For more than ten years, there have been debates in the Internet measurement space about panel-centric versus server-centric measurement. Recently the IAB released its Audience Reach Measurement Guidelines, a document two years in the making that will go a long way toward clarifying the measurement issues. On the very first page, the IAB document lays out several “foundational principles,” the first of which reads:
“Client-Initiated counting is crucial. These guidelines rely on the central concept that counting should occur on the client side, not the server side, and that counting should occur as close as possible to the final delivery of an ad to the client.”
Audience counting should occur on the client side. I think that pretty handily settles the panel-versus-server debate, at least with respect to audience measurement.
Still, in the past eighteen months or so, some of our clients have expressed to us an interest in what has come to be called “hybrid” measurement. The thinking behind hybrid measurement is that some combination of panel data and server data would enable development of a measurement solution offering “the best of both worlds.”
Panel measurement is best for understanding the behavior of people. Server measurement is best for getting at the behavior of, well, servers.
So we at Comscore sat down and asked ourselves this simple question: Can panel measurement be improved by the introduction of server data? And the answer is: “Often, it can”.
So in June of 2008, we introduced a beaconed component to our Video Metrix service, allowing us to enrich the panel-derived people data with census-level server data providing total stream counts and allowing us to include a measurement of niche audiences. We use the panel to develop the kinds of metrics that only people measurement can generate: person-level duplication across entities, “intensity” metrics (streams per person, minutes per person), audience composition (i.e., demographics), and duplication — the disentanglement of reach from frequency, the fundamental metrics that still drive most media planning and analysis. (Our Video Metrix Reach & Frequency module will be out soon.)
Inevitably, as various “hybrid” offerings come to market, it will become important for the industry to understand the pluses and minuses of each approach and to realize that not all hybrids are created equal. For example, some companies today claim to offer a hybrid solution, without even having a panel. Frankly, I find this to be disingenuous; the thought process that led our marketplace to collectively endorse the concept of hybrid measurement was, quite clearly, the idea of deriving the benefits of panel measurement AND the benefits of server measurement together. Lose the panel, and it is not a panel-server hybrid.
It is easy to talk a good game about models and inference, about making numbers up, sanctioned by the requisite number of PhDs. (No offense, doctor.) But eighty years of audience measurement tells us that, in order to derive valid and reliable audience measurement, you have to MEASURE THE AUDIENCE. And maybe I’m hopelessly old school, but as far as I can see, the audience that advertisers, publishers and marketers are interested in reaching is still comprised of people.
There is no substitute for measuring people.
This is why we at Comscore fundamentally believe in person-centric measurement, and why our renditions of hybrid measurement will be panel-centric hybrids.
Our chairman Gian Fulgoni has encapsulated this philosophy recently with the phrase “Power to the People!” Which is the title of a new Comscore White Paper, one that I’d urge you to read.