I am sure you have all seen the latest on ad rotation here:


http://adwords.blogspot.com/2012/04/new-changes-to-ad-rotation.html


We are getting a number of questions on what advertisers should do and how our platform will handle the change. I want to discuss both as there are some things that can affect your account with this new setting that could negatively impact your performance.


How will BoostCTR handle the change?


As of right now we have a few options. The amount of information from Google is limited at this point so I will outline a few ideas. Once we figure out the details we will implement one of these (or a new idea as we learn) to handle the new rules. These suggestions are mainly meant to suggest what we think we will be doing and are mostly automated in nature. Implementing them by hand is going to be difficult for all but the smallest of accounts BUT we are looking into releasing them into our system for use by all users to help them run their own tests. We currently offer a free version of our software that does not include our writer network so in the near future we may be able to automate this for you. Stay tuned for that in the next couple of weeks.


  1. Add a new ad to the ad group every 29 days and delete it immediately.
  2. Use param tags to modify the ads using a space-no space flip flop every 29 days
  3. Create an identical copy of one of your ads every 29 days
  4. Modify a destination url of an ad slightly using a query string parameter that is put there for this purpose


Each of these would simply serve to trick the system into resetting the clock so that actual ads in the process of being tested can continue to run. There is another point about how multiple ads running simultaneously can result in a higher CTR than just one ad that I will make in a moment.


The idea with these approaches is that we would force even rotation and it would allow us to find a better ad where it takes more than 30 days to decide. In our own testing we find that a majority of testing takes more than 30 days because of volume issues. Head ad groups and many trunk ad groups can decide much faster but the majority of ad groups take more time. This is why I find the time based approach that Google is taking very interesting. Why would you use time as your metric when we all know that click volume is what is being used to find “winners”? Seems they want to make sure that things don’t linger in even rotation perhaps for smaller advertisers that are not actively managing their accounts. What makes that strange is that optimize rotation has been the default for years so perhaps this is targeted at larger advertisers that run on even rotation with a large number of very bad ads.


When we look in our system we see a number of large advertisers that are running too many ads in a single ad group where they can achieve statistical significance with more than 30 days of data . As an example we have an e-tailer that has MANY ad groups that have 10+ ads in them from template based deployments over the years. They keep adding more but never seem to delete effectively which is one of the reasons they work with us. Since they run with this many ads today, forcing them to optimize after 30 days will probably result in a higher overall ad group CTR. This is probably the major use case that Google is trying to remedy. The issue we have with this is that now the advertiser gives up effective testing so that improperly managed ad groups will yield better results. Seems that with a little active ad management we could have the best of both worlds.


In response to the issue with this and other advertisers we have just released a new tool internally that looks at all ads over a long time period and tells us which ads need to be paused. This accomplishes the same thing that Google is attempting because it eliminates poor ads yet it leaves the ability to test properly. Why Google did not release a tool to manage instead of forcing rules that benefit the lazy is beyond me. We are currently running a version of a tool to find poor performing ads thathave been running a very long time for our Enterprise clients. This tool let’s you look at campaigns to show you where you need to pause ads based on historical data. It compares dates where two ads ran at the same time so it will filter out times when data is not appropriate for ad comparison. It is pretty cool and has some more things in the works. Now back to another reason even rotation is important.


Why would you want to run more than one ad evenly?


Imagine you are an e-tailer and you offer free shipping and/or 10% off on the same day and perhaps this is your angle and you do it every day. Now imagine someone searching for a product and they are in the middle to late stages of the buying funnel. They see your ad on their first query and it says 10% off. They then search a few more times and now your free shipping ad shows up. They now click through because that message resonated really well with them whereas 10% off did not. This scenario is one where you can imagine that your ad group CTR is now 50%(saw you twice and clicked once) as opposed to 0%(just saw the first ad). In this example, the sum of its parts adds up to more than the whole. By having multiple ads rotate you are able to present multiple messages to see which ones work well. Without even rotation you will not be able to support this type of ad setup beyond 30 days. What this really points out is that multiple ads may always be better but without even rotation all the time you don’t even have the option of working this way. Ever wonder why Google always tells you to run multiple ads at once? This is part of the reason.


Another use case that Google apparently does not care about is landing page testing. Most advertisers that don’t use a third party testing tool(offsite page redirection) use even rotation with two identical ads but different destination URLs to test landing pages. Now you only have 30 days to decide or you have to implement a suggestion above on how to reset the clock.


What should I do?


Depends. If you are a smaller advertiser that rarely writes then do nothing. If you are a large advertiser and currently use optimize to clicks or conversions you can also do nothing but realize that you are leaving money on the table by not ad testing. Moreover,any testing you do conduct will not be optimal since Google decides on the which ad to serve much too quickly when using the optimize settings. If you are an advertiser that uses even rotation for a number of reasons start preparing to be able to handle this change through automated means. Living with the change may permanently affect your performance and only through active ad management can you overcome this difficulty.


-Rob Lenderman
Founder BoostCTR