AdWords
2.5K members online now
2.5K members online now
Improve your AdWords performance and boost your ROI, CTR, and Quality Score
Guide Me
star_border
Reply

Experiment impressions unlogical

Participant ✭ ✭ ✭
# 1
Participant ✭ ✭ ✭

Hi

 

Sometimes when I run AdWords Experiments (ACE), and the bid is decreased in the Experiment (as compared to Control), I see that the impressions for Experiment is actually higher than for Control.

As in attachached image.

 

However, considering a) the bid is lower, b) the experiment should alternate between Experiment and Control, I wouldn´t expect there to be much of a difference with regard to impression volume, or if there were it would be higher volume for Control, which has higher bid and thus gets higher AdRank and can be shown more often.

 

Can anyone explain why this happens to me please?

 

(for non-swedish reader: Visn. = Impre.)

 

ace-impressions.jpg

2 Expert replyverified_user

Re: Experiment impressions unlogical

Top Contributor
# 2
Top Contributor
David;
Hmm... Something with the setting of the experiment between testing and control.
Statistically this doesn't make sense....
Have you called Google asking to look into the setting?
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: Experiment impressions unlogical

Participant ✭ ✭ ✭
# 3
Participant ✭ ✭ ✭
Hi

No I think it´s more probable I get a swift and correct reply here : )

I´ve seen this before, so it´s not only just this experiment or campaign

Re: Experiment impressions unlogical

Participant ✭ ✭ ✭
# 4
Participant ✭ ✭ ✭
If it would be possible to get an Experiment expert to look at this, it would be really great

Re: Experiment impressions unlogical

Community Manager
# 5
Community Manager
Hi David,

Please look out for an email from me so I can get some more details from you and look into this in your account!

Thanks,
Cassie

Re: Experiment impressions unlogical

Top Contributor
# 6
Top Contributor

@David_Precisera;

Once resolved it would be interesting to know the answer

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: Experiment impressions unlogical

Participant ✭ ✭ ✭
# 7
Participant ✭ ✭ ✭
Hi
@MosheTLV

the answer from the engineers is basically that if the experiment has lower bid (which it has) it may not be qualified for the same auctions, but only for less competitive auctions, in which it will be easier for it to win an impression, easier than it is for the control with higher bid. Thus the experiment can get more impressions.

Makes sense to a degree... But I still don´t understand why the higher bid would give fewer impressions, because it can enter and win all auctions that the lower bid can, AS WELL as many other auctions where a higher bids is required.
And with rotation set to 50/50 the higher bid should compete in all those auctions, not only the more competitive ones.
As far as I understand, but obviously I´m missing a piece here

Re: Experiment impressions unlogical

[ Edited ]
Community Manager
# 8
Community Manager

Hi David,

I believe I mentioned this when I emailed you the solution, but with the 50/50 split, the experiment vs. control do not enter the same auctions.

Hopefully that clears up any confusion!
Cassie

Re: Experiment impressions unlogical

Participant ✭ ✭ ✭
# 9
Participant ✭ ✭ ✭
Yeah, that was very kind of you, the email was explanatory, but I guess I just can´t understand really ow this works.
I mean, I can even see this impression inequality for keywords that are exact matched.