I'm starting out with using some ACE's, but am not 100% sure how to set up my experiment to give me the comparison that I'm looking for. I'll describe my aims below... All opinions appreciated!
I have a "search network only" campaign to which I'm wanting to implement a major restructure and , obviously, it would feel a lot safer to do this as an ACE.
The campaign currently has around 400 keywords grouped into 10 ad groups. The groups were set up a long time ago and not particularly well - there are only loosely related words grouped in each ad group. The good thing is that I now have very comprehensive search query data that will allow me to set up some very tight ad groups based around converting keywords. So the plan is to take the existing keywords from 10 ad groups, plus any extras from the search query data, and restructure them into around 40 ad groups.
So to measure the performance of the 10 old ad groups versus the 40new ad groups, do I simply set them both up in the same campaign, then mark the old AGs as Control and the new AGs as Experiment?
Re: ACE Experiment[ Edited ]
January 2015 - last edited January 2015
One thing that strikes in my mind is, comparing keywords performance is the ultimate goal that you want to accomplish(which result in comparison of ad groups performance), so my opinion would be setting ACE at individual keyword would be better option of comparing the performance. Setting 50% imp share on old keywords and 50% imp share on new keyword.
Let's wait for other community members to give their suggestions on this.
Edited: One more suggestion I forgot to mention,try not to make changes to your best performing keywords(as per goals) in old ad groups, leave them as it is and make changes to rest other keywords.
Re: ACE Experiment
Thanks for your answer and it's a good point you make. Comparison of individual keywords going to be the best way to improve overall performance.
So even if I mark an entire ad groups as "experiment" or "control", do I also need to mark each individual keyword inside each ad group as either "experiment" or "control" in order to compare at the keyword level?
Re: ACE Experiment
That would be a good idea if it was all the same keywords. But now that I think more about it, I'm actually be adding a lot of phrase and exact match keywords to capture clicks previously coming from broad match keywords, so experiments at only the keyword level will not really work.
So is the answer just to mark the old AGs and all the KWs in them as "control only" and then the new AGs and all the KWs in them as "experiment only"?