Google does not get the purpose of testing
February 27th, 2009Have yet to really see anyone screaming about this, but then again I have not been looking that hard…
I was in SF this week meeting with both Google and Facebook and as a throwaway Google dropped the bomb that they are no longer allowing multiple domains in the same adgroup. Adwords Blog talks about it here.
What this means to the average affiliate marketer is that you no longer can test which Display URL (DURL) works best for you.
Lets take an extreme made up example of a dating site with the goal of a credit card sign-up.
If your DURL is FreeSex.com then you will get lots and lots of clicks which Google loves and thus your CPC will go way down but your conversion rate will be horrible since the DURL implies free when a sign-up is required.
What about a DURL of FatUglyChicks.com? You probably get very few clicks, meaning your CPC goes up, but someone desperate enough to click on that ad is probably pretty likely to convert since they have been somewhat pre-qualified.
This is nothing more complicated than an algebra problem - does lots of cheap clicks with a bad conversion work better than more expensive clicks with great conversion. As a marketer you really only care about the bottom line here.
This is an extreme example, but for a dating site there are literally millions of different possible combinations for URLs - some of which are naturally more likely to succeed than others.
Google used to allow you to test lots of URLs in a single adgroup and rotate those ads to get equal impression so that you could determine which URL worked best for your metrics.
The new revised version of the DURL policy now requires you to set up different URLs in different adgroups - meaning that even though you might be bidding on the same keyword that Google’s algorithm now decides your quality score and will not evenly rotate them and ultimately end up serving the ad that meets Google’s best interest.
Unfortunately this is short sighted for Google. I have a URL that is awful - no one would think it would win in a test, yet for some reason it resonates well with consumers and qualifies them and induces them to click frequently as well as convert - great combination. Without testing, we never would have found this hidden gem that allows us to pay Google millions of dollars per month. Without testing, Google would have chosen a better clicking DURL and I would have never optimized my business - which ultimately optimizes Google’s business.
Any one else concerned about this change?








