homepage Welcome to WebmasterWorld Guest from 54.167.173.250
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member
Home / Forums Index / Google / Google AdWords
Forum Library, Charter, Moderators: buckworks & eWhisper & skibum

Google AdWords Forum

    
Quality Score Observation
Quality Score Changed Within an Hour With Zero Impressions
metakomm




msg:3639795
 8:19 pm on May 1, 2008 (gmt 0)

I created an ad group.

Purple Hairy Widgets = Ad Group Name
Keywords Purchases = purple hairy widgets, exact, phrase, broad
Ad Copy - Headline has Purple Hairy Widgets

Quality Score Great... $0.05

I go back to the ad group and add another ad WITHOUT the keyword in the headline and no where in the adcopy.

Quality Score OK... $0.15

So i guess I diluted the strength of the ad group by creating an ad that doesn't have the keyword in it. Now the ad without the keyword will show 1/2 the time, and the relevancy between the keyword purchased and the ad copy decreased resulting in a decrease in quality...

 

metakomm




msg:3639796
 8:21 pm on May 1, 2008 (gmt 0)

and interesting..i just noticed when i do the reverse, and create an ad without the keyword first, and THEN add the keyword to another ad later, my quality score increases...

Rehan




msg:3639818
 8:59 pm on May 1, 2008 (gmt 0)

If you have multiple active ads for a particular ad group, the QS shown for the keywords is the QS for the ad displayed by default in the AdWords web interface. When you're viewing the list of keywords, the interface shows the most recently created ad (of the active ones) and the QS data to go along with it, so that's why you see the QS increase and decrease in the manner that you did.

But even though you see "Quality Score OK... $0.15", it may be great or terrible for the other ads. That data for each individual ad is stored and used by the system, but it's just not shown in the interface online. When minimum bid and ad rank are calculated in preparation to show your ad, the system uses the quality scores for the combo of the keyword and the specific ad.

For example, if you have 9 evenly-rotated ads that give a min bid of $0.15 for "purple hairy widgets" and then you ad another one that gives you a min bid of $0.05, then you will see "Quality Score Great... $0.05" in the AdWords interface. If you bid $0.10 then the keyword will be shown as Active, but your ads will be shown only 10% of them time because your min bid will be $0.15 90% of the time.

[edited by: Rehan at 9:25 pm (utc) on May 1, 2008]

dertyfern




msg:3640072
 3:55 am on May 2, 2008 (gmt 0)

Rehan, so then would your comments suggest that more ads in an adgroup is best in order to cover all keyword/ad combinations?

wanderer1985




msg:3640127
 6:39 am on May 2, 2008 (gmt 0)

Rehan, you have given a very nice explanation. But can anyone confirm this? Adwords adviser, can you confirm what Rehan has stated here?

Rehan




msg:3640171
 8:05 am on May 2, 2008 (gmt 0)

Sorry, just a correction to what I wrote earlier -- "When you're viewing the list of keywords, the interface shows the most recently created ad (of the active ones) and the QS data to go along with it".

That's not quite correct for situations other than metakomm's. The ad shown when you're viewing the list of keywords is the active ad with the most clicks for the date range you've selected; if there is a tie for the highest number of clicks (including if it's zero, like for metakomm), then the most recent one of the tied ads is displayed. I'm not sure which ad's QS data is shown in non-simple cases, but based on my observations the 2nd and 3rd paragraphs of my last post are at least close to accurate.

dertyfern wrote:
Rehan, so then would your comments suggest that more ads in an adgroup is best in order to cover all keyword/ad combinations?

Hmm...I'm not sure how you'd draw that conclusion. I'd say that what is important is the standard practice of testing different ads to see what works best; keep the winners and get rid of the losers. In terms of keyword/ad combinations, you may find that one keyword has a better QS when matched to a particular ad variation and another keyword in the same ad group has a better QS when matched to a different ad variation (that's hard to determine when you have multiple active ad variations...it's easier if you run each ad variation by itself for a day or two at a time). In that case, you should probably split the ad group into two to improve the keyword/ad matching and optimize quality scores.

[edited by: Rehan at 8:07 am (utc) on May 2, 2008]

wanderer1985




msg:3640186
 8:21 am on May 2, 2008 (gmt 0)

Do you mean that if i change the date range, my Q.S of keywords change? I dont think so.

Rehan




msg:3640279
 11:02 am on May 2, 2008 (gmt 0)

No, that's why I wrote "I'm not sure which ad's QS data is shown in non-simple cases". But if you have multiple active ad variations and you change your date range, then that can change the ad shown at the top when you're viewing your list of keywords. (Assuming my observations are accurate,) The important thing to keep in mind is that the quality scores shown are associated with only one of the ad variations.

