Encouraged by the success of AdWords, their paid search program, Google soon after launched AdSense, which extended pay-per-click search listings beyond search to the web as a whole. Any publisher, large or small, could partner with Google to provide advertising for their pages. Google would display the same advertisements used in their search listings, in the hopes that users browsing on the publisher’s website would see the ads and be interested enough to click. In this model, the publisher relinquishes control of the ad unit to the ad provider (Google, or any of a number of other CPC text-ad providers). The responsibility lies with the ad provider to decide which ads to display.
The method most commonly used to select the advertisements most likely to be interesting to the user is called “Contextual Targeting”. In contextual targeting, the ad provider examines the content of the page the ad is being shown on, and looks for keywords that provide clues to the subject. Then, the ads associated with those keywords are displayed, as if the user viewing the page had performed a search.
As an example, if a user views a page containing reviews of the newest MP3 player, the ad provider could find the keyword “MP3 Player” in the content, and display the same set of MP3 Player related ads they would upon a search query. The idea is that if the user is viewing a page about a certain concept, he or she is likely interested in that concept.
Contextual targeting works very well for a wide variety of pages, but many times it is difficult to find context that can be monetized, and in some cases, it can be difficult to find useful context at all. For instance, if someone views someone’s profile on a social networking website, the content of the page is unlikely to provide any clues to the viewer’s interests.
There are many other targeting techniques used in CPC text advertising, and new methods are being developed all the time. All of the methods are designed with the same goal in mind: provide advertisements that are relevant and cater to the viewer’s interests. Two examples are “geo-targeting” (showing ads from local advertisers, based on location information gleaned from the viewer’s IP address), and “site targeting” (showing ads that tend to appeal to the audience of the website as a whole, as opposed to trying to target exclusively to the particular page).
More complex targeting methods are also possible. For instance, if a user is “seen” on a page that provides strong clues as to their interest, such as a page containing reviews of popular MP3 players, they can be “tagged” (via a cookie in their browser). If the user is later “seen” on a page with little or no inherent context, ads for MP3 players can be shown.
While extending paid search advertisements beyond search results in Germany has been somewhat successful, the value of the inventory rarely matches that of search. When a user uses a search engine, he or she is in the mindset to buy, is ready to commit. When the user is browsing the web at large, making a purchase isn’t the immediate goal. An analogy would be the difference between showing someone an advertisement for ice cream when they are at the grocery store, in the freezer aisle, trying to decide which brand of ice cream to buy, and showing the same person an ice cream ad when they are simply walking down the street past the grocery store. They might still be very interested in ice cream, but it’s not their immediate focus, so the percentage of people who will immediately follow through and buy the ice cream will be much lower.