Semantic Targeting for Ad Ops

One technology that’s growing in relevance for publisher ad operations is semantic targeting. Built into every ad server is the ability to target or exclude campaigns from appearing based on keywords. But semantic targeting goes beyond keywords in ascertaining the meaning of the content on the page the ad is to appear. By mentioning that I want to really “kill” with my column, it’s possible that relevant advertising would not appear because of any keyword filtering going on. A good semantic targeting technology would ignore that one instance of the word and just assume that I have the same vocabulary as a 13-year-old surfer.

It’s this intelligence about the context of the page that makes semantic targeting so useful. When the wrong impression appears next to the wrong content at the wrong time (travel ads on pages about airplane crashes is the most used example), ad ops is held responsible. It’s unfortunately a scenario that can’t be prevented 100 percent of the time, but semantic targeting can go a long way in reducing the possibility. One less thing to worry about means something else can get the focus it deserves.

But semantic targeting goes beyond prevention. It can be a tool to help drive revenue in real and meaningful ways.

A business-oriented site taxonomy. As editors add content to the site, they take responsibility for tagging and sectioning the content. They are, of course, thinking of how consumers will navigate the site and decide which content they will view. Ad operations follows the editors’ direction in organizing ad inventory. Unfortunately, that means certain revenue opportunities won’t be realized. A sports article on the business of baseball, for example, may only be labeled as sports. Ad ops would love to see this same content labeled as business to help grow inventory that commands a higher CPM.

One can solve this problem by creating tools that allow operations to tag content in the same way editors do. While that’s the ideal solution, it unfortunately doesn’t scale. Ad operations can’t be spending time re-categorizing content. Using a semantic targeting solution allows for content to have additional categorization done automatically that ad ops can use to increase the amount of high yield inventory.

A publisher-focused data revenue source. As much of the industry moves to a data-driven, audience-targeted marketplace, publishers are finding others are determining the value of their inventory. Semantic targeting becomes one way in which publishers can now play in the game and help determine value. Since the semantic categorization of the page is done with full visibility to the publisher, the publisher can determine if and how it goes to market and at what price. As an ad operations person, I would be working with solutions that allow me these controls.

Context will be king again. While audience and data are the basis for a growing amount of targeting in the industry, context will always bring more value. Auto intenders are a valuable audience segment. Auto intenders viewing the right content are more valuable. An important point is that it may not be auto content that they should be viewing. I learned this years ago as one of Tacoda’s first clients that auto intenders were more likely to convert outside of our auto section. Context of where audience-targeted ads will be delivered is an opportunity for buyers and sellers to further negotiate the value of an impression. Semantic targeting will allow this to be done at a meaningful scale.

Privacy regulation may increase its value. Regardless of your view of how regulation may or may not impact our industry, it’s probably in your best interest to start working on targeting solutions that don’t rely on third-party data sources. Semantic targeting which doesn’t require a cookie should be a safe bet as something that will increase the value of impressions if and when regulations come down. As with most things operationally-focused, it’s best to get some experience with a technology before you are in crisis mode.

Related reading

hp
Why do we love visual content
Stats_Simpsons_Mob_Featured Image
electronic circuit board