Everyone talks about the precise measurement that comes with Internet advertising. It has become the perfect contrast for all the problems with measuring television. But what are they really talking about, and why can't we have it with TV? Certainly they're not talking about Internet display advertising. You have two competing panel-based currencies whose audience estimates are wildly different, server logs from clients that produce another very different estimate, and finally direct measurement providers with yet another very different estimate. How many people saw that banner ad? Pick a number, any number.
The case for superior Internet measurability started when banner ads were tracked, and often bought/sold, based on click-throughs. Click-throughs are "perfectly" measurable. But it didn't take long to figure out Internet display advertising is really about impressions, not clicks, so various competing services and methodologies were developed to estimate the delivery of impressions. About that same time, however, a new form of Internet advertising was developing, a little thing called "search." Search is now held up as the shining example of the world's most perfectly measurable advertising medium: you know exactly how many prospects Search delivers to your site, you know exactly what you paid for them (and can adjust that in near real time), you know exactly what they did (and bought) when they came to your site (or returned in the future) – all the elements for a perfectly calculable ROI.
Amazingly, in all the discussions about the perfect measurability of Search, I have never once heard anybody point out what's obviously MISSING -- age/sex demographics. Nobody asks Google or Yahoo or Bing, "what demo's does your portal deliver?" Can you imagine bidding for keywords based on age/sex? ("I'll pay $2.25 per click for M18-34, but only $1.50 per click for M50+"). Of course not. It doesn't matter. What matters is the value of attracting an interested prospect to come to your site, and measuring exactly how many of these prospects you generate, and how many sales are eventually created.
Why is almost every other media, except Internet, so singularly focused on age/sex demo's? I believe it can largely be explained by the evolution of the measurement techniques themselves. In the early days of television (and radio, and print), the only way to attempt to determine the audience for a commercial message was to ask a group of people to self-report what they watched/listened to/read. Due to shear economic and logistical forces, this group of people was a relatively small sample of the entire population. Over time some of the self-reporting was replaced with electronic measurement, but the size and nature of the sample populations remained more or less the same. A rich field of statistical science burgeoned on topics such as selection bias, cooperation bias, weighting, smoothing, etc.. While very smart statisticians and researchers worked hard to create the best projections they could, the fact remains that the projections were, and are, projections, and subject to significant variation based on myriad methodological decisions.
You had a situation where there was great (believed) certainty in the composition of the specific panel itself, whereas the projections of total audience were less certain. Human nature being what it is, no surprise the interpretation and use of the data gravitated toward the most known element. We believed with much confidence that 18-34 year old men behaved in a certain way, because a group of a couple thousand recruited 18-34 men reportedly behaved that way (if they correctly cooperated with the research demands). Less certain was how many 18-34 year old men, or total individuals, or total households, across the country consumed a piece of commercial content, let alone how many took action with the brand.
We are witnessing a time when digital technology, in the form of set-top boxes, may finally be ready to transform the TV measurement paradigm into the Search paradigm. Set-top boxes can free us from the constraints of a panel. Like Internet Search, smartly used collections of set-top box data can provide high certainty on the precise quantity of an advertising message consumed. Innovative providers are taking this a step further and marrying precise advertising consumption with other behaviors, such as cross-media behavior (example: TiVo-Quantcast) or product purchase (example: TRA).
Yet there are still those whose first response to set-top box data is, "But you don't know the demographics." Right. And neither does Search. If age/sex demographics are really that important, they can be projected. Given the choice between 1) certainty of demographics, but projected total audience; versus 2) certainty of total audience (and connections to other behavior), but projected demographics, which is more appealing? The success of Search tells us the answer.
Todd Juenger is VP and General Manager, Audience Research and Measurement for TiVo, Inc.
Read all Todd’s MediaBizBlogger commentaries at InteracTiVoty - MediaBizBlogger.