The two major companies that focus on online advertising are taking two different strategies towards winning the market, according to the Wall Street Journal (Monday, November 8, 1999), "Online Ad Titans Bet Big Race to Trace Consumers' Web Tracks). Both strategies are based on an assumption of what predicts your online purchase behavior.
Doubleclick believes it can figure out what you'll buy online based on what you've bought before in both the online and off-line world. CMGI (mainly through its investment in Engage Communications) thinks they can better predict your purchase behavior by knowing where you spend your time online. Let's face it – this is a huge battle for big bucks that essentially relies most heavily on consumer behavior, and that is squarely a marketing issue.
Of course, since these are basically technology companies, the battle seems more to be based on whose technology will work. As Rich LeFurgy (chairman of the Internet Advertising Bureau) says in the WSJ article, "Doubleclick has a tremendous head start, and the best predictor of purchasing behavior is past purchasing behavior. But CMGI has stunned the dot-com world with their vision and their ability to integrate a bunch of different technologies."
The business model for these companies (as it relates to online advertising) involves deciding what ad to put in front of a person when you're browsing on the web. Currently the basis for that decision is such things as time of day, the web site they're looking at, etc. But they both look to change all this.
Doubleclick will focus on marrying different databases (a big problem) to get both online and off-line purchase behavior. CMGI will use cookies and page transitions to get their information. Privacy and technology issues still remain.
But let's assume all technical and privacy issues disappear. Do either of these companies have it right? That remains to be seen. For years we have had customer data on buying behavior (using scanner data in grocery stores) but its ability to predict customer behavior is questionable.
Even if we assume that knowledge of past behavior might be useful for understanding future behavior (and that probably also varies heavily across different product categories), will any knowledge of off-line behavior predict online behavior, a la Doubleclick? On the other hand, most people may just search for information but then then buy in the offline world (see the article by Stewart Alsop for a convincing view of this).
In either case, there doesn't seem to be anybody really concerned about consumer behavior here, but lots of assumptions are flying around. But it will be a better understanding of consumer behavior that would help people predict the winner in the end.