Delivery of basic online campaigns can be evaluated against simple demographic audiences, but we're a long way off from an audience-based currency in online media.<BR>
For the record, let me say that I believe demographic targeting on the web is marginally useful, and that most advertisers would get more mileage out of their online ad campaigns if they learned to target their ads by lifestyles, psychographics and behaviors.
After all, that's how people align themselves when they use the web.
That said, many advertisers like to target their ads demographically. It gives them a common currency for making comparative judgments about online versus print, television and other vehicles. Advertisers might also target advertising demographically for strategic reasons. For example, they might notice different product buying or consumption habits between different demographic groups, and they might want to shore up some marketshare among a select group of people.
A decade ago, my boss at the time, Joe Apprendi, said to me, "Make no mistake, demographics are coming to the web in a big way." I countered that web planners would never move to demographics, since there are so many more compelling ways to target the right audience online. We were both sort of right. We can target ads demographically on many sites that collect registration information or have some sort of external database profile to key to, but we never saw the emergence of the audience-based model that governs offline buying.
For the folks who haven't planned or bought television, the essence of the audience-based model is that everything is guaranteed against a specific demographic. Buyers and sellers use ratings to predict what a given TV spot will deliver in terms of audience. If the spot delivers, that's great. If it falls short against the expected audience, sellers typically give buyers a make-good ad. In this way, demographics form the basis of the currency traded by buyers and sellers.
This stands in stark contrast to how things are done online. These days, if I want to buy against a demographic audience, I can't tell my publishers that I need to buy 100 GRPs against Women 35-49. (Well, I can, but they probably couldn't do much to help me.) I have to instead look at which sites can generate reach against that demographic efficiently, model an effective buy within comScore's or Nielsen's R/F tools, and negotiate with each site over impressions. If for some reason I buy inventory on a site and a significant percentage of their W35-49 population fails to show up while I'm advertising, I'm out of luck, since my ads aren't guaranteed against W35-49.
This begs the question: If we have predictive reach and frequency tools and the ability to demographically target ads, why don't we have an audience-based currency in online media buying?
There are a few things that we need to have in place before the road can be paved toward an audience-based model:
1. Better Tools
Frankly, the predictive reach and frequency tools from comScore and Nielsen are somewhat lacking, IMHO. They need to be able to account for targeted impressions purchased on sites. In comScore's R/F tool, I can see what 1 million impressions purchased on Yahoo! against A18-49 will do for the reach and frequency of my overall plan, but if the demographics get much more complicated than that, I'm in trouble. Add, say, an income qualifier like a household income of $50K or more, and I'm in hand-calculation land.
Additionally, these tools are able to account for impressions within specific sections of sites, or within channels of networks. Let's say, though, that I'm sponsoring the Pregnancy & Parenting section of iVillage for two months. The tools don't have enough data on that specific section of the site to be able to tell the difference between it and the larger section in which it resides. Meanwhile, the demographics might be significantly different. This needs a lot of work.
If the tools get a little better at dealing with complex demos and niche areas of websites, this will help.
2. Something to Post Against
The predictive aspect of online planning against demographics is a lot further along than gauging the actual delivery. If we got to the point at which a site could guarantee ratings against demographics, what would we reconcile delivery with after the ads run? Perhaps comScore and Nielsen will step into the space of providing actual delivery, or maybe publishers will suggest something based on audited server logs. It also might be a business opportunity for the agency-side ad serving companies. There are advantages and disadvantages inherent in these approaches. But right now, we have nothing. If we want an audience-based model to emerge, we need something to post against.
3. Code for the Ad Servers
With the way things are currently bought and booked online, it's easier for a publisher's ad server to be able to serve 1 million impressions within the sports section than 1 million impressions against Men 18-34. Ad servers would need to adjust from a revenue optimization perspective so they can cost-efficiently serve out campaigns against demographics while simultaneously serving campaigns bought under other models. This is an inventory management challenge best addressed by ad management propeller-heads at companies like DoubleClick and 24/7.
It seems to me that we've jogged only halfway down the path of bringing demographics online. For the advertisers that are most comfortable with demographics, finishing up the work we need to do to bring an audience-based currency online might also help bring additional ad dollars to interactive media. To me, this sounds like an initiative that should be right up the alley of the IAB, but it will also require significant effort from the agency, publisher and research communities in order to have a prayer of coming into being.
Tom Hespos is the president of Underscore Marketing and blogs at Hespos.com. Read full bio.