What Google AdWords Really Changed Last Night
This morning I reported that the “fairer” AdWords quality score changes went live. In that post, I asked you to submit your questions, because I had scheduled a call with Google’s Nick Fox to answer some questions I had over this quality score change. The call is now over and I have some pretty transparent […]
This morning I reported that the “fairer” AdWords quality score changes went live. In that post, I asked you to submit your questions, because I had scheduled a call with Google’s Nick Fox to answer some questions I had over this quality score change. The call is now over and I have some pretty transparent answers from Mr. Fox of Google.
Did you know that the way Google normalized the click-through rate data in the past was to test lower placed ads in higher ad positions? They did, but now they no longer do.
When I asked Nick Fox to give me details of what changed in the CTR normalize calculation, he told me it was two fold. The first was that they are now using fresher data. The second point kind of surprised me.
In the past, Google normalized the CTR calculation by placing a lower ranking ad in a higher ad position and showing that ad to a very small percentage of searchers. They would then gather data and use that data to figure out a what-if scenario, i.e. what-if this ad was in position X, what would be the click-through rate. Nick Fox told me, they never charged for clicks on ads they moved from their real position, and it was only presented to a very small subset of searchers.
Now, Google no longer puts the search ad in a ‘what-if’ ad spot. Instead, they use a statistical formula to normalize the CTR data based on the position of the ad, as to when it is shown to a searcher.
The next question was sent to me by a couple of readers. In short, they asked what was changed in the ad promotion algorithm, what specific attributes and what are the specific quality threshold. As I described earlier, ads no longer are dependent on the ad above it to make it to be promoted. They can by pass a higher ranking ad, that does not qualify to be promoted, and show up in the top paid listing. Nick Fox explained that with the promotion algorithm, CTR and quality is rated much higher than the bid price. In fact, quality is weighted much higher in the promotion algorithm than it is in the ad ranking algorithm. Nick told me that the ad ranking algorithm weighs bid and quality about evenly, while the promotion algorithm weighs quality more heavily then bid.
The next question was does this change only apply to search or also the content network. Nick confirmed that this only applies to search.
Then I received two questions outside of the scope of this specific quality score change, but Nick Fox was kind enough to answer them.
(1) Why would there be a price discrepancy between the exact same keywords, with the exact same landing page, when the URLs are different? Nick explained that often searchers click based on the URL, and they aren’t able to see the landing pages until after they click. Since CTR is a large part of quality, and since your URLs are different, that can play a role as to why the prices are different, assuming everything else is equal.
(2) Is historical CTR for a display URL calculated at the account level or system wide level? Nick Fox said it depends because Google gets as granular as possible. This was part of the real time quality score that will actually take into account the historical CTR of an ad, to the search query generated. If Google does not have that level of granularity, then they will move up to a higher level, such as account level or overall system wide level.
Hopefully that answered your questions, if not, feel free to bug me via Twitter and I’ll try to add more Q&As to this article.
New on Search Engine Land