AdWords
1.9K members online now
1.9K members online now
Improve your AdWords performance and boost your ROI, CTR, and Quality Score
Guide Me
star_border
Reply

How does the experiment-control split (percentage) in ACE work?

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭
I am just studying the ACE for some potential business use. When I read the document about ACE, I think I get one question about the experiment-control percentage split. In document, it said that this number means every search google will decide which one (experiment or control) will be applied for ads. I think there are 2 possible meaning:1. For every keyword search there is x% possibility to use test or control 2. The keyword in this campaign will be split into 2 groups according to the percentage. I think the first one is correct, right?
I ask that because I could also put some keywords just for experiment or control. If I do that way, does it mean that these keywords will just for one purpose discarding the the percentage setting?
One further question is maybe if I want to create one experiment in one campaign. One subset keywords will only for testing, one subset only for control, and the rest keywords will not be impact. Is it possible?
1 ACCEPTED SOLUTION

Accepted Solutions
Marked as Best Answer.
Solution
Accepted by MosheTLV (Top Contributor)
September 2015

Re: How does the experiment-control split (percentage) in ACE work?

Participant ✭ ✭ ✭
# 2
Participant ✭ ✭ ✭

A keyword (or ad or adgroup or whatever you´re testing) can only be control/experiment OR not in experiment.

If a keyword (let´s say it is) that is in an experiment triggers ads at search quries 100 times and the control / experiment is set up to 20% / 80%, 20 of those times  the control group (and the bid set for that) will be used in the auction. 

View solution in original post

Marked as Best Answer.
Solution
Accepted by MosheTLV (Top Contributor)
September 2015

Re: How does the experiment-control split (percentage) in ACE work?

Participant ✭ ✭ ✭
# 2
Participant ✭ ✭ ✭

A keyword (or ad or adgroup or whatever you´re testing) can only be control/experiment OR not in experiment.

If a keyword (let´s say it is) that is in an experiment triggers ads at search quries 100 times and the control / experiment is set up to 20% / 80%, 20 of those times  the control group (and the bid set for that) will be used in the auction. 

Re: How does the experiment-control split (percentage) in ACE work?

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭

Hi David,

 

Thank you for your response!

I think I understand it mostly. But to confirm it, I want to use one example to make sure.

I want to make the experiment on one campaign.

1. Assuming there are 4 keyword groups in this campaign: A,B,C,D.

2. I set the percentage of experiment is 20%. 

3. The tested thing is max CPC.

4. Group in experiment:

  • Group A is in experiment.
  • Group B is in control.
  • Group C is in both experiment/control.
  • Group D is neither in experiment/control.

4. Result

1) All keywords in Group A will only be applied the tested CPC value.

2) All keywords in Group B will only be applied the original CPC value.

3) The keywords in Group C will use tested CPC value for 20% querying, and originial CPC value for the rest 80%.

4) The keyword in Group D will use the original CPC value and it will not be included in the experiment.

 

Could you confirm that?

 

Re: How does the experiment-control split (percentage) in ACE work?

[ Edited ]
Participant ✭ ✭ ✭
# 4
Participant ✭ ✭ ✭

Hi

 

As far as I understand B and D is the same thing. Or rather: D is not possible (unless the keyword is paused)

 

Apart from that your setup seems correct and making sense