By Winston Owens, VP of Content, IRIS.TV
A well-tailored taxonomy is the foundation of any successful publisher’s video strategy. If metadata is the lifeblood of content discovery, taxonomy is the circulatory system. A taxonomy’s configuration of categories and subcategories provides context for the discovery and recommendation algorithm which interprets a video’s more granular associated metadata (keywords/tags, title, description, etc.) to facilitate video recommendations.
Over the past ten years, metadata has transitioned from being largely viewed as an outsourced and/or “intern task” into one of the most important components for intelligently managing online video. The foundation of any well-rounded video strategy is centered around metadata management and taxonomy design.
Video sites like Netflix* — and other algorithmically-centered platforms like Pandora — have contributed to a sea change in the perception of just how powerful metadata can be when used properly. By employing a team of dozens of people who apply tens of thousands of tags and descriptions to videos, Netflix is not only able to generate accurate recommendations on a fairly granular scale but also simultaneously analyze the performance of those user-interactions and recommendations using their metadata as a tracking mechanism for content performance.
While smaller video publishers might not have the resources or scale to employ hundreds of people to apply and manage their metadata — as Netflix has done — there are basic tips and tricks that can be utilized to maximize the library value of any online video publisher.
The real ROI comes from a publisher’s ability to create a taxonomy that simultaneously takes into account its business goals (branded video campaigns, increasing views/viewer retention, etc.) while also considering the subtleties of the brand’s editorial that resonate most with their target audience.
When creating a taxonomy, publishers should keep in mind that “categories” essentially function as content buckets (audience verticals) whereas metadata capture the more descriptive and nuanced qualities of a specific video (e.g. people, locations, institutions, ceremonies, etc.)
While there is a general template for creating taxonomies, editorial input is integral for capturing the nuances that enable the taxonomy to be an effective tool for video publishers. Video publishers are encouraged to build taxonomies that broadly address their brand’s primary LENS category — which stands for (L)ifestyle | (E)ntertainment | (N)ews | (S)ports — while also including brand-specific categories that capture the peculiarities of the publication (i.e. branded special events, ongoing original series, etc.).
Things to consider when creating a taxonomy:
1) Scope of publisher’s coverage — The wider the scope of the publisher’s coverage, the broader/more disparate the publisher’s audience’s interests are. Larger taxonomies are more appropriate for publishers with multiple verticals within their publication. For example, a larger more sprawling taxonomy would be appropriate for a sports publisher that covers multiple sports and leagues vs. a sports publisher that focuses on a single sport and/or league. Another example would be a publisher that has coverage of the entertainment industry AT LARGE (film/tv, music, sports, tech, etc…) vs. a publisher that exclusively covers independent film.
2) Volume of videos in the library — The higher the volume of videos in publisher’s library, the more descriptive/targeted and higher in number the categories can be (examples would be: categories that cover “pop up” branded campaigns, special events, annual events in addition to their regular coverage). Publishers with smaller libraries are encouraged to start with a smaller, broader taxonomy that can be made more specific with additional categories over time.
Listed below are a few examples of how IRIS.TV generates taxonomies for publishers based on their corresponding LENS category (LENS: Lifestyle, Entertainment, News, Sports)
Lifestyle publishers tend to approach broad cultural subject matter through the perspective of a specific demographic. The most important thing to remember when building a taxonomy for lifestyle publishers is to keep the target demographic in mind. For example, some publishers are targeting women between the ages of 18–35, while others are broadly targeting “college-aged men.”
General Lifestyle Categories
Example of Fashion Taxonomy
Entertainment publisher taxonomies are fairly straightforward depending on the scope of the subject matter. For publishers dealing with content across mediums (e.g. films, tv show, music, etc) it’s best to classify videos as broadly as possible, while publishers who are exclusively covering an entertainment medium like comedy or music should aim to create taxonomic categories that specifically address their content.
General Entertainment Categories
Example of a Music-centric Publisher Taxonomy
Taxonomies for publishers in the “Hard News’ business are fairly universal as newspaper sections are reflective of specific audience verticals for ad-targeting (e.g. the “Sports Section” frequently has advertisements marketing local sports events, local sports gear, etc for that individual market.) Daily news publishers are encouraged to publish frequently and multiple times a day to maximize the turnover of often-fleeting “News” content.
Example of a Major News Outlet Taxonomy
Example of a National Election Taxonomy
Sports taxonomy generation is fairly straightforward. The broader the scope of a publisher’s sports coverage, the more general the categories (i.e. sports news aggregators). In instances where sports coverage is focused on a single sport or league, publishers should aim to use more sport-specific/league-specific categories.
Sports News Aggregator Categories
Example of a Boxing / MMA Taxonomy
Netflix Source: https://www.washingtonpost.com/news/arts-and-entertainment/wp/2015/06/11/netflix-tagging-yes-its-a-real-job/