Monday, June 25, 2007

Advertisements need to continue increasing personalization

According to Sean Ammirati at Read/Write Web this statement was made at Supernova conference by Sheryl Sandberg , Google's Global VP of Sales and Operations while giving a presentation entitled What's next for advertising?.

Sean Ammirati Says
“This [statement by Sheryl] was surprising to me, given that I don't believe Google has publicly announced any plans to incorporate behavioral targeting into their ad delivery system.”

Those who have been reading my blog know that I have been speculating about Google entering the Behavioral Targeting field. This statement by Sheryl reaffirms my belief.

Why do I think Behavioral Targeting makes sense over any other form of personalization?

Let’s start with the type of data that can power Online Personalization. There are 3 main form of data collection that can be used for personalization:

1. Explicit User/Visitor preferences – This is what a users tells a website about their preferences, interests etc and the site serves them appropriate experience, products, content, ads etc.
Example of this are – you customize what you want and how you want it.

2. User/Visitor Demographic data - This form of personalization takes users Demographic data and serves up appropriate product, content, ads etc. recommendations based on those demographics. This could be a combination of explicit and implicit (location) form of data collection. Most of the data however is entered by users e.g. age, gender, income etc.

3. Behavioral Targeting – This form of personalization does not require any explicit user data. In this form of personalization Users’ intent and preferences are inferred from their browsing behavior via an anonymous cookie and not tied to any PII (Personally Identifiable Information) data. Based on users’ behavior marketers put the users in one or more of the predefined segments and then serve content, products or ad appropriate for the segments that the visitor falls in. Publishers/Networks/Advertisers don’t know who the visitor is all they know that cookie ABC123 is somebody who might be in the market for a new car. Read Behavioral Targeting 101 for more details on how behavioral targeting works.

1. The problem with “Explicit User Preference” is that it is mostly dependent on choices that a site provides to the visitors. If those choices are not indicative of what users really likes then it will give the wrong indication of visitors’ preferences. Say the site only provides me to choose Red, Green and Yellow colors, while my favorites is Blue then I will choose one of the given colors even though it does not show my true preference. Many times the visitors, in order to get through the registration form quickly, will fill whatever choice they see first. Expecting them to enter their preferences so that you can serve them appropriate Ads is not going to work. Preferences also change with time but visitors don’t go back and change their preferences unless there is a very compelling reason to do so. Getting an appropriate ad is not a very compelling reason. This form of explicit data collection border on the line of PII data and hence can suffer backlash from privacy groups.

2. The problem with “User/Visitor Demographic Data” is that a lot of visitor do not provide correct demographic information. It also suffers a lot of the same issues as “Explicit User Preferences”. User’s Demographic also borders on the line of PII data.

Behavioral Targeting on the other hand does not rely on visitors’ explicit information. As visitors preferences change so will their onsite behavior, say if was looking for a sports car about a year ago and now my situation is different and I am looking for a SUV, it will be reflected by my browsing behavior. I don’t have to go change my explicit preferences that I entered about a year ago (no waste of my time); my behavioral data will show that. Absence of PII makes it even more preferred way of personalization of Ads.

No comments:

Post a Comment

I would like to hear your comments and questions.