After Google’s earnings call last week, Google’s stock jumped on strong results and comScore’s took a temporary hit. Many analysts and pundits were using comScore’s estimates of flattening paid-click advertising growth at Google to conclude that Google’s overall revenue growth would also be flat. Yet Google reported 42 percent year-over-year revenue growth, and 7 percent quarter-over-quarter revenue growth. In contrast, comscore had estimated 1.8 percent growth in U.S. paid clicks for the quarter.
Since Google gets a huge chunk of its revenues from search ads, there should be some correlation between a slowdown in paid-clicks and a slowdown in overall revenues. Yet as we first noted last February, historically the correlation between comScore’s U.S. paid-click estimates and Google’s actual quarterly results has been hit or miss. This time was no different. During the earnings conference call, CFO George Reyes addresses the paid-clicks issue directly by saying:
Now let’s look at aggregate paid clicks growth. Aggregate paid clicks include clicks related to ads served on Google properties, as well as ads served on our partner sites. Aggregate paid clicks grew approximately 20% over Q1 2007 and approximately 4% over Q4.
So why the disparity? It turns out many analysts took the comScore data and ran with it, using it incorrectly, as comScore itself points out. As the table above shows, comScore only measures paid clicks on Google search ads in the U.S., which it says accounts for no more than 15 to 30 percent of Google’s total reported clicks. It does not measure global paid clicks (yet). And that’s missing half the story since last quarter 51 percent of Google’s total revenues came from international markets. The comScore figure also doesn’t take into account clicks on AdSense ads that appear on other Websites, or even search ads on AOL and Ask (which are also powered by Google). Nor does it measure clicks on ads on other Google properties such as YouTube or Gmail.
So the comScore figures are an indicator of what is going on with ads on Google search proper, and you can’t really go beyond that. Comscore goes through some math to come up with what it considers a better comparison between its estimates and Google’s’s reported numbers. If you add in comScore’s estimates for AOL and Ask, that would have brought its paid click growth estimate up from 1.8 percent to 5.4 percent. Then, backing out an estimated 18 percent rise in cost-per-click pricing, comScore infers that Google’s actual U.S. paid click growth in the quarter was 10 percent, which narrows the gap considerably. But it is still not perfect.
The lesson here is you cannot take one data point and use that to predict Google’s revenues.
I would argue that google did well in Q1 because when the comscore report came out, Google began shading the background of their high CPC adwords which will draw extra attention to them and increase their CTRs. All the top 3 positions on any keyword that cost greater than a $1 received the yellow background treatment. Simple as that.
the lesson here is, that analysts are redundant.
You can not predict progress I think. Google is the web search leader and its revenue is diversified. I don’t think that anyone can correctly estimate what Google chart will show. When Yahoo ads and search division will be handovered to Google the we will see more big rise in revenue and shares.
There is no current way to predict web trends on a global scale.
Only Google has that information, and they’re not going to share it citing privacy.
The best way to be predictive in respect to Google is to watch for sudden changes in their business. They actually have the data from analytics, and other javascript people have put on their websites from the world over.
They will know about a drop off months before any of us will with comscore data and will take corrective action.
The best way to predict Google is to watch what they do. 3rd party data will never be accurate or demonstrative.
Rubicon project and others have the best strategy. Piggy back Google, adbright and other javascript code under the guise of optimization then get people to distribute it in a wrapper. Then you can go ahead and collect massive amounts of accurate data. A website that does multiple vendor advertisement optimization could effectively steal millions in web trend data from the different vendors.
The problem is getting people to sign on and trust that.
That would be a good project to invest in.
interesting — I wonder when google will begin doing more ad tactics around their video stuff on youtube.
The article correctly points out that Comscore clicks don’t predict Google revenue, but Google also reports the number of paid clicks on Google properties, and Comscore doesn’t do well there either.
not related to this article.
but yahoo mail beta has a photo slideshow feature now. just wanted to let you know if you have not noticed.
comscore also misses the traffic from dell/gateway and others where google has a partnership for start page
Yet another company showing that the days as the US as the #1 in online are over.
I had better like google . I did my own search to find whatever I need. You can see it in my own site .
So, this brings up an important question: Is Techcrunch going to stop using Comscore and compete.com when covering websites? Because every time I see a compete/comscore graph on TC comparing two or more sites, I cringe. It’s totally unfair to any company who is focusing on the world as a whole. It’s true that until around 2004, a website had to win US traffic to succeed in general, but it’s no longer the case (look at hi5/friendster/any number of other sites which are quite successful despite a lack of US traffic).
The problem is not ComScore’s methodology so much as the data they have, which is a very small and unrepresentative sample of Google’s worldwide audience. The best data to predict Google earnings is with large-scale SEM optimizers like Efficient Frontier. This article shows how Efficient Frontier data has consistently outperformed comScore data in predicting Google earnings:
http://anand.ty...data-beats.html
Why do people pay attention to Wall Street analysts again? Because they are almost always wrong? Because they are incapable of or are unwilling to understand the methodology used to collect the data they spend 90 hours a week analyzing?
First off, the graphic in the image above was taken from the ClixMarketing blog and no credit was given:
http://www.clix...he-differences/
Second, I agree 100% with the statement that “you cannot take one data point and use that to predict Google’s revenues”. However, it seemed pretty apparent that an earnings surprise was in order, especially given the research report from AdGooroo which showed that the trend in advertiser count reversed itself over the first quarter.
http://www.adgo...tiser_share.php
The reason this was apparent was because there was clearly a drop in advertisers as a result of the changes in AdWords policy last July. Many advertisers migrated to Yahoo and MSN as a result. The big gamble here was whether or not Google would recover, and the AdGooroo data showed they did. It’s not a far jump to imagine revenues would be significantly higher, especially in light of the growth in international and contextual advertising.
Google also decided to reduce the clickable area of the ads and to increase the bid threshold for some low-priced keywords — giving up some low-value clicks and messing up everyone’s clicks/dollar ratios.
We haven’t seen anyone predicting convincingly what the global search market share for Google is. Using the US market share alone to predict the global market share can mislead. There is also a more interesting life outside the US!
So why did TechCrunch never question nor investigate Comscore’s methodology. They weren’t even in the same ball park with their estimates yet had falsely led everyone in the industry to believe that their traffic monitoring/analysis was accurate.
The real lesson here is that you cant trust anyone to tell the truth because everyone has an agenda and profit motive…
Google should thank ComScore for the stock jump because it’s not all about financial numbers, 50% (or more) is managing expectation (psychological impact)
when do people start thinking out of Google? don’t get biased Erick.
Scott – the image is from the comscore blog post.
Michael,
You’re right and I stand corrected. My apologies.
Scott