comScore Blog Post On The Google Paid Clicks Issue
comScore, a company that I am on the board of, has published a lengthy blog post about the paid click data that drove down GOOG earlier this week. I think its well worth a read for anyone interested in this issue.
Their conclusion is:
While we do not claim that these concerns are unwarranted, we believe a
careful analysis of our search data does not lend them direct support.
More specifically, the evidence suggests that the softness in Google’s
paid click metrics is primarily a result of Google’s own quality
initiatives that result in a reduction in the number of paid listings
and, therefore, the opportunity for paid clicks to occur.
Comments (Archived):
I thought the report was a little weird because the clicks on my adsense-supported sites have been through the roof since December.I realize it’s not search, but it’s still a decent measurement.
Well, Fred, at least there can be no mention that your interest in comScore is allowing you to game your personal investment strategies with Goog. 🙂
Yeah. I get the data the same time every other customer doesFred
I don’t doubt “there’s a perfectly good explanation for all this,” as the saying goes, but read of the “oops” here is that it further-entrenches the eye-rolls and disclaimers that everyone in interactive media employs when discussing comScore data.Classic quote from audience measurement discussion cited below for context:NewMediaPro 1: “comScore says we have 5mm uniques, but you know…”NewMediaPro 2: “Totally”Nick Denton’s call to openly-publish all site traffic data looks more prescient every day, as do calls for a more openly-audited methodology that salesfolks could site with a straight face. I know it’s complicated, but perception is dang-near the reality in this category.
Nick’s idea is good. But server logs aren’t accurate. You need a user based approach not a server based approach to deal with issues like cookie deletion/churn and cachingIt is popular to roll the eyes when talking about comscore but what if their numbers turn out to be closer to the truth than server logs?Fred
I understand and agree that client-side solutions do a better job, and server logs are inadequate, but is anyone seriously using “server logs” any more? Isn’t that a response from the 90’s, and not reflective of what tools are the real alternatives?My sense of the practical alternatives are (i) Google Analytics (for the frugal) and (ii) Omniture (for the less-fugal). Both of these tools, in my understanding, employ client-side mechanisms (eg, javascript tags) that rely on page loads (not server log writing). The firm for whom I work has relied on both over its lifetime (in addition to Webtrends), and we see discrepancies between clicks (from partners) and visits (on our pages) all the time.Cookies/caching-type ‘internetness’ is like the weather — always shifting and hard to control — but “server logs?” I don’t think we’re still talking about that distincition. I mostly wish everyone would publish what activity they see, and the tools providers could agree on a common set of methodologies. It would save all the distrust that lives among new media professionals, and would have saved comScore from the above mess.
Server logs was the wrong term.I should have said server side methodologiesfred
To me, this reads like Comscore covering their ass: when Google’s revenue numbers come out, if they’re down, then they were right: if they were up, well, now they’re still right.Everybody I know believes Comscore numbers are laughable, but they get attention because they’re the only number that’s public and the media doesn’t know any better.
The reaction out here in the marketplace is that Comscore screwed up and then tried to cover their tracks. I have no idea what the truth is. The only way to know is to wait for the next Google conference call discussion sometime in late April.