AdWords
5.5K members online now
5.5K members online now
Get started with AdWords - learn the basics to get set up for success
Guide Me
star_border
Reply

Minimum Impressions for Testing 4 Ads?

Follower ✭ ✭ ☆
# 1
Follower ✭ ✭ ☆

Hi. I just started my first campaign and I have an adgroup that I'm using to test 4 different ads. Each ad is getting shown roughly 25% of the time. Right now my adgroup has 50 impressions and 0 clicks.

 

How many impressions should this adgroup have (or each ad have) before I can make a decision on which ad to keep (if any)?

 

Ignoring conversion tracking, what should I be looking for to determine which ad(s) to keep?

2 Expert replyverified_user

Re: Minimum Impressions for Testing 4 Ads?

Top Contributor
# 2
Top Contributor
Hi Alex K,

Fifty impressions is rarely enough to make a decision, but it depends on the response rate. The higher the response rate, the fewer impressions needed to make a decision. 50 impressions in the adgroup means each ad has rec'd only 12 or 13 impressions, That's not enough to make any decisions. Give it a little time to gather data, and never ignore conversion tracking.

Best of Luck!

Pete
petebardo -- Deadhead doing AdWords

Re: Minimum Impressions for Testing 4 Ads?

Follower ✭ ✭ ☆
# 3
Follower ✭ ✭ ☆
Hi Pete, thanks for the reply. So how many impressions do you feel each ad should receive before I can make any decisions?

Re: Minimum Impressions for Testing 4 Ads?

Top Contributor
# 4
Top Contributor

Hi Alex;

The number of samplings (impressions)   should be enough to get a conclusion which is statistically significant.
It's hard to tell without calculating the a statistical sampling. A rule of thumb - I draw conclusions about ad-copies only after a few hundred impressions,

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: Minimum Impressions for Testing 4 Ads?

[ Edited ]
Follower ✭ ✭ ☆
# 5
Follower ✭ ✭ ☆

Hi Moshe, thank you for the insight and the links. I'll use them to learn more about statistics and sampling.

Just to clarify, when you say "a few hundred impressions" is that for each ad or the adgroup in total?

Marked as Best Answer.
Solution
Accepted by topic author Alex K
September 2015

Re: Minimum Impressions for Testing 4 Ads?

Top Contributor
# 6
Top Contributor

@Alex ;

I don't want to get into statistics; so I will try to explain by an intuitive example;

Let's say that you have an ad group with 4 ads;  which rotate evenly;

  • Ad A was triggered 400 times with CTR of 5%
  • Ad B  was triggered 350 times with a CTR of 3 %
  • Ad C was triggered 60 times with a CTR of 5%
  • Ad D was triggered 40 times with a CTR of of 1%;

It is clear that you can draw conclusions about ads A and B, while ads C and D are still questionable. You also cannot say whether ad A preforms better than ad B, since the difference in performance is not statistically significant. However, you can say that  ads A and B  perform statistically significant  better than ads C and D.

 

I hope this somewhat intuitively  illustrates the approach in analyzing ad performances.

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: Minimum Impressions for Testing 4 Ads?

[ Edited ]
Follower ✭ ✭ ☆
# 7
Follower ✭ ✭ ☆

I see what you're saying. Each ad should probably have a couple hundred impressions before any conclusions are made. But looking at the adgroup as a whole can give some insights without all ads necessarily getting the few hundred impressions.

I'll wait until I get some more impressions before making any changes. Thanks.

Re: Minimum Impressions for Testing 4 Ads?

Collaborator ✭ ✭ ✭
# 8
Collaborator ✭ ✭ ✭

Hello Alex K,

 

This tool by Brad Geddes could help you in the future : 

 

http://adalysis.com/

Re: Minimum Impressions for Testing 4 Ads?

Follower ✭ ✭ ☆
# 9
Follower ✭ ✭ ☆
Hi Adrian, that's an interesting link. Thanks for sharing it!

Re: Minimum Impressions for Testing 4 Ads?

Top Contributor
# 10
Top Contributor

Alex/ Adrian;

 To keep my  answer simple without getting into a discussion, which requires a background in statistics, I intentionally avoided (as much as I could) statistics.

 

@ Alex;

If the point has been raised and you do wish to get  into the statistics, the proper  / correct way to do that is using Adwords internal tool "campaign experimentsand  not to rely on external tools.

 

@Adrian:

Have you  tested the service you linked?

They do not explain which statistical  distribution modeling is behind it, or is it just a different UI basically using "campaigns experiments modeling? Does the user need to grant API access?

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’