[edited by: Rehan at 11:06 am (utc) on May 2, 2008]

mrclark




msg:3640284
 11:07 am on May 2, 2008 (gmt 0)

"If you have multiple active ads for a particular ad group, the QS shown for the keywords is the QS for the ad displayed by default in the AdWords web interface"

I don't agree with this. I think Google take into account EVERY Ad that's in your Ad Group which is ACTIVE and I've done enough experiments to prove this to myself.

If you have 4 ads in your adgroup and 3 have a good CTR - pausing the one with the worst CTR could very well improve your quality score and decrease the minimum bid. Google would take into account the 3 active ads you have and give it an average Quality Score. They don't just look at the featured Ad. All the featured ad is, is the one with the highest CTR thats all.

mrclark




msg:3640287
 11:08 am on May 2, 2008 (gmt 0)

"The important thing to keep in mind is that the quality scores shown are associated with only one of the ad variations"

Again, that's incorrect. Its the average of every Ad that's currently active.

Rehan




msg:3640294
 11:20 am on May 2, 2008 (gmt 0)

If you have 4 ads in your adgroup and 3 have a good CTR - pausing the one with the worst CTR could very well improve your quality score and decrease the minimum bid.

Okay, you're probably right there in terms of the displayed QS. But I still stand behind the 2nd and 3rd paragraphs of my first post above. :)

mrclark




msg:3640306
 11:51 am on May 2, 2008 (gmt 0)

"When minimum bid and ad rank are calculated in preparation to show your ad, the system uses the quality scores for the combo of the keyword and the specific ad"

It doesn't it uses the combo of ALL ACTIVE ads. If it's about to rank [Widgets] with Ad-A it'll still take into account the quality of Ad-B Ad-C and Ad-D.

"For example, if you have 9 evenly-rotated ads that give a min bid of $0.15 for "purple hairy widgets" and then you ad another one that gives you a min bid of $0.05, then you will see "Quality Score Great... $0.05" in the AdWords interface. If you bid $0.10 then the keyword will be shown as Active, but your ads will be shown only 10% of them time because your min bid will be $0.15 90% of the time."

I can't really agree or disagree with this one as I'm not completely sure what you mean.

Either way, a keyword should belong to it's own Ad Group, if it's getting enough impressions.
Each match type should also go in it's own Campaign.

This way, it's easier to determine which ads are working for that single keyword. If you have a bunch of keywords in the same Ad Group it's virtually impossible to keep track on which ads are working well for certain keywords and not for others.

Put them into their own adgroup and test. The QS will improve; the min bid will improve and ultimately the Ad Rank will improve. Remember AdRank = Max CPC * QS

wanderer1985




msg:3640317
 12:10 pm on May 2, 2008 (gmt 0)

If I am not mistaken, the google system considers the particular keyword -ad combination while ranking the ad. It does not take into account the other ads in the adgroup at that instance while trying to get the adrank for that particular search query. It makes sense to consider the following scenario

two ads A and B. One keyword. The Q.S of keyword is calculated depending on both the ads A and B. This Q.S is used to determine the min bid of the keyword. Now when the ad has to be ranked, the consideration would be the combination of ad-keyword-landingpage relevancy. Hence the the ad A might rank higher than B. Correct me if I am wrong

jkwilson78




msg:3640421
 2:25 pm on May 2, 2008 (gmt 0)

"The important thing to keep in mind is that the quality scores shown are associated with only one of the ad variations"

I actually think there is some validity to this based on what I have observed in my own account.

One thing that I do is test different domain names. So I will have two ads in the same adgroup. The ad copy and keywords are exactly the same the only difference is the display and destination URLs of the two ads.

I have had Google "slap" one of the domains in my ads and not the other. If the ad that got slapped had more clicks and thus is the one shown as the default ad, my keyword quality scores drop to "Poor" and all traffic from that adgroup grinds to a halt.

If I pause or delete that ad and wait a few minutes to an hour with the other ad running by itself all of the keywords will go back to Great or OK quality scores.

This happens like clockwork every time.

I have never noticed the reverse happening, where a great quality ad still shows even if another has been labeled as poor quality. So I think it is safe to assume that on some level something considered "poor" quality can over ride everything else regardless of Great or OK quality scores.

Now this is easy to spot because I use different domain names (I do register and setup live sites on all the domains I test; no tricks or anything)

My take away would be that if you are testing multiple ads and suddenly get "Poor" quality scores. See if there is a really low performing ad in your adgroup and delete it and wait a while to see if things go back to normal or even make a slight change to the best ad so you can force the Adwords system to reevaluate your landing page, ad, etc.

mrclark




msg:3640430
 2:51 pm on May 2, 2008 (gmt 0)

"The Q.S of keyword is calculated depending on both the ads A and B. This Q.S is used to determine the min bid of the keyword. Now when the ad has to be ranked, the consideration would be the combination of ad-keyword-landingpage relevancy. Hence the the ad A might rank higher than B. Correct me if I am wrong"

Landing Page relevance doesn't come into play when Google calculates the Ad Position - not one bit. It's only used to determine the Min CPC.

mrclark




msg:3640470
 3:26 pm on May 2, 2008 (gmt 0)

"The important thing to keep in mind is that the quality scores shown are associated with only one of the ad variations"

I disagree but I'll explain why.

