By Ted McConnell
In our last episode we established, that the media business sells audiences for a living, and that a lot can go wrong between the time we define an audience and the time we use an audience definition to purchase media.
This time, we will explore some basics. Why is data important? How does it figure in the big picture of marketing, who uses it, and how they use it? But first, let’s agree on the basics.
To have (audience) and have not
Media companies earn audiences by putting content in places that people can access it, and either charging viewers for access (subscription model), or advertisers for the privilege of presenting a message to the audience (ad model).
The game for media suppliers in the advertising model is to produce valuable audiences.
Audiences that will certainly purchase a profitable item are fantastically valuable, and audiences that will probably never buy an advertised product; not so much.
Globally, the media industry might be as big as eight hundred billion dollars. So, what we have is a global industry worth a little under a trillion dollars that makes its money by selling the attention of audiences. Yet, advertisers do not know when they buy media (or even after that!) whether the audience they bought was worth it. Surely, this is the only eight hundred-billion-dollar industry in which customers can’t know if they got what they paid for – but they try.
Ultimately, of course, it was an unfulfilled need that drove demand. The difficulty in advertising is getting the message about the need to the person who has it.
If getting the message to the consumer need is the quality of reach, then the number of times anyone gets a message, regardless of whether they could have used it, is the quantity. Shockingly, as an industry, we lack a standard, objective quality measure for audiences.
Why Content is King
Knowing who has a need, and when, is the craft of modern marketing. Contacting those people is the craft of the media business. Mainly, intercepting need starts with making content that people will come to when they have that need. With entertainment content, a program is designed to draw a specific demographic.
Any way you slice it, content is bait, and the right kind of bait will bring the right kind of fish. We can measure which kinds of fish go to which kinds of bait, broadly. Women like the Oscars, so if a woman might buy your product, advertise there. Oscar bait, face cream fish.
But, what kind of person wants a hotel in Omaha next week? What demographic is likely to need a dishwasher? The mapping between needs and content as expressed in demographics falls flat for most kinds of products and services. Yet demographics are still the standard way of mapping needs to advertisements.
The New Indicators of Need
Enter, stage left, niche content. Online, we have hotels.com, and toiletseats.com. Visitors to those sites are vastly more likely to need a hotel or toilet seat than a demographically defined audience. The modern advertising business is all about capturing indicative behaviors. If I sell toilet seats, I would love to advertise to people who seem to be looking for one. The use of demographics as a surrogate for actual need works well for products that everyone needs, but things that everyone needs (e.g. toothpaste) are a small portion of our global economy.
For any product or service aimed at a sparse window of need (toilet seats or hotels), it makes sense to be very selective regarding who gets what message. Online, to do that, we create lists of people (identities). Those lists are called “segments”, and in the marketplace, those segments are structured to contain a single attribute, which can be anything from “nursing mothers” to “people who viewed content about cars.” The segments feed buying platforms.
But what if the data was wrong? Did those consumers actually do the behavior used to capture the interest? Does that behavior correctly indicate a need?
Approaches to Audience Validation
To date, there are only a few ways to gauge the quality of an audience.
One is anecdotal, in effect, a narrative. “Women are interested in beauty, and my site is about beauty, so my site has ‘beauty-interested’ visitors.”
Another, is to refer to vast warehouses of information about people; so-called “profiles.” Profiles are populated with claimed interests, behavior data, etc. Profiles have a lot of issues, but the main one for brands is that they are not updated frequently enough nor are they detailed enough to capture a lot of unmet need. No profile has “needs to buy an air conditioner tomorrow” as a variable.
Another possibility goes straight at the problem. When it comes to finding out how a person is feeling or thinking, the best way is to ask them. This is survey-based audience validation.
Another method is cross-validation. If several segments say the same thing about me, there may be a better chance of that being true. This is also fraught with possibility for error.
All the methods except the survey method have significant, endemic validity issues (including bots, bad inferences, cookie deletion, and shared machines) except in specific circumstances.
In our next chapter, we will talk about how survey methodology can be used to yield reliable, timely, and accurate results for campaigns, segments, and publishers.