1) Is there any ideal number of ads to split test at one time in one ad group? I've heard that just 2 at a time is best, and then when you start to see a trend, delete the worse one and create a new one to try to beat the better one. Is there any reason not to run 10 at once and accumulate data on all of them at once? 10 at once strikes me as more efficient - at least I won't have to go in and make changes as often. Agree? Disagree?
2) How many impressions do you give an ad before you feel confident you can see a trend in terms of CTR?
...and a couple about trademarks:
From reading the information on the Adwords website, I under the policy about bidding on trademarked names to basically be: “Go ahead, but do so at your own risk – the trademark owner could get you in trouble.”
1) Do you have any suggestions regarding trademarks?
2) What about typos on trademarks?
Thanks!
When one of your ads has reached this number, it's time to make a decision.
I remember reading that there was a mathmematical reason why 20 was the best, but not sure why.
What I tend to do is give it to 20 and if I really can't make a decision, give it an extra 10 to see it makes any difference.
[edited by: tedster at 11:43 pm (utc) on April 2, 2008]
In order to find statistically significant results on a test of this nature you must focus on the volume of clicks and the difference in the click through rate. The more clicks and the greater the difference between the CTR, the less time it would take to find results. The less clicks and/or the smaller the difference in CTR, the longer it would take to run this test. It looks like the program PPC_Chris is using does this calculation to determine significance for you.
So to answer these questions - You should look at the impression and click volume of the KW or group of KWs you are testing. Lots of volume means you can find results quicker and/or add more variables to the test without having to wait a year for any answers.
As a general rule of thumb I find anywhere from 1-3 test copies and 1 control copy will work fine. If you begin to see one piece of copy under-performing in the test group you can always pull the piece of copy from the test, which will increase the number of clicks you are seeing as well as the number of clicks being allocated to each test cell.
Regarding your questions related to trademarks -- As long as the owner of the trademark (TM) has not filed with Google you have free-reign of using the TM in your copy. Of course the TM in your copy should have a direct correlation with the content of your landing page and the desired action you would like to occur after the click event in order to ensure that your ad drives relevant, high converting clicks and has a good quality score.
You cant 'get in trouble' for using a trademark but you run the risk of the owner of that TM shutting you down by filing with Google. As a best practice check your ad's status after implementation to make sure that it is live. To mitigate the risk of losing traffic if the TM owner does file, unless you plan on checking up on the ad often, it may help to put a second ad in the ad group that does not contain the TM. (Look to optimize ad rotation here.)
