homepage Welcome to WebmasterWorld Guest from
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member

Visit PubCon.com
Home / Forums Index / Google / Google AdWords
Forum Library, Charter, Moderators: buckworks & eWhisper & skibum

Google AdWords Forum

How Google AdWords treats new ad text
new ads vs. old - unacceptable discrimination

 12:40 am on Feb 12, 2009 (gmt 0)

By now, everybody who uses AdWords has figured how tough is to get new ad perform well "against" the one picked by AdWords' system, regardless how good that ad is.

I believe it was some of the previous posts in this forum pointing to a blog where it was suggested that the testing is done by creating copies of old active ad(s) and new ones, in order to see the real value "assigned" by the system. Otherwise, if you let old ad running and put new one against it, the later will never get a real chance.
Fine, and thanks to the guy(s) that invented and suggested this - based on their real life experience - I agree with their theory.

But, who can keep creating endless copies, and is this really the best way of doing it?

And, what about ads that are really better from human perspective - when compared to those "old" ones? What about CTR?

The weight that has been given to existing ads is just too heavy, just as the total weight being given to ad text in general, compared against the rest of what makes a complete Google AdWords account.

I have numerous examples where "so so" ad has been getting onto #2 with $0.95 bid for months, and yet, no other ad text can do that. Even with $1.25, they're still at #5.

In addition, ads with better CTR make specific keyword's previous QS 7/10 (OK) become 4/10 (poor). One keyword in an ad group. Hey?!

This is not right.

Google AdWords:
Can you please rethink the methodology that has been implemented into current "machine" and make it so we can let your users determine the value of our ads by CTR produced based on their clicks?




 1:51 am on Feb 13, 2009 (gmt 0)

I agree - apples to apples ad copy testing is now almost impossible. New ads will often run in lower positions, so you can't compare CTR and conversions in a true test environment. Even if you modify, delete and re-add your existing ad, it will still get preferential treatment. I think you are now best off creating lots of multiple ads when you start a brand new ad group. After they run for awhile, throw out your underperforming ads and then leave the best ad alone for a long, long time.


 12:43 pm on Feb 13, 2009 (gmt 0)

Unless I've missed the point, why don't you just change the setting in your campaign to rotate the ads rather than choosing the automated Optimum ad strategy?


 2:28 pm on Feb 13, 2009 (gmt 0)

changing your settings will get the ads to be equally served, but does not mean they will be in the same average positions.


 2:59 pm on Feb 13, 2009 (gmt 0)

Not only that, if one is doing SIGNIFICANTLY better, I've found that Google ignores the setting and shows the better performing one more often anyway. I sent in a support issue on that once, and was told that's the way it works.


 3:16 pm on Feb 13, 2009 (gmt 0)

Google is missing the boat for some reason. I have had new copy that beats old copy CTR by 20% with same average positions. However the new copy's cost per click is 25% higher than old copy. I had no choice but to discontinue new copy. Both Google and I are losing out on those 20% extra clicks.

Google's answer is that it is the quality score accounting for the difference. I suggested human review and was told that was not possible.


 4:49 pm on Feb 13, 2009 (gmt 0)

I would like to add to that. If AWA would be kind enough to pass some feedback to the appropriate people in Google.

I'm seeing the same thing with content campaigns.

Old ads seem to be given preference even when new ads have higher CTR (and sometimes higher conversion rate once the visitor has landed). So it's a clear winner vs. clear loser. Yet the loser gets more exposure in the ad auction.

And if I set the ads to rotate evenly, it just drops the overall number of impressions and as a result clicks that I get. It happens because the new ads' QS doesn't give it rank high enough to win bids in the auction against other advertisers' old ads. Setting it to rotate basically cuts the exposure my old ad gets in half, but doesn't make the new ad show more often.

So I'm often forced to continue using old ads. Because the new ones just get 1/10th of the clicks.

Higher CTR, higher conversion rate, but still 1/10th of the volume.


 10:22 pm on Feb 16, 2009 (gmt 0)

I agree, this is a very annoying issue, especially because in the UI, it doesn't say what position each individual ad is in. So you see 2 ads and one has higher CTR and you think, "well that must be the best one", but it was only better because Google was showing it 2 positions higher the whole time.

