AdWords is now Google Ads. Our new name reflects the full range of advertising options we offer across Search, Display, YouTube, and more. Learn more

Ads
1.6K members online now
1.6K members online now
Improve your Google Ads performance and boost your ROI, CTR, and Quality Score
Guide Me
star_border
Reply

problems with my A/B test for Ads

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Hi All

 

I am running a quite simple A/B test in a Campaign. There are a number of Ad Groups under the campaign - each of these Ad Groups have exactly 2 adds (its 2 types of adds, with some adjustment for the KW in the ad group). 

My problem is that across the Campaign one (type) of the ads consistently gets more impressions and clicks.

 

My settings are:

Equal CPC for all KW

Bid strategy (on Campaign level) is "Focus on clicks, manual maximum CPC bidding"

Ad Rotation is "Rotate indefinitely: Show lower performing ads more evenly with higher performing ads, and do not optimize"

 

I hope someone can tell me how to get a true A/B test so I can analyze.

 

Thanks

 

2 Expert replyverified_user

Re: problems with my A/B test for Ads

Rising Star
# 2
Rising Star
Hi Rasmus P,

I couldn't quite understand '2 types of ads'. Are you testing one ad text against the other?

Sumanth
Sumanth Sridhar

Re: problems with my A/B test for Ads

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭
Hi

Exactly. I have a 'champion' and created a new campaign (in a different account) to test a new ad against the benchmark.

Re: problems with my A/B test for Ads

Top Contributor
# 4
Top Contributor
I too havbe found this - despite setting the ads to rotate evenly, Google and I clearly have different concepts of "evenly".

The simplest solution if you want to run a true A/B test is to split the ads and then use scheduling to run one ad at a time - switch them between days so that the same time slot is not always running the same ad and see if this helps get more balanced numbers.

The problem seems to be that Google take the ad copy into account when determining the QS which in turn can improve the ad rank of one ad above the other - despite that fact that you want them to run evenly, the system will generally prefer to serve one over the other....

Re: problems with my A/B test for Ads

Rising Star
# 5
Rising Star

I haven't experienced this of the few times I've run a test. However, I go by stickleback's words here.

 

@stickleback: Howe about having two identical ad groups with different ads (one in each), then set them to equal halves for control and experiment?

 

Just a thought!

Sumanth Sridhar

Re: problems with my A/B test for Ads

Follower ✭ ✭ ✭
# 6
Follower ✭ ✭ ✭
Hi Rasmus

Let me see if I understood the situation correctly.

You have a campaign with many adgroups. You would like to do an A/B test of ad copy, and for this you (wisely Smiley Happy) created just 2 versions of the same ad (with variations in the wording).

Then you took these 2 ads and included them in different adgroups (I'm assuming each adgroup got your 2 ads and NO other ad ran while the test ads ran)
Am I correct so far?
If so, and If one of the ads gets more impressions and clicks than its "sister" version, shouldn't IT be declared the best ad (on average)? After all, that is what A/B testing is about.

Or am I getting confused somewhere?

Cheers

Jorge

Re: problems with my A/B test for Ads

Top Contributor
# 7
Top Contributor
Not quite... at least in my opinion and experience.

An ad might have the better ad rank and, therefore, get more clicks and impressions - it may even have a better CTR all of which would lead you to suggest that it is the "better" ad. But what if the conversion rate for the "worse" ad were double?

Because of the way in which calls to action might work and the pure phrasing of the ads, the ad with fewer clicks might be the one which happens to resonate with your best prospects.

When we try and run any A/B test we have to decide what the benchmark is going to be.

The OP frustration was that Google was not really splitting the traffic evenly between two ads as the setting suggest they should.

Re: problems with my A/B test for Ads

Visitor ✭ ✭ ✭
# 8
Visitor ✭ ✭ ✭
@Jorge I will agree with Stickelback. What I am trying to do is get an even split and then make my own decision. This decision can be based on conversions not visable to Google. If i completely trusted google I would just set it to optimize for conversions or clicks.
In addition then the one set of adds have gotten so few impressions and clicks that no conclusion can be made.

For now I have paused the one set of ads that were getting all the impressions. Its a work around, but not a solution. So if anyone have a solution I will be happy to hear.

Best

Re: problems with my A/B test for Ads

Rising Star
# 9
Rising Star
Thanks Steve Smiley Happy
Sumanth Sridhar