The basic use case for contextual advertising is well understood: put an ad next to content that reflects either the product or the audience it is trying to reach. For example, if an advertiser were keen on reaching an LGBTQ audience, the advertiser might consider advertising on an explicitly LGBTQ site such as Out Magazine.
That’s how context drove the industry prior to the cookie and programmatic. Context was effective but blunt: it lacked the journalistic know-how to discern which articles in which publications would be most relevant to an audience. Then came programmatic contextual.
The industry has been using a very basic form of context in search advertising for more than 20 years. It’s a simple premise: enter a bunch of words and terms into a walled garden interface and they’ll target those words across their distribution. But this happens without consideration of the multiple connotations those words may have. This could lead to your ads appearing on a page where those words and terms take on a very different, and possibly negative, meaning. For example, you wouldn’t want an ad targeting an LGBTQ audience to show up next to an article about a hate crime.
Resolving ambiguity is straightforward. All you have to do is check the surrounding environment of content so you can be sure those words mean what they’re supposed to and that they’re in an environment you want to be associated with — not one you want to avoid.
The solution is to understand the topic of the content first and then examine the words you’re interested in. That’s what our engine does 1 million times a second across 3.5 billion pages from 55,000 sites. It checks the topic of the content first and then determines if the words are appropriate to target with ads. So, an ad targeting an LGBTQ audience would never show up next to an article about an anti-LGBTQ hate crime, which makes sense to target based simply on keywords but not based on a deeper topical analysis.
Once you understand the topic of the page, you can also leverage other aspects of context like tone, sentiment, and psychographics: all excellent inputs for deciding a programmatic bid price. Then you have a winning combination. You’re confident about your keywords as well as other attributes of the content you want to target or avoid.
Now you’re ready to take the contextual signal and apply it to both audience models for walled gardens and inventory models for digital.