Personalization and Its Effect on Ad Sales

It’s with a sigh of relief that we leave the thorny subject of email marketing and open up a new topic within our larger “defining the Internet” series. But, in true “from the frying pan into the fire” spirit, we have a sneaking suspicion that this new topic profiling and personalization may prove to be equally controversial.

For starters, we’re going to tackle what exactly is meant by the terms “profiling” and “personalization.” Then, we’ll turn our attention to what all this means for ad sales the good (higher CPM potential), the bad (flagrant privacy abuses), and the gray (pretty much everything else). Like email marketing, we can’t promise clean or easy answers, but we’ll share our perspective and provide a forum for incorporating other points of view.

So, let’s get started on definitions. First, we’d like to make you aware of a great site for learning more about this topic We’ve used some of this site’s definitions as a starting point for our own thinking about this topic.

In this week’s column, we’re going to talk about profiling. For all the brouhaha, online profiling is actually a pretty simple concept. Profiling involves collecting customer information on a systematic basis so that this information can be used to enable one-to-one marketing efforts.

There are several methods for collecting customer information. The first way is by asking users for information directly, and they provide it. Registration data, customer surveys, contests, and offering premium services in return for additional information are all common ways that sites directly ask site users to provide information about themselves.

Savvy sites with a strong advertising focus understand that the direct collection of user data is an ongoing process based on a delicate equation of value and trust. They know that the key to direct information requests is to tie information requests to something of commensurate value to the user.

If the site is providing a substantial free service such as email, for example, users generally feel that the site is within its rights to ask for some basic demographic information at the time of registration. They usually don’t feel this way if the site is merely offering a service that implies limited commitment and lower value, such as search capabilities. So, your site’s leverage in asking for and receiving user information is a function of what you provide back.

Timing is also a key consideration. Experience shows that new users will exit your site in droves when faced with the prospect of filling out a lengthy registration form. But they are less likely to be irritated if, 25 user sessions later, your site finds a way to ask for another piece of demographic information that ties into some new offer or asks in the context of “help us meet your needs more effectively.” In this situation, a track record and trust have been established, paving the way for a deeper relationship between the customer and the company.

Another major way to obtain customer information is to infer it based on the user’s behavior on your site or on other sites. We’ll be talking a lot more about this, but, essentially, inference-based information requires a site to be able to track and analyze how a customer actually behaves. Where customers go on the site, how long they stay, what actions they take, where else they go on the web, and usage patterns over a period of time are all examples of trackable behaviors. This type of tracking is implemented through the use of cookies and through specialized software and services that allow sites to make sense of user behaviors and patterns.

Finally, it is possible to build customer databases by purchasing and appending information from other sources. This has been a mainstay of the direct-response business for decades. List owners frequently increase the value of their lists by purchasing and appending information from multiple sources to add demographic and behavioral depth.

It’s not quite as easy to do this online since often a site has only an email address versus actual identifying information, such as name and address. Last year’s uproar when DoubleClick purchased Abacus, an offline direct-response company, was largely about the privacy issues that ensue when rich but largely anonymous online behavior is married to actual identifying information from the offline world.

Next week, we’ll look at what profiling enables companies to do and how it is reshaping online advertising.

Related reading