Posted April 18, 2008 2:20 pm by with 4 comments

Tweet about this on TwitterShare on LinkedInShare on Google+Share on FacebookBuffer this page

By now you know that the paid click metrics released by comScore came nowhere near Google’s actual growth for Q1 2008. So what happened? Did comScore cry wolf, or is there a legitimate reason for the gaping void between what it predicted and what Google announced?

Before, I share my thoughts, here’s what comScore’s Andrew Lipsman has to say

comScore reported that Google’s U.S. paid clicks in Q1 were up 2% vs. year ago, and down 9% vs. Q4 ’07. During the earnings call, Google noted a 20% increase in aggregate paid clicks vs. year ago and a 4% sequential gain.

Why the discrepancy, you may ask?

As is always the case, we need an apples-to-apple comparison. comScore’s paid click report refers to domestic paid search clicks only, while Google’s “aggregate paid clicks” refers to global search and also includes affiliate site ads (i.e. AdSense). Two fundamental components of Google’s reported number – international clicks and AdSense clicks – are not currently included in the comScore report.

Here’s my take.

If you’re in the business of estimating the number of “apples” sold in a quarter, and the market wants to know the TOTAL number of apples sold worldwide, don’t provide a report that only includes US apple consumption.

comScore is technically correct that its data is not an “apple-to-apple” comparison, but when you’re building your name–and getting the fame–by letting analysts assume the worst about Google’s numbers, shame on you for not providing the real data. Either you had the full data and held back–because it wouldn’t create as much publicity for comScore–or you didn’t have the full data, so you’d better work on how you can get your idea of “apples” matched-up with Google’s.

After all, if Google reports on “worldwide apple sales” then you’d better provide data to track worldwide apple sales. That’s what the market wants. If not, step out of the spotlight and make room for a company that can.

And lastly, shame on us, who took comScore’s incomplete data and, in the absence of anything substantial, assumed the worst from Google.

So, to answer the question. Did comScore cry wolf? Yes, and no. It went something like this…

comScore: “Wolf, wolf, wolf.”

Analysts: “Where? We do don’t see it.”

comScore: “Oh, it’s out there somewhere. Don’t blame me if you assumed it was actually in our village.”

  • ComScore got burned on this, they need to just admit it.

    Jaan Kanellis’s last blog post..20 Answers from Google Search Guru Udi Manber

  • Comscore has been off before. I hate how the WSJ always sites comscore statistics, as they are almost always wrong.’s last blog post..Be Prepared

  • Any data collection must operationally define the terms (or the data is likely to mislead). It is amazing how often this is not done. This is a very public example of something that is a problem much more often. Often the fact that people had different understanding of what the data meant is never understood.

    John Hunter’s last blog post..The Defect Black Market

  • Give me a set of statistics and I’ll prove anything you want. When I’m done I’ll take those same numbers and prove the opposite. Numbers only tell part of the story.

    Steven Bradley’s last blog post..Remarkable Is…