“The Sell Sider” is a column written by the sell side of the digital media community.
Today's column is written by Alysia Borsa, chief marketing and data officer at Meredith Corp.
During the recent Advertising Week and Programmatic I/O, everyone was talking about data and insights. But often overlooked in this discussion is a key source of insight that’s as old as websites: taxonomies.
For publishers, the depth of their data and insights depends on the depth of their taxonomy.
Taxonomies initially served as a site navigation tool for editorial teams to classify and place content, while search engines used taxonomies as a discovery mechanism to help users find content.
Today, a publisher’s taxonomy is quickly becoming a business differentiator. Deep taxonomies enable publishers to offer more robust and precise targeting for advertisers, provide deeper insights about users and inform editorial strategies.
Investing in a robust taxonomy isn’t the right strategy for every publisher. It’s costly in terms of people, time and technology, and since taxonomies are ever-evolving, they are never truly “complete.” If you don’t have large scale, it likely isn’t worth the effort. After all, tagging a fitness article with 100-plus attributes is overkill if you only have a few hundred pieces of fitness content.
For publishers with large scale, however, building and maintaining a deep taxonomy is a worthy investment and makes their content and experiences more valuable.
The interplay of editorial expertise and technology is the foundation for a strong taxonomy. Third-party taxonomies from Google and the IAB are a useful starting point, helping standardize content to five to 10 levels of depth and suggesting keywords publishers may not have considered.
But five to 10 layers isn’t enough. Large publishers can easily apply 100-plus attributes to a piece of content, which requires a human, editorial layer of expertise to get it right.
Publishers should ask their editors, “How would you categorize this piece of content?” They may be surprised by the response.
For example, a single piece of food content, such as a chili recipe, may warrant hundreds of tags, including “Mexican,” “easy cooking,” “crock pot,” “low-carb” or “ground beef,” as well as occasion-driven and psychographic tags like “Sunday BBQ,” “family meal” or “feel-good meals.” Only by involving the editorial team that creates the content can publishers be sure their taxonomy is nuanced and accurate.
Automate what you can
Once publishers have a set of “seed” content – anywhere from 100 to 1,000 pieces of content – that is properly categorized, they can use machine learning and automation to apply their taxonomy to the rest of their content. Automation greatly reduces the amount of time it takes to apply a taxonomy, and it removes human subjectivity from the equation, ensuring all content is categorized in the same way.
Even so, automation isn’t a panacea. Human taxonomists must go back and audit the deployment, spot checking that it works correctly.
Ultimately, consistency is the key to a successful taxonomy, both at the outset when a taxonomy is created and on a continuous basis moving forward. Taxonomists and editors must work hand in hand to ensure existing and new content is tagged consistently.
For example, there are many pieces of content across Hello Giggles, People and other sites about Snoop Dogg’s and Martha Stewart’s friendship. But Snoop Dogg’s public name has changed over time, ranging from Snoop Doggy Dogg to Snoop Lion and Snoopzilla; what rules can publishers set to teach their systems that all these pieces are Snoop Dogg-related, despite the different names?
Taxonomies aren’t static; they are dynamic and always evolving. Just as taxonomists must train editors how to classify content, editors must also train taxonomists about emerging keywords.
New terms appear all the time. New terms and their synonyms must be identified and added to the taxonomy, then sent back through all existing content. Recently, for example, the surge in interest in the ketogenic diet has created a need for not only a “ketogenic” tag, but also a “keto” tag, which is what most people search for, and new rules to add these tags to many existing “low-carb” recipes. Loading up-to-date tags and keywords into a CMS makes this process easier for editors to employ.
Benefits for publishers
It’s a lot of work, to be sure, but a rich taxonomy is worth the investment, bringing greater return to publishers and their advertising partners. A deep taxonomy is the hallmark of any serious data player in the industry.
A rich taxonomy, for example, allows publishers to get very granular about what’s resonating with users. Instead of targeting a hair care brand’s ads based on “hair” content, for instance, they can be targeted as “keratin-treated,” “frizzy hair,” “bad hair days,” “busy moms” or “quick fix.”
Publishers will also achieve deeper audience insights. A rich taxonomy helps identify trends at more granular levels and predict trends before they happen. Publishers can ask editors for “edge case” topics before tagging and tracking engagement with those topics. For example, we saw a rise in gochujang, which is a common Korean ingredient, and jackfruit as a vegan meat substitute, which is now sold at Whole Foods alongside pulled pork.
It also helps recognize emerging search trends more effectively to inform editorial strategy. Editorial teams can then fill content gaps or promote existing content based on the growing demand.
Perhaps most exciting, a rich taxonomy can enable publishers to make a seamless move to voice search and voice-activated skills. Taxonomies provide rich context, helping publishers present the most relevant content in this new channel.