Is the Demo Dead? Profile Targeting and the Web

During the second half of the 20th century, the advertising industry grew up. Before the advent of television, it used to consist of simple price, item, and benefits messaging placed in any potentially public forum, from signs to newspapers to live radio broadcasts. Coinciding with the advent of the boob tube, though, advertisers wanted more information about whom they were reaching and what the effects of advertising was on those people.

At first, research was based on households, and that was really as specific as research got. A given number of households listened to a radio station or received a newspaper or tuned in to a television program. Households can be segmented by ZIP codes. To this day, magazine, outdoor, and direct mail advertising most often target customers this way.

Soon, more specific information was desired. Demographics fast became the currency of targeting. Media and the vehicles of which they consisted were selected based on a given reach a specific media or vehicle had against the target demographic. If I want to market a new product, it is standard to determine what the likely demographic of the potential consumer is so that I can then find that consumer “being in the world,” as Heidegger would put it, using certain media where I can then place my messages regarding the new product.

This system has worked pretty well for a couple of decades, so when the Internet debuted as an ad vehicle, it only made sense that segmenting vehicles and targeting advertising would be based on demographics. In fact, the exciting thought was that I could target my message ONLY to a selected demographic because the web site being used as my ad vehicle could collect that information via registering users and then technologically parse out my advertising assets to be sent only to those who had identified themselves as, say, women aged 25 to 49 who had kids in the house.

But what if, in spite of proprietary research conducted and syndicated research studied, half of those who would actually be interested in my new product fell outside of the demographic? What if what I have to market is a lifestyle product that appeals more to a consumer’s state of mind rather than to his or her age?

Sure, products endorsed by the American Association of Retired Persons (AARP) can be targeted based on an age-50-plus demographic, but what about something like, say, snowboards? There are kids 10 years old who want them and guys in their thirties who want them. These are demographics so divergent, I can’t really target BASED on demographics. But perhaps these consumers share a similar state of mind: new, exciting, daring, on edge. Whether you are 6 or 60, if you are interested in my product, then I want to talk to you about it.

Enter profile-based targeting. Imagine if I could read click-stream data of a given user (anonymously, of course) and, based on that data, determine statistically what kind of advertising that user would be most receptive to? Well, it can be done, and it is being done. In fact, something like this has been done for years in one form or another, but now it is gaining some prominence.

For information on how something like this can work, see Tom Hespos’s September 28 ClickZ article. It is a cookie-based system that would use cookies already in circulation from most ad servers. Or the specific advertiser could set its own cookies and begin profiling visitors to its site for its own remarketing efforts.

Back before Disney bought Infoseek and turned it from one of the best search engines into one of the worst, an advertiser could purchase inventory that was targeted to certain “categories” of users, identified by their preference for certain content and responses to certain kinds of advertising. It was called Ultramatch, and it worked very, very well.

DoubleClick is going to start doing this for all its properties and already has this kind of targeting available. DoubleClick’s Sonar Network has profile-based targeting in place, and inventory is available now for advertisers to purchase.

Engage is also selling profile-targeted inventory. And AltaVista, a CMGI property, is using Engage technology to provide profile-based targeting of inventory. There are others, I’m sure, who already have or are rolling out this kind of program for buying inventory, and there are sure to be more. But here are a few points to be aware of before building a buy this way:

• Overtargeting: It is actually possible for the algorithms being used to “create” this kind of inventory to go too far and reduce your possible audience by so much that there is no longer any value in advertising because the audience is too small.
• Lack of volume: As a result of overtargeting, you could end up with an audience that might be able to satisfy some efficiency objective because the advertising is so well targeted, but the audience is too small to generate the volume necessary.
• Media mix: Not everyone makes a purchase or is interested in a brand at first sight. You need to create a multimedia environment that touches the potential consumer in a lot of different ways. Not everyone responds to the same thing, and not everyone is in a position to “Buy now!” Make sure you place your profile buy in the context of a “surround sound” marketing environment so that all the media can reinforce one another and maximize effect and results.