Thrilling Real-Life Tales From the Trenches: Panel Data Vs. Server Data
In a comment on my last contribution to this column, one reader wondered whether, and I'm paraphrasing here, my decision to write these columns with a subjective voice and point of view (that is, the POV of the comScore Chief Research Officer) was appropriate. I thought it was a fair point, but as I indicated in response, this is part-column, part-blog; and as long as my affiliation is sufficiently "disclaimed," best to write what I know. And what I believe. And to let you all keep me honest should that become necessary.
So in that spirit, I thought I'd devote this week's column to excerpts from my diary.
This week a publisher, let's call them "W," wanted to understand why their server-centric Page View (PV) counts were so much higher than their panel-derived PVs. We asked them for a listing of their top urls ranked by PVs as reported by Omniture, and compared that to the top urls receiving credit in the Media Metrix panel.
Well, diary, it turned out that the very top url in Omniture was a 302 redirect to the landing page; of course we credit the landing page, but not the 302. And two other top-5 Omniture urls also represented back-end calls, not actual content served to a visitor, and were thus filtered from our reported data.
Site-centric data counts everything that machines do; but panel data filters out the stuff behind the scenes that never makes it to the person out there in front of the screen.
They were very appreciative of our efforts to reconcile this disparity, and even bought us all a delicious lunch!
Just came back from a big meeting with "S." They've been after us for weeks to help them understand why they consistently saw more traffic from their internal data than in Media Metrix. Seems as if, month in, month out, their internal data showed about 35% more visits than they were seeing in our data. Then it occurred to us, "S" only had access to our data in the US. Well -- wouldn't you know it? When we looked at worldwide data, we found the "missing" 35% of visits. They were coming from outside the US, and of course their internal data was worldwide, and they were lining it up against our US data.
They weren't quite as happy as "W," but they did send over a lovely muffin basket.
"Hollywood" called again. She understands why panel data will report lower Unique Visitor (UV) counts than server data. "I get cookie deletion and all that," she said -- sn't it great when someone just totally "gets" you? "But I need you to come out and explain all this to the big dogs." So we flew out to the Coast and walked them all through the standard panel-verses-server litany; we showed them how cookie deletion inflates reported reach in site-centric data, and we told them all about the IAB Audience Reach Measurement Guidelines. We got boffo reviews. Variety said it all: "Cookies Crumble, Panels Take Cake."
****These are all guaranteed, 100% true stories. OK, so maybe I exaggerate a little in my diary. But "W" really did buy us lunch.