AdWords
4.5K members online now
4.5K members online now
Dive into advanced features like Remarketing, Flexible Bid Strategies, AdWords Editor, and AdWords Scripts
Guide Me
star_border
Reply

Testing Ads - A constant thang???

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Dear Adwords brothers & sisters,

 

Over a month is it better to split test ads with even rotation for first 2 weeks and then run on optimise for clicks with the winner for the remaining 2 weeks or is better to continually be testing???

 

I would love to hear your views Smiley Happy

 

C.

2 Expert replyverified_user

Re: Testing Ads - A constant thang???

[ Edited ]
Top Contributor
# 2
Top Contributor

For me a strong and clear vote for continuous testing!  Hi Chris, and welcome to the Community.

 

For a start, I think optimising for Clicks is an odd setting.  The only situation I'd want to optimise for clicks would be in a Campaign where my primary objective was only to get people to the site without any worry about whether they actually bought anything or signed up to my newsletter, etc.  In other words, a site that has no measurable conversions.  Since I think it'd be pretty tough to find a site where there is nothing you can measure success by, it's hard to imagine where you'd use this setting.  It's also dangerous, in my opinion; since "clicks" is effectively ignoring conversions, you could have a situation where Google happily does its best to get you as many clicks as possible in a manner that is the worst possible for conversions.

 

Optimising for conversions sounds like the ideal, but I've just never seen it work well.  I've tried it many times but on each occasion I got better results manually and in some cases conversions actually went down.  I have wondered if I didn't give the tests a decent time period but when you're playing with a clients' livelihood it's hard to explain to them that they lost $15,000 last week because you were "trying something out".

 

My primary reason though is that I struggle to resolve the apparent conflict between the concept of their being a "best" Ad for a situation and the well-established good practice of having more than one Ad in a Group.  Although it could be argued that Ad A works better with Keyword A and Ad B with Keyword B, my argument in this situation would be that if there is that marked a difference Ads A & B should be in their own Groups.  And I'd apply the same logic to locations, schedules and (where possible) devices, etc.

 

I guess this all assumes that you're watching the Account like a hawk, which most normal people won't so in fairness I can see where optimising for Conversions would be the best choice for the "casual" advertiser.  But since that in itself suggests a willingness to accept poor performance I tend to go right back to square one...

 

Jon

AdWords Top Contributor Google+ Profile | Partner Profile | AdWords Audits

Re: Testing Ads - A constant thang???

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭

Thanks for the response Jon.

 

 

So if you had an ad group with 3 ads on rotate evenly and one of those ads had the best CPA and the highest CTR when compared to the others, what would you do? Clone the other ads to get them as close as possible to the top performing ad but with one small difference? Reduce the number of ads in the group so the 'best' ad is shown more often? Or something else?

Re: Testing Ads - A constant thang???

Top Contributor Alumni
# 4
Top Contributor Alumni

If Jon will allow me to jump in here?

 

Over a month is it better to split test ads with even rotation for first 2 weeks and then run on optimise for clicks with the winner for the remaining 2 weeks or is better to continually be testing???

 

I would never, or almost never, do this.  Not unless the account I was working was one I'd managed for quite a while--one where I knew the conversion flow extremely well and was sure that this was not a market where the bulk of the conversions came at a particular point in the month. 

 

It's not at all unusual for a market to have a monthly "seasonality" in that buyers might be doing research all month, but the bulk of the purchases (or leads or inquiries or whatever) actually come in one week of the month.

 

The problem with testing ads based on the calendar is that a flat "these two weeks and then those two weeks" might not match the 'pattern' of your conversions. If the bulk of your conversions come in the last week of the month, the data you're gathering by only testing ads in the first two weeks of the month doesn't include the all-important "do people who want to make a purchase find this ad interesting" metric.

 


Theresa
Google AdWords Top Contributor
*~*~*~*~*~*~*~*

Re: Testing Ads - A constant thang???

Visitor ✭ ✭ ✭
# 5
Visitor ✭ ✭ ✭

Hi Theresa,

 

Thanks for the reply.

 

I take your point on a month having peaks and troughs for conversions,

 

But if rotate evenly is to be kept as the default setting all the time how do you maximise the exploitation of a top performing ad?

Re: Testing Ads - A constant thang???

Top Contributor
# 6
Top Contributor

By looking for what makes that a top-performing Ad and creating/editing others to be similar.  Back into the cycle and see what comes out; it may be that after another 6 - 8 weeks that Ad is no longer the top, it's now one of your new ones so now you look at that/those Ads to see what makes them the best and edit/create others to be similar and round and round it goes...

 

The trick, of course, is spotting what makes them top-performers...

 

Jon

 

AdWords Top Contributor Google+ Profile | Partner Profile | AdWords Audits