Lately, contextual advertising has made a major comeback. Once the golden child of the early 2000s advertising world, the practice—in which targeted ads are placed on pages based on the content of those pages—is now again attracting attention, this time as a potential solution to our inevitably cookieless future.
But some believe that going back in time isn’t the best way forward—myself included. As CEO of digital marketing company Semasio, I see the return to legacy contextual advertising as an unnecessary regression. At Semasio, we’ve developed a novel approach to consumer and content profiling based on semantic analysis. A semantic understanding of the terms that pages contain and internet users consume gives us a unique opportunity to seamlessly target across users and pages, a practice we call unified targeting.
With our roots in audience targeting, we see the cookieless future as an opportunity to move into new, innovative approaches in the intersection between audience and contextual targeting—a synthesis of the two.
This synthesis is called contextual audience extension, and rather than basing itself on the content of a specific page, contextual audience extension looks at behavioral patterns of, for example, a first-party audience, and then uses those behavioral patterns to project the audience onto pages for contextual targeting.
For comparison, legacy contextual advertising is a one-trick pony of matching the marketing message to the page content, while contextual audience extension transcends that limitation through behavioral analysis of an audience to find net new context independent of topical connections. This enables us to see connections between audiences and contexts, which are invisible to traditional contextual targeting providers.
The bigger picture? I believe this synthesis of audience and contextual targeting will pave the way to the cookieless future.
Semantic consumer understanding came first
To fully understand what unified targeting is all about, it’s important to understand the history of Semasio itself, which initially was solely a semantic audience targeting company. In 2019, we started to see that user identifiability, and thus audience targeting, would come under increasing pressure, so we took our semantic profiling and targeting expertise and expanded it into contextual.
The approach uses natural language processing to understand the most significant terms in the content a user is consuming. If a marketer, for example, is looking for people with a current affinity for baking recipes based on apples, he might input “apple” into the system. He’d then get a keyword cloud with all the keywords co-occurring with and co-consumed with the word “apple.” But due to the multiple meanings of the word, this keyword cloud will be messy, containing a smattering of terms related to consumer electronics, trees and baking recipes. By marking terms like recipe, flour, butter and the like as correct, and terms related to consumer electronics, banks and trees as wrong, the user is disambiguating the word “apple” by providing context around it.
The importance of the seed audience
The final part of the equation is the incorporation of a seed audience. Let’s say you want to sell small electric cars. If you were going with traditional contextual targeting, you would put your advertising on a page that’s about small electric cars or related topics. The problem is that this is exactly what everybody else is doing, which means large demand for small supply, driving up prices and hampering scale.
Fortunately, contextual audience extension does something very different. The platform might look at the behavioral patterns of the users actually signing up for a test drive in the small, electric car (first-party audience or seed), and uses these patterns to project the audience onto the contexts. Think of this process as building a dynamic heat map of the internet for that unique audience. If a member of the seed audience visits the page, it heats up the page ever so slightly. If a nonmember visits the page, it cools down the page slightly. This means the more people from the audience you have relative to people out of the audience, the hotter this page will become.
This novel approach establishes a connection between audience and contexts, which transcends that of legacy contextual targeting technologies. It enables the creation of net new supply that is invisible to the competition, producing lower costs and greater scale.
A lot of people are conflating the deprecation of third-party cookies with the deprecation of any kind of user identifiability, but the truth is, there are interesting approaches out there, compelling alternatives to third-party cookie-based user identifiers. The key is to build on the successes of the past to create a whole new future.
It’s all too easy to regress into a past marketing technique that was undoubtedly successful. But I truly believe that the post-cookie future of digital marketing is bright and full of innovation.