Web Informant #333, 16 June 2003:

We know where you live




Web site operators are getting better at figuring out where you live and what your interests are, all in the process of better targeting their advertising and content. And while on the surface that sounds like a welcome development, like anything else on the Internet it has a dark side as well. The trick will be how the good and the bad balance out in the coming months ahead.


There are two general technologies at play here, and both have gotten a boost in the past couple of months, thanks to such diverse places as the Wall Street Journal Online and Major League Baseball. First is the art of geolocation, or tracking down Internet users by their Zip codes and other location-specific data. Geolocation isn't something new -- various Internet advertisers have been doing this for years -- but the technological sophistication is increasing and the targeting process is improving, The problem in the past has been a lack of solid data about where particular IP addresses actually originate from. Given that there are over a billion such addresses, you need to collect a great deal of information and be able to sort through it quickly to figure this out. For those of you that don't know or who have been under a rock: every Internet user's computer is assigned a particular four-digit code whenever that person is connected to the Internet. The codes follow a particular pattern based on the nearest router that is used to connect the computer to the Internet. The IP addresses of some of these routers are fairly well known, while others can change quite often and are harder to pin down.


In the past the simple mapping of IP address to a particular user was crude -- early software counted everyone using AOL as living in Virginia, where the company used to have its headquarters. The problem is compounded by the millions of dial-up users that don't have the same IP address from one session to the next, and from the millions of DSL and cable modem users who also switch IP addresses frequently.


But lately the technology has gotten better, and now a variety of companies use geolocation software to do various tasks. One example is streaming content delivery service Akamai, which maps IP addresses to its own database to determine the connection speed of a particular user, according to stories in today's New York Times. This enables Akamai to figure out what content to serve up to a particular user. You wouldn't want to receive that full-screen video over a 14.4 link, as an example. Another case in point is blacking out Web users from watching live baseball games in their home markets. While I question the wisdom of this -- anyone who is desperate to watch a ball game over the Web should be given more, not less encouragement due to the lousy definition of the video stream -- I understand why they have to protect their franchises.


The second and complementary technology has to do with clickstream analysis, or the ability of a site operator to figure out what a particular user is doing on their site and where they are going and what they are interested in. Again, the key here is being able to track a particular IP address of a user as they move about your Web site, and bundling together all those page views into a coherent picture of what each individual is doing. This requires a huge database and the ability to sort through it quickly as well as being able to determine whether a session is still active. The problem is that Web servers don't think about sessions quite the same way that advertisers or content suppliers do. This is one of the reasons for setting up a cookie to keep track of this information, but cookies have their problems too. (I won't get into that here, enough has been written about this over the years.)


The Journal has an interesting twist on this. Each subscriber to the online edition is placed in one of eight categories automatically by associating various pages that they visit on the site with the topical focus of the category. The categories are car buffs, techies, investors, heath nuts, leisure, mutual funds, opinion leaders and travel. An advertiser can design their own category based on custom criteria of the user's clickstream. I think this is the first time that any site has done any kind of partitioning in this fashion. In the past, many site operators have analyzed their page views and placed ads on the most popularly viewed pages. The Journal is breaking new ground here by setting up particular topical areas and selling these areas to advertisers. The topical areas have something to do with the site content, to be sure, but first and foremost it has to do with the interest of the site's visitors, and there is the innovative aspect of the deal.  It is probably a good way to make some money and charge a premium for their advertisers, and I predict that others will soon adopt this way of looking at their sites too.


The combination of the clickstream and geolocation technologies says good things about Internet advertising. According to various sources, Internet ad spending is up in the last quarter, a welcome sign of life in our industry. And with these new technologies, or improvements to older ones, we will see more targeted advertising in the months to come.


So what could be bad about all of this? A few things come to mind. First, the databases could be used to track people down. Although the current technology is used in the aggregate, there is nothing to stop people from drilling down to individual IP addresses as the technology continues to improve. A number of people have begun to use the geolocating software to ferret out fraud or the potential for fraudulent transactions. And of course there are a number of evasive counter-measures that users can employ to hide their IP address if they are so concerned. Second, Internet advertising could become more intrusive as site operators learn more about their visitors and more closely target their ad servers. And finally, the support and implementation costs for these technologies are enormous currently and the monies spent on them might be better used elsewhere, such as in building some more market share and brand awareness for your site.


Entire contents copyright 2003 by David Strom, Inc. 

David Strom, dstrom@cmp.com, +1 (516) 562-7151

Port Washington NY 11050

Web Informant is (r) registered trademark with the

U.S. Patent and Trademark Office. 

ISSN #1524-6353 registered with U.S. Library of Congress


If you'd like to subscribe (issues are sent via email), please send an email to: