like Disney and platforms like Netflix and Amazon Prime Video have teams of employees or contractors whose job is to watch and annotate metadata for their original movie and TV titles. Companies who distribute titles via TVOD or retail stores know better metadata makes their titles more easily findable by consumers and thereby more marketable. That means more sales, more rentals, more views and more revenue.
Global listings and metadata is one of Spherex's core businesses and provides studios or platforms easy access to a massive data store of over 1 million unique titles, including artwork variations and trailers in 45 languages spanning over 140 locales. Covering many languages, versions, and formats, the Spherex datastore contains nearly 25 million title records for Hollywood's top movies and tv shows and titles produced worldwide. Title records are cleansed, normalized, localized and ready to use.
It's easy to dismiss or not be too concerned about a title's metadata quality because it's not something people see. But whether they realize it or not, it is something they use every day. With nearly 600,000 new titles released worldwide each year, competition for the top placement on results pages is only going to get more intense. Understanding the importance of metadata, taking advantage of its proper use can help get your content the audience attention it needs to positively impact your bottom line.