A keyword's quality score uses all Ads which are active in the Ad Group and averaged out. This Keyword Quality score is used to determine where it will appear when somebody searches for it.

So the actual Keyword Quality score (for ranking purposes) will be something like:

(keyword + Ad1) + (keyword + Ad2) / 2

The CTR of each Ad and how well it relates to the keyword, is what will be taken into account. So say Ad1 has a great CTR and is very relevant then the first part of the equation above will be good .. but Ad2 might have a poor CTR and not be relevant. As both parts are added together and averaged (divided by 2), the QS for the keyword in question will be OK .. not great ... not bad.

So you see, both Ads are taken into consideration, not just the one that's shown.

What happens when you pause/delete an Ad? IMO it's no longer included into Google's CTR/Keyword Relevant calculation.

Of course there's 2 other things that Google take into account when working out the Ad Position. These are:

# Your account history, which is measured by the CTR of all the ads and keywords in your account
# Other relevance factors

How significant these are, is anybody's guess. Probably only Google knows.

But the point I'm making, is that both ads in this example are used when calculating the keyword's position. So adding/deleting/pausing ads (when split testing) have indirect effects on how well the keyword is going to perform.

Also remember, that AdRank (where it will show) is calculated:

Max Bid x QS. Ultimately, it's the higher you bid that will go a long way as to what position you'll be in.

Rehan




msg:3640500
 4:09 pm on May 2, 2008 (gmt 0)

But the point I'm making, is that both ads in this example are used when calculating the keyword's position.

If that is the case, then why would different ad variations have different positions for the same keyword? I've had cases where I was running two ads and for a particular keyword one ad always appeared in position #2 (when viewed through the Ad Preview Tool) whereas the other ad always appeared in position #4. If the QS in the AdRank=Bid*QS calculation was an average, then the ads would not have different positions.

jkwilson78




msg:3640608
 6:37 pm on May 2, 2008 (gmt 0)

If that is the case, then why would different ad variations have different positions for the same keyword? I've had cases where I was running two ads and for a particular keyword one ad always appeared in position #2 (when viewed through the Ad Preview Tool) whereas the other ad always appeared in position #4. If the QS in the AdRank=Bid*QS calculation was an average, then the ads would not have different positions.

I have to agree with this as well.

Keyword performance can vary drastically based on your ad text so it would not make sense to factor in each relationship each individual ad has with the keyword at the time your ad is being served to a searcher.

And that is the key distinction.

When keyword#1 is searched on and Ad#2 is scheduled to be shown it only makes sense that the quality score with that keyword and that ad are only used to calculate position.

If all other ads are considered in the calculation as well then Google would essentially be penalizing advertisers for split testing new ads since all new ads you test will not be an improvement.

Now how the quality score that is show to us in our accounts is calculated could very well be determined as a hole but based on the observations I detailed in my previous post I'm not convinced.

This is a great discussion by the way.

Rehan




msg:3641730
 3:47 am on May 5, 2008 (gmt 0)

Well, it looks like what I've written earlier in this thread is wrong.

Here's a test I tried today...
Ad Variation #1 - $0.05 min bid when by itself in Ad Group A
Ad Variation #2 - $0.15 min bid when by itself in Ad Group B

When I combined both ad variations in either Ad Group A or Ad Group B, the min bids displayed for the keywords became $0.15 (after an hour or two). So the AdWords web interface didn't show me the min bids for the 'featured' ad variation, and it didn't show an average either...instead, it seemed to be based on the worse of the two ad quality scores. When I paused one of the ad variations in each ad group, the min bids for ad variation #1 eventually went down to $0.05 again.

The second point I had wrong was whether each ad variation's quality score (for min bids) was used separately when preparing to display the ad. Unfortunately, that's not the case. When I had only Ad Variation #1 in Ad Group A and a bid of $0.10, the ad was being displayed. After I added Ad Variation #2 to Ad Group A and the min bid went up to $0.15, neither Ad Variation #1 nor Ad Variation #2 was displayed unless my bids were $0.15 or higher. So adding a 'bad' ad can have a significant effect on the active 'good' ads in the same ad group, at least as far as the minimum bids are concerned.

wanderer1985




msg:3641745
 4:15 am on May 5, 2008 (gmt 0)

it makes sense as the keyword Q.S and hence the min bids depends on the ads. So if the min bid is $0.15 and you bid $0.10 then the keywords will be inactive an hence ur ad wont be shown.

malfeyns




msg:3647546
 10:49 pm on May 11, 2008 (gmt 0)

Thanks for sharing, Rehan. Interesting.

Did Ad #1 have a significantly higher CTR than Ad #2?

What were the differences between the ads otherwise (e.g., keywords in title)?

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