I talk a lot about online data mining and aggregation. Let’s talk specifically about the techniques online advertisers use to collect and “monetize” user information. First up is “behavioral marketing”, a hot buzzword in marketing since 2004.
Behavioral marketing is marketing to people based on their behavior. Say I sell pizza. I want to market my pizza directly towards people whose behavior shows that they are receptive to, or interested in, my pizza. Offline, it’s not that easy to figure this out: you could mine my grocery store rewards card to find that I like to buy frozen pizza, or you could access Domino’s or Local Pizzeria phone records, but all this information is locked up in “information silos” and not easy to combine. However, online the problem is that there is too much data, not too little. Thus, marketers concentrate on “high value data points”.
The overriding and proven assumption here is that what pages Web site visitors click on and where they go from those pages indicates at least a presumptive interest in buying products related to the topics that they click on. For example, repeat visits to a Web page with reviews of sport utility vehicles, coupled with a cruise to the automotive section of classified ads on a site, clearly indicate at least a curiosity about SUVs.
Now, let us suppose that same visitor is also going to pages where she clicks through to an online book seller to a book about how to help your child adjust to kindergarten. Behavioral targeting specialists may look at this data and start to conclude that the site visitor is looking for an SUV to fit the transportation needs of her growing brood.
Often, this information is not just gleaned from one visit, but repeat visits over time. Perhaps on the first few visits to a newspaper site, most clicks are to articles about SUVs. On the second visit, or maybe the third, the articles are revisited, but the customer also clicks on the automotive ads. It does not take a degree in rocket science (or in marketing, for that matter) to recognize the likelihood the customer is on a likely trajectory from “investigate” to “purchase.”
This behavior, then, is extremely individualized, and marketing is designed to interface with users at the point where their behavior indicates they might be ready to purchase.
(Com scholars will appreciate this piece comparing mass advertising to the magic bullet theory & behavioral marketing to uses & gratifications theory.)
Here’s an illustration of how this works:
1. Users visit websites
2. Their site visits are tracked and aggregated across the sites using tracking cookies
3. Users are shuffled into psychographic/demographic groups based on behavior (like “hip mamas” or “car enthusiasts under 40″)
4. Users see different ads based on their demographic group.
This method is a perfect fit for Web2.0 companies (via Charlene Li):
A case in point: digg.com produces no content of their own but has a very unique way to look into the interests of its users. Kevin showed a very cool software tool they use internally called “Trace” that looks at the stories a specific user is reading, and shows in real time how that user’s attention jumps to other topics. Kevin also showed how “diggers” were related to each other based on the stories they mutually “dugg”. The traditional “audience management” advocates like Tacoda have shifted toward behavioral targeting, but at the core, understanding users at a highly granular level will be an essential skill for media companies.
Awesome! So keep this in mind next time you put together a kick-ass, totally personalized, super Web2.0 experience for yourself.