The re-emergence of Contextual Ad Targeting
Recall, advertising, in its many forms, is about awareness. Effective advertising informs an audience about a brand, product, or service. The intent is to educate, persuade, and remind consumers. The goal is to match the most relevant ad to the right audience. While many ad platforms would have you believe that behavioral targeting, based on collected personalized information, is the holy grail in the digital landscape, this is self-serving and only works to protect the moat that ad platforms have built. There is another form of advertising that is as effective and far less invasive, and it doesn't have such a high dependency on large ad platforms like Google's.
The operating theory of contextual advertising is that if an advertiser knows what a user is doing at any particular moment, they can use that knowledge to present an ad that will be relevant. For example, if a user is reading an article about motorcycle maintenance, an ad from a motorcycle brand would likely have high relevancy to that particular consumer. The "context" of the user's experience is the key to ad matching, rather than personal information captured via cookies, fingerprinting, and consumer profiling.
In the earlier days of digital advertising, context signals were used to curate a finite number of consumer experiences, often through trial and error. Today this process is greatly improved with the use of machine learning to create a more personalized result.
How Context Is Defined
In modern contextual advertising there are many signals, that when used together, produce a more accurate prediction of what matters to a consumer. These signals are only about the active session with the consumer and include data such as consumer input, location information, clicking and scrolling, and the types of content a consumer engages with. Unlike with behavioral targeting, no previous data about the user is used as part of the ad matching process. Consumers understand the differences and an overwhelming majority prefer contextual targeting over behavioral targeting, they are uncomfortable with ad placements requiring personal information.
Contextual targeting is greatly enhanced by the meta data associated with a piece of content. During the editorial workflow, content creators use both manual and automated tagging to identify the subject(s) of an article. Natural language processing (NLP) can score the article on several dimensions as well: additional subject tags, reading level, or sentiment score are all examples. All of this information is associated with each article, providing elements that can be used to match a consumer's signals to a more relevant ad.
Applying machine learning to this form of advertising trains algorithms to identify the types of content that most resonate with a consumer. This is achieved by using the consumer session data stream, article meta data, and the historical impact that previous ad matching achieved. Over time, the algorithms improve their ad matching capability, learning from real world data, and dynamically updating so that future matches are improved. All of this analysis happens each time a consumer begins a session on a first party domain website. And in a purely contextual advertising scenario, it will be repeated for that same consumer the next time they return to the site. The only history of the user the site has is the near-time activity the user generates.
Many sites will combine the practices of both behavioral targeting and contextual targeting. They do so by leveraging a longer data history for consumers that are willing to authenticate and allow first party cookies, and by enhancing the experience from real-time actions a user is taking on a web site. In a world without third party cookies, advertisement funded web sites will likely invest in additional contextual ad targeting and create new consumer experiences that rely on authentication where a consumer profile can be created to leverage as part of behavioral advertising.
Proven Results
Where the approach with behavioral targeting drives volume, contextual targeting improves relevancy and therefore ad quality. Publishers benefit by providing advertisers a more engaged audience, establishing a premium for contextual ads. Advertisers are willing to pay a higher ad impression price when consumer engagement increases proportionally. In addition to the privacy improvements and a better user experience, contextual ad targeting also offers greater brand safety for advertisers and lower technological hurdles to implement.
With increasing attention on consumer data privacy and data regulations, companies are shifting their ad targeting strategies. For example, DuckDuckGo, since it inception, has eschewed behavioral targeting and generates part of their revenue via contextual ads. New York Times International disabled behavioral ads in 2018 to ensure compliance with GDPR requirements and has "not been impacted from a revenue standpoint, and, on the contrary, our digital advertising business continues to grow nicely.” Results from a separate study by GumGum showed "contextually relevant ads generated 43% more neural engagement and 2.2 times better ad recall." NPO, the Netherlands public broadcaster, cut behavioral ad targeting in 2020 and saw ad revenue rise several months in a row, even during the early months of the COVID-19 pandemic.
While the ad platforms will wait until they are forced to improve consumer data protections, publishers and advertisers can benefit now by adjusting their marketing strategy to have a heavier reliance on contextual ads. Not only will their businesses benefit financially, they will also generate goodwill from consumers and be on right side of any future privacy and data regulations.