Time to give up on demographics

Demographics were essentially the first tool developed for eliminating advertising waste. That was decades ago, when advertisers discovered demographic correlations such as the notion that purchasers of consumer packaged goods tended to be female. Broadcast outlets began to guarantee their reach against demographic audiences using ratings, and broadcast media's audience-based model hasn't changed much since.

What has changed is the rest of the media business and its approach to targeting, particularly digital media. We can focus our media laser on behavioral, interest-based, lifestyle-based and other varying criteria, and the guarantees for our online buys are against whatever criteria we choose. In other words, if we contract for 400,000 impressions against beekeepers in the state of Oregon, and we get 350,000, we're owed 50,000 impressions against Oregon beekeepers or its cash equivalent.

No such luck with television. Asking for an audience guarantee that's more granular than age/gender demographics is usually a no-no. This week, I was thinking about how inefficient that is.

To break it down further, let's look at how poor demographics are at zeroing in on the right prospect. We tend to think of specific media entities as having specific audiences locked up, or at least performing very well against them.

For instance, we think of MTV as having the youth market sewed up. Would it surprise you that the latest comScore numbers (December 2007) show only 36.8 percent of MTV.com's audience falls into the 18 to 34 range? It surprised me.

Let's look at some other sites that are famous for their association with specific demos, shall we?

  • ESPN.com is practically synonymous with the male demographic. But guess what? More than a quarter (26.4 percent) of the site's unique visitors are female.
  • Think AARP.org is visited only by old fogeys? Think again. Less than half (46.4 percent) of the site's visitors fall into the 55+ category. And only 15 percent are 65+. (Before anyone brings it up in comments, yes, I'm aware of AARP's efforts to bring folks into the fold at a younger age.)
  •  iVillage.com is a site for women, right? Well, it is. But it happens to reach a lot of guys, too. The site is just over 56 percent female.

My point here is not to say any of these sites are deficient in some way. They're not. When you break apart the composition of just about any media vehicle, you'll find surprises. The point is to demonstrate how crummy age and gender demos are as a method for targeting advertising. In short, it matters little what your age and gender happen to be. It's more important to focus on your interests, lifestyles and behaviors if I want to get a relevant ad message out to you.

The ability to focus like a laser on the most qualified prospects for a product or service -- the "sweet spot" of the target audience -- has been a big selling point for digital media for more than a decade. While digital has consistently made strides to help advertisers get to the right people at the right time, broadcast media have failed to do the same. This has resulted in a rather frustrating double standard, where broadcast media still perform heavy-lifting chores on many integrated campaigns, while digital media supplement the effect of the broadcast.

Something's gotta give. Broadcast can't continue to stagnate with respect to its targetability, while digital media have resisted putting anything resembling an audience-based guarantee in place.

Maybe there's a compromise. There is value in an audience-based guarantee system for digital media. However, I don't think it should necessarily be built on demographics. Rather, guarantees could be based on ratings against behaviorally and contextually defined audiences.

In turn, it would be nice to see broadcast adopt more granular audiences as the basis for guaranteeing delivery. The demographic of 18-34 just doesn't cut it anymore. Broadcast needs to layer on the filtering criteria so that its delivery can more closely approximate what we're able to accomplish in digital. With the advent of digital TV, acquiring the data necessary to do this should be a much easier job than it might have been a decade ago.

I think we've demonstrated, though, that demographics are a crummy indicator of audience receptivity to a specific ad message. So why are we married to it when we evaluate media delivery?

Tom Hespos is the president of Underscore Marketing and blogs at Hespos.com. Read full bio.

 

Comments

Gene Cameron
Gene Cameron January 21, 2008 at 4:52 PM

Tom,

I could not agree more! In the automotive space clearly see that demographics are not indicative of model purchase. If you would like to have some examples. Just let me know. You can see some of what we are doing at www.onlineautomotivereview.com

Gene

Ingrid Michelsen
Ingrid Michelsen January 21, 2008 at 11:58 AM

For Pharma and certain CPG products demographic targeting really does need to remain a targeting option, but for the rest behavioral, interest-based and 'lifestyle' targeting makes much more sense. In fact, at WhitePages.com we recently ran a female-targeted CPG campaign that was under-performing until we opened it up to both genders (keeping other targeting in place)- suddenly performance exploded. The agency was genuinely surprised that men were also interested in their advertiser's cleaning product. This is the kind of thinking that keeps demographic targeting in the marketplace, even if it ultimately hinders campaign success.

Langston Sessoms
Langston Sessoms January 21, 2008 at 2:39 AM

I think saying to give up on demographics is a wrong. Lets not pretend like behavorial targeting doesn't have its own flaws as well. Demographics data and users behaviors need to work together for a deeper understand of the audience that's being targeted.

Site demogrpahics informatin should be open like what Quantcast is trying to do. Ultimately I think companies like comscore and nielsen hard the niche sites who get left out in the cold because third party panelists don't visit them enough but tons of other people do.

Bruce Gerth
Bruce Gerth January 17, 2008 at 7:20 PM

Butter, toilet paper and any cpg can be targeted if its value proposition is well defined. Is the butter organic? Go for a Green sustainable crowd. Is is heavy in cream or specialized in measured packaging? Target the cooking/gourmet crowd. Looking at a product in a generic way will lead to marketing it in that manner. Lifestyle characteristics cross demographics boundaries, but once analyzed and identified zero in on the target audience. Identify those specific lifestyle traits of your audience and put your product in their path.

Tom Hespos
Tom Hespos January 17, 2008 at 3:31 PM

More sites are cutting deals with Nielsen. When the intersection of the comScore and IRI panels gets deeper, you'll be able to do the types of things I'm describing web-wide (or something close to it). So I'm not sure that "only a few sites can offer real planning features for cpg" will hold true much longer.

I don't think it will be much longer before we can request inventory against various types of category or brand users, and have sites fulfill on it accurately. When that comes to pass, I doubt I'll be doing much "gimme women 25-44."

Stephan Noller
Stephan Noller January 17, 2008 at 3:13 PM

To sum it up i would say: Time to start with demographics. Of course in combination with online specific capabilities like the one you mentioned. But i don't think it would be a good idea not to develop the demographics approach any further.

Stephan Noller
Stephan Noller January 17, 2008 at 3:11 PM

Tom, i agree - what Yahoo and others together with Nielsen and Comscore established is exactly the direction it has to evolve. But in my opinion online is only at the beginning of this - in fact only a few websites can offer real planning features for cpg and the approaches with market research data are not more than a first try. It is necessary to have targeting technologies that go beyond simple click behavior - not only for cpg by the way.

Tom Hespos
Tom Hespos January 17, 2008 at 3:05 PM

Kevin - thanks for your kind words.

Stephan - actually, there's plenty of behavioral data for CPG, and the notion that behavioral targeting or segmentation can't work for CPG is a long-held belief that's just not accurate anymore. Look at what Yahoo! did with Consumer Direct, and the subsequent targeting deals between other media properties and Nielsen and IRI. The industry is evolving as you described, since now I can go after heavy/medium/light category or specific product users with a reasonable degree of accuracy.

Stephan Noller
Stephan Noller January 17, 2008 at 2:56 PM

Nice idea Tom, but unfortunately your approach does not work very well for a lot of products, especially cpg. In fact i would argue exactly the other way round: Online is lacking of planning capabilities for those products - a tremendous amont of money that could be spent only but isn't because of that. There is no behavior for butter, toilet paper or detergent. The only industry has to develop and implement targeting technologies that provide advertisers with more sophisticated demographic targeting and product interests that are not just collected by tracking peoples website clicks. Otherwise those budgets will keep loving print and tv ads...

Kevin Doohan
Kevin Doohan January 17, 2008 at 9:39 AM

"It's more important to focus on your interests, lifestyles and behaviors if I want to get a relevant ad message out to you." I completely, completely, completely agree. Nice piece.