Adwords split testing question

[dgrumbli]

In the last week I began testing ads using split testing. I am using the same keywords for both ads but have slightly altered the copy of the ads. I have noticed something that I don't understand, In one of my campaigns I am getting about 5 times the impressions on one ad vs the other. The CTR seems to be about the same for both ad groups.

Does anyone know why I would be getting such differences in the number of impressions for the two ad groups? From what I have read I was expecting the impressions to be roughly equal.

[bcombs]

There is a setting in the AdWords control system that allows you to choose whether the serving of the two ads should be optimized (for Google's profitability) or split evenly. The default is to maximize clicks.

Interestingly, even when you set it to serve the ads evenly, it doesn't always do so. This makes A-B testing a bit of a challenge.



Brian

[AdWordsRep]

Hi all,

Just stopped by for a moment, and spotted what seems to be a misunderstanding here.

It seems to me that dgrumbli is 'split testing' by using similar ads and the same keywords in two different Ad Groups.

On the other hand, bcombs is referring to a system that can be applied when advertising with two (or more) ads under one set of keywords within a single Ad Group.

dgrumbli, I have only a few minutes just now, and can't write this in detail - and I hope that others will jump in as well. However, when you create two ads in two Ad Groups, with the same keywords in each Ad Group, only one ad may show. So you are essentially competing with yourself for impressions, but not really able to control which ad shows. This is not ideal.

If your aim is split testing ad copy, it is probably a better idea to create multiple ads within the same Ad Group, and then use the feature to which bcombs referred to decide how they are to be rotated.

More info can be found in the AdWords Help Center, as for example on this page:

Can I create multiple ads for a set of keywords?
https://adwords.google.com/support/bin/answer.py?answer=6128

OK, gotta run. Hope this helps. ;)

AWR

[dgrumbli]

Thank you AWR and bcombs, Initially I was trying to run the ads in two different groups. After I combined them I was still having about an 80/20 split in the ads event with equal CTR. I looked and I do have the account set up to optimize so I will change that and try the testing again.

[eWhisper]

I find it's the simplest things I overlook, so just a friendly reminder.

The date range is sticky, so if you look at a date range where one ad wasn't showing, the percentage shown will be off - you need to consider a date range when both ads were being shown to see the real numbers.

[Mel66]

Also make sure to turn off Content Targeting when you are testing creatives. Unless something has changed (correct me if I'm wrong AWR), due to the setup of the content network, Google can't rotate your creatives equally in the content network like they can in search. If you turn off auto-optimization, and turn off content, you should find that your impressions are nearly equal.

HTH

Melissa

[dgrumbli]

Thanks to all, I have split testing working great. Turning off auto-optimization was the key however I had previously turned off Content Targeting due to the poor conversion rates.