Demographics Are Still Killing Us

The fact that broad demographics are still the way we evaluate the delivery of media plans is really hampering the online advertising industry. Let's use the hypothetical example of Slurm, the fictitious energy drink that was the subject of a Futurama episode a while back. An agency developing a media plan for Slurm will do a lot of research on the target audience. They'll find that a lot of different types of people drink Slurm, but that the core target is, say, Adults 18-34. They might find that the core target for Slurm is a certain type of young person who likes music, attends concerts, tends to be a young, urban professional, etc. When it comes time to evaluate the delivery of the plan, we're still talking reach, frequency and GRPs against A18-34.

One of the reasons this sucks for online is that people align themselves with sites and communities online according to their interests and lifestyles. As an online media planner, I might look to MTV and VH-1 as potential media venues. But guess what? Lots of people who aren't between the ages of 18 and 34 like MTV and VH-1, too. Since we pay for every impression in online, we end up paying for a lot of people who aren't inside the demographic parameters. I don't think this is such a horrible thing, but when it comes time to evaluate the delivery of the plan, online can look inefficient at reaching the target because it has higher costs to reach 1,000 people in the target demo. (Cost Per Point.)

If I buy a bunch of TV spots in Prime time, (hey, Slurm has a pretty big budget) I pay based on the audience of A18-34 who are watching the networks and shows we purchase. I also get a ton of other people who just happen to be watching the shows, like folks who happen to be 39 and still might be interested in lugging a case of Slurm home from CostCo or whatever. Slurm benefits from this, because while their media is evaluated against the core target, they also get a lot of Slurm drinkers outside the demographic that they don't pay for.

On the online side, a planner might decide to overlay a demographic targeting filter to increase delivery against the core target. This will make Cost Per Points competitive with television, print, radio and other offline media, but it has several nasty side effects:

1. You lose the benefit of the "spill" outside the demographic. Most sites that can target demographically use site registration data to target ads to people occupying specific demographics. For all intents and purposes on the delivery evaluation side, ALL of the impressions you buy end up going to that core demographic.
2. You limit your site selection. There are a lot of sites that would be well-suited to reaching the Slurm drinker in a relevant context, but guess what? They don't have demographic targeting capabilities. When it comes time to eliminate options based on their ability to reach the target efficiently, these sites are selected against.
3. You boost frequency at the expense of reach. Fewer sites to choose from means less reach and more frequency (in general). Further, very few sites have registration data on 100% of the people who visit their site. When a demographic targeting filter is added, the potential pool of people who can receive the message is not only limited by the demographic filter, but also by the total number of people submitting valid registration data. And when you talk to a smaller pool of people, you decrease your reach and increase your frequency.

These are just three of the undesirable side effects. There are more. I won't get into them now.

So, what's to be done? To me, two feasible options exist:

The first is an audience-based currency for online. In other words, we pay for online the same way we pay for TV. If I buy a rotation through Yahoo! Music against A18-34, I pay for delivery against A18-34, but I also get a lot of other folks outside the demographic who happen to be hanging out there. This would have the benefit of increasing the value of Run of Site and other untargeted inventory that doesn't tend to be used on targeted online media plans because of its high cost to reach a specific demographic. It would also help make CPPs more competitive between offline media and online media. Plus, it's easy for clients to understand.

The biggest drawback of this approach, IMHO, is that it abandons what makes online media unique and desirable - the ability to reach people in alignment with their interests and lifestyles - by concentrating on demographics, which are really just a surrogate for the real audience you're trying to reach.

The audience-based approach is nothing new. There's been a lot of talk in the industry over the years, debating a move to this approach. There are a lot of problems with it, including the notion that publishers don't trust measurement companies enough to be able to determine delivery against a demographic. It would open up a huge can of worms with respect to whose numbers to trust.

IMHO, the second option is the better one - Evaluating entire media plans on something other than demographics. Unfortunately, this would take a seismic shift in the entire media industry, which isn't exactly known for its ability to adapt its systems quickly (or at all).

Back to Slurm for a second... As an online guy, I'm in the position of having to alter my plan on MTV and VH-1 because my demographic delivery doesn't look as attractive from an efficiency standpoint as the :30 TV spots. But let's say that it was the TV people who had to adjust their delivery evaluation instead of me. Because people online congregate around interests and lifestyles, online will deliver the psychographic target a lot better than offline. My cost to reach the people I REALLY want to reach is at worst comparable and at best, a lot more efficient. It's only because of the demographic surrogate that television appears to deliver the target more efficiently than online.

Now, since television doesn't have vast reams of data to the extent that online does, it's a huge challenge to be able to sell or evaluate television by delivery against an interest or a psychographic audience. Even if you could develop a model for evaluating it that way, there not a snowball's chance in hell that the networks will sell it to you that way. They'll only guarantee against demographics. Thus, it's television's inflexibility in both measurement and pricing model that preserves its own dominance.

If we could change the way we evaluate the delivery of media plans, interactive media could be the core of many ad campaigns. You wouldn't see 5-15% of media budgets spent online, you'd see 50, 60 or 70 percent.

Then we could get to work on how to avoid shortchanging two-way dialogue, which is easily facilitated online but not in other media. Before we can get there, though, we need to show how interests and lifestyles are better measurement evaluation criteria than age and sex.