There is no way to get a fair test to see which ad text truly has the highest CTR.


 3:01 am on Feb 17, 2009 (gmt 0)

especially because in the UI, it doesn't say what position each individual ad is in

I asked this at least two times in "AdWords features".

We need to see average position and average Cost Per Click for each ad text, just like we see it for individual keywords.


 1:47 am on Feb 18, 2009 (gmt 0)

Yikes - this is a bit of a worry for an Adwords novice.

Here I was assuming I could simply test a new ad by adding it to the adgroup.

I think you are now best off creating lots of multiple ads when you start a brand new ad group

Could you please clarify scout? Does this method help address the "old ad bias"?

So the problem is introducing a new ad without a history against an ad with history? In short, it cannot be done (properly, scientifically)?

(OT: so many traps for new players)


 5:55 am on Feb 18, 2009 (gmt 0)

it cannot be done

It's probably that nobody here can say that for sure. What we say is that it's tough and that it takes a heck of time and nerves to get it done - have new ad perform better than the old one(s).

So far, I found that:

- I have no idea for how long I would need to let new ad run until I figure it's worth of keeping or not - in other words - how long should it be let run?
- I have no idea if I should create one, two, or three (I don't think over three is good) ads at the same time, and let them run against one or two of the old ones.

The worst part above all is that I can't think "customer" anymore, but "AdWords" only.

Let's say you're asked to build an ad that should be very appealing to the audience that walks every day through the center of your city, or an ad for some newspaper or magazine, or TV ad.
During the planning and creation process you would think about your ad goals and what your ad should do to the brains of all those people that will come across it - or those that actually should be attracted while the rest should stay away and not spend your valuable time.

Now I go back to my previous paragraph about "worst part". Forget customers, forget CTR, forget everything but the formula that nobody knows, even the engineer at AdWords at this moment - I'm sure he'll fix it once he figures it.

Damned HAL 9000 [en.wikipedia.org].

It's all his fault. ;)


 6:19 am on Feb 18, 2009 (gmt 0)

In addition, ads with better CTR make specific keyword's previous QS 7/10 (OK) become 4/10 (poor). One keyword in an ad group. Hey?!

Personally, I can't recall that happening with any of my keywords and ads, and I do create new ads quite often. For me it's been working just as you're requesting when you say "let your users determine the value of our ads by CTR produced based on their clicks".

I had one case where one ad variation had a CPC of $1.00 with a 10% CTR... During a promo period, I created a new ad variation that started out with about the same CPC, but because it was getting a CTR of 20% the CPC went down to about $0.65 in a few days.

I haven't experienced "old ad bias" either. If I create a new ad variation identical to an existing one, they end up having pretty much the same stats.

Maybe I'm just living in the right part of the Matrix.


 11:02 am on Feb 22, 2009 (gmt 0)

It might be a bit risky- but when I introduce a new ad to compete with the well performing ad- I pause the old ad for a day or 2-- forcing the new one to be displayed. Often I can see at this point that the new ad is poorer (low CTR) etc.


 10:40 pm on Feb 22, 2009 (gmt 0)


that seems to be a good idea. Any ideas about the risks of such an approach?



 7:04 pm on Feb 23, 2009 (gmt 0)


Well risk maybe the wrong word. But it means you are pausing your trusted proven adds and letting a new ad have front page.
1 day is usualy long enought if the new ad is any good for it to get a reasonable score and can compete on equal footing with the previous ad


 11:55 pm on Feb 23, 2009 (gmt 0)

if the new ad is any good for it to get a reasonable score

Have you ever seen a newly created ad in a NEW ad group having QS of i.e. 6 or 7 at 9AM (referring to a single keyword here), and then moving up to 8 or 9 within the day or any other period without other changes applied (like landing page or whatever else)?

On the other hand, you've seen your existing QS 7 going to 4 after you've created new ad, but paused (or left running) the existing one that held 7.

Google has been "predicting" things for long time now. This creates a possibility of having your very new ad "slapped" or "featured" even before it gets its run.

One of the ways to test this would be to stay in premium (yellow) area for whatever period you think is long enough in order to get your NEW ad marked as good or bad. See if your starting QS will improve...

Global Options:
 top home search open messages active posts  

Home / Forums Index / Google / Google AdWords
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved