AdWords
4.3K members online now
4.3K members online now
Improve your AdWords performance and boost your ROI, CTR, and Quality Score
Guide Me
star_border
Reply

Display Ad Experiments

[ Edited ]
Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Hi,

I'm split testing display ads in a retargeting campaign and I have a question regarding the ad experiment.

 

 

Assume I want to test a branding ad vs a product ad. I would like half the visitors from my list to only see branding ads, and the others to see product ads. Will this happen how ad experiments are set up, or will the ads simply rotate among my entire retargeting  pool?

 

Thanks,

 

1 Expert replyverified_user

Re: Display Ad Experiments

Participant ✭ ☆ ☆
# 2
Participant ✭ ☆ ☆
Hello Neil T,

Same Visitors in experiment can see different ads. At least I think this is what you are asking.

It is not like in website pages A/B tests.
** I´am learning Adwords and find it very interesting.
** Ik ben Adwords aan het leren en vind het erg boeiend.

Re: Display Ad Experiments

Top Contributor
# 3
Top Contributor

Hello;
When experimenting with ads you are testing changes in ad-copy ; not  comparing versions of new ads.
On the search network, testing performances of various (new) ad-copies could be done by using "even rotation".


It's harder on the display network, since the there's also a factor of the banner size, that might  trigger one banner ad and not the other; (due to limitations in slots available on the sites - there are a number of  banner sizes that are more popular.) 

-Moshe

Moshe, AdWords Top Contributor , Twitter | Linkedin | Community Profile | Ad-Globe
Did you find any helpful responses or answers to your query? If yes, please mark it as the ‘Best Answer’

Re: Display Ad Experiments

Collaborator ✭ ✭ ✭
# 4
Collaborator ✭ ✭ ✭

You can split the remarketing audience in 2 using the advanced segment "visitor bucket <51 and visitor bucket >50 . You will need 2 ad groups with the same targeting and bid but with 2 different sets of Ads.

Re: Display Ad Experiments

Participant ✭ ☆ ☆
# 5
Participant ✭ ☆ ☆
Hi Moshe,

When you are testing two versions of an Ad, lets say in a 50 - 50 split. It compares how these two ads are difference in the process that starts from going in the auction, shown and clicked.

50% of the control ad goes to the auction
50% of the experiment ad goes to the auction
Every time it can be one of these two ads.

So I think it does not matter if there are new ads. Because the test does not specific use the history. The control works as control from the start date of the experiment.

Can you explain me why you cannot use ACE here? When I read how it is working, I do not see a reason why.
** I´am learning Adwords and find it very interesting.
** Ik ben Adwords aan het leren en vind het erg boeiend.

Re: Display Ad Experiments

Top Contributor
# 6
Top Contributor

@Elja G ;

What I mean is that  for running experiment you need a control group;

The control group is your current ad-text  (running). The goal of the test is to determine whether the new ad-copy will result in a new ad performing (statistically significant ) better compared to the current (running ) ad-text . 

 

The only advantage of doing that using ACE (compared to analysing the data yourself),  is that the system will determine  whether  "better" performing can be backed by reliable statistical  data (i.e statistically significant performing better).

 

-Moshe

Moshe, AdWords Top Contributor , Twitter | Linkedin | Community Profile | Ad-Globe
Did you find any helpful responses or answers to your query? If yes, please mark it as the ‘Best Answer’

Re: Display Ad Experiments

Visitor ✭ ✭ ✭
# 7
Visitor ✭ ✭ ✭
Thanks but I can't figure out how to split my retargeting list into two segments. Would be interesting to see the effect of not serving any retargeting to some people as well.
Marked as Best Answer.
Solution
Accepted by topic author compcc
September 2015

Re: Display Ad Experiments

[ Edited ]
Participant ✭ ☆ ☆
# 8
Participant ✭ ☆ ☆

Hello Neil T,
Adrian is talking about re-marketing groups in Google Analytics.

 

You can  read about this here :
http://support.google.com/analytics/answer/2611820?hl=en

 

Analytics randomly tags users with numerical identifiers from 1-100. You can take advantage of those tags by including the Visitor Bucket metric when you create a list based on a visitor segment.

 

You can use Visitor Bucket to run A/B/N tests of your Remarketing Lists and the associated ads, and test the efficacy of one remarketing campaign against another. For example, you can create two visitor segments to identify any population of users (e.g., Revenue > 100; Transactions > 5; Country/Territory Exactly matching United States), and then use the Visitor Bucket metric to differentiate them:

 

Users tagged 1-50 would be candidates for List A; users tagged 51-100 would be candidates for List B. Use List A in one ad group, and use List B in another, and then analyze which was more effective.

 

analytics-remarketing filter.png

 

 

analytics-remarketing dimension.png

 

 

analytics-visitor-bucket.jpg

 

And they say :
You can use that same methodology to measure how effective your remarketing campaigns are compared to no use of remarketing.  You create a remarketing list to include only a specific set of users and ignore the others. For example, add any criteria to the list, and set Visitor Bucket to Less than 51. In this case, users 1-50 would be candidates for remarketing ads, while users 51-100 would not.

 

But maybe someone else can explain me this last sentence  I do not understand how to make a remarketing list that is not a remarketing list.

** I´am learning Adwords and find it very interesting.
** Ik ben Adwords aan het leren en vind het erg boeiend.