In my last post, I outlined my current plan for my thesis on targeted online advertising. This time, I’d like to answer what might seem like a very basic question, but it is still a very fundamental one: what do I mean by targeted advertising in the context of the web?
Targeting isn’t a new concept at all - marketers and advertisers have been finding ways to segment audiences for decades. What makes targeting in the context of the web so interesting to me is that it offers opportunities for near mass personalization from a variety of different angles and media formats in a very short period of time, and the capability to get real time data on how users respond to those changes. The flip side is the privacy issues that arise from those capabilities, which I will talk about in later posts.
I’ve divided online targeting into four categories:
- Geographic, where advertising messages are presented to users based on their estimated geographic location, derived from their IP address.
- Profile, where advertising messages are presented to users based on “traditional” demographics like age, gender, or geographic location, or on other self-descriptive information they may have entered in an online profile, such as interests, hobbies, favorite films, vacation plans, etc.
- Contextual, where advertising messages are presented on a site because it is complimentary to the content of the page being viewed. Examples include putting ads for cable packages on electronics sites that sell televisions, or hotel ads on flight booking pages.
- Behavioral, where advertising messages are displayed based on a user’s online actions, which could include what links they click, what time of day they are online, what sites they visit and how often, what search terms they enter, what they buy, and how much time they spend on a site.
If you follow this field at all, you probably know that behavioral targeting is the “hot” form right now that is getting a fair amount of press right now. There definitely is promise in analyzing behaviors to determine the intentions and motivations of a user, but I would content that targeting becomes the most accurate when we can effectively mix and match these techniques based on the behavioral data.
In my next post, I’ll elaborate a bit more on these categories, how they are related, and where the data for each can be derived. Thanks for reading, and please post any comments or questions.
