Comments on: Match Types & Quality Score – The Truth At Last

By: Matthew Mierzejewski

Matthew Mierzejewski — Wed, 22 Dec 2010 18:56:00 +0000

I thought I’d jump into the discussion, since the RKG post was referenced nnWe did find a new AdWords Help Center page, that appeared after our post went live, and I assume this post as well: http://adwords.google.com/support/aw/bin/answer.py?hl=en&lev=+topic&cbid=xro0z0hlwy6k&answer=68095&src=cbnn“Q: Does my choice of match type impact my Quality Score?nA: Match type does not directly impact Quality Score.”nnAlso, as my post title gives away, negative keywords do not have any direct effects on QS either.nCraig, looking forward to any updates from Google on this popular post!

By: Craig Danuloff

Craig Danuloff — Wed, 15 Dec 2010 03:07:00 +0000

Not yet. We’re scheduled to talk later this week. I will post an update as soon as I have it.

By: Jrod

Jrod — Wed, 15 Dec 2010 01:59:00 +0000

Hello, nnI’m very curious, any update from Google?

By: Frederik Trovatten

Frederik Trovatten — Tue, 14 Dec 2010 15:25:00 +0000

Hi Chad,nnHas Google confirmed your statement:n”When the CTR of those queries is lower than that of the identical query, you get an undeserved boost. When the CTR of those queries is higher than that of the identical query, you pay a quality score price.”nnIf yes, then that’s a huge game-change, especially regarding the mining of CTR data etc..

By: Ludvik Hu00f8egh-Krohn

Ludvik Hu00f8egh-Krohn — Sat, 11 Dec 2010 12:13:00 +0000

Hi Chad!nnThis post is definitely one of the top AdWords articles this year! Great work!nnHave you had any success in getting more information from google? Especially whether QS is calculated from exact or phrase match queries!nnLudvik

By: Craig Danuloff

Craig Danuloff — Thu, 25 Nov 2010 17:35:00 +0000

Google and I have missed each other several times over the past week, but we’re scheduled to try to talk again early next week. Hopefully the mystery will be solved soon.

By: Christian Waijers

Christian Waijers — Thu, 25 Nov 2010 15:23:00 +0000

Is there any news about the ‘true’ calculation of QS? I subscribe to Martin’s remarks about how exact (or phrase match, whichever it actually is) as criteria for QS calculation would seriously impact the way the system works and does not make much sense to me.nnDid you get an update from a Google authority on this?nnGreat that you are really trying to push Google into finally providing us with some more facts though!

By: Martin

Martin — Mon, 15 Nov 2010 10:59:00 +0000

Could it be that these ad groups have different ads? Or maybe the same ads but multiple ads each and Google opted for different winners?

By: Martin Roettgerding

Martin Roettgerding — Mon, 15 Nov 2010 10:56:00 +0000

Craig, thanks for your reply. You mention a separate relevance score for a keyword being considered in an auction which is an interesting idea. I would’ve thought that if you’re eligible for an auction, that’s it. Then it all goes into one quality score for the minimum bid and, if needed, the one for the auction itself. Guess I’ll have to mull that over rnrnStill, I think I’m right: If you take CTR from exact matches you get suboptimal ad positions. However, no need to go in circles… I would love to hear what Google says about that.

By: Craig Danuloff

Craig Danuloff — Sat, 13 Nov 2010 17:11:00 +0000

Hm. Hard to explain that. Obvious idea would be landing page issues or relevance, I assume both are ‘OK’? Will try to find out and report back.

By: Craig Danuloff

Craig Danuloff — Sat, 13 Nov 2010 17:11:00 +0000

Hm. Hard to explain that. Obvious idea would be landing page issues or relevance, I assume both are ‘OK’? Will try to find out and report back.

By: Craig Danuloff

Craig Danuloff — Fri, 12 Nov 2010 18:17:00 +0000

Martin – thanks for the detailed explanation of your view. I’m afraid a full reply would be longer than appropriate for comments, but let me try to convey my thoughts briefly.nnI’m 99.99% sure I’m exactly right within the scope and language of the post, based on direct conversations with the most senior and experienced members of the Google Ads Quality Team and management. I assure you these sources, who I’ll name if/when they give me permission but would be recognizable. However, I have some more discussions upcoming with them and I will attempt to get specific confirmation and clarification.nnI think your correct citation of the role of the search query in the determination of relevance is confused with the conversation here on click-through-rate. In other words the query can influence the relevance ‘score’ for that instance of the keyword being considered in an auction, but the quality score used for ad rank and CPC calculation will use the QS from the identical queries. nnYour point here and even the explanation I suggest does bring up some interesting questions and I’ll pursue them. nnI think the other confusing factor (to all of us) is that there isn’t one quality score – as you cite there is visible and the one behind the curtains, and there are many components behind the curtains that are often publicly referred to as quality score. I have a whole chapter in the upcoming book about this very issue. But I do not believe that the root point of this discussion – the fact that identical queries only are used in the calc of quality score – is incorrect. Again, I will verify in as much depth and clarity as possible and report back in the near future.nnThanks for the tweet and comment!nn

By: Martin Roettgerding

Martin Roettgerding — Fri, 12 Nov 2010 17:59:00 +0000

Hey Craig,rnrnyou asked on twitter why I thought you might not be serious with this post. I think all the conclusions would be right, but the premise is wrong: “Quality score is only calculated when the search query is identical to the keyword.”rnrnI find it hard to argue with this premise as there are no sources given. But looking at the consequences you outlined, it seems very wrong.rnrnLet me explain with a more dramatic example. Let’s say your keyword is “dog” (phrase match) and your quality score for that exact search query is through the roof. But for the search query “dog food” your ad performance is very bad, abysmal. Now if you don’t have the keyword “dog food” in your account, Google would have to rely on the great QS for the exact match even for the query “dog food”. For you this means a big, undeserved boost. But what would this mean for Google? The boost might lead to your ad being number one for the phrase “dog food”, even though the ad performs badly. The top spot would be given away to an ad that makes little money for Google. If that spot would’ve been given to another advertiser who actually gets higher CTR’s, Google would earn a lot more.rnrnThe bottom line is this: If your premise were true, Google would be making less money. Or, as Google would put it: the users would get worse search results, the advertisers would be treated unfairly and Google would be unhappy that both users and advertisers are unhappy. Whichever way you put it, this would not be a good way for Google to handle quality scores. That is why I strongly believe that they would never do it this way.rnrnThat Quality Score is only being calculated for the exact match query has often been said about the visible score on the adwords interface. I’ve never heard that in connection with the actual quality score. The official version is that quality score is calculated every time an ad is eligible to be displayed for a search query (see http://adwords.google.com/support/aw/bin/answer.py?hl=en&answer=49174, first sentence). According to the AdWords help pages, one factor in determining quality score is “The relevance of the keyword and the matched ad to the search query” (same page as above). This includes the actual query. If only the exact match query was used for calculation, this would become “The relevance of the keyword and the matched ad to the keyword”.rnrnMy guess is that this post is based on a misunderstanding. Someone on the chain from the actual Google QS engineers over some Google reps to you has mixed up actual QS and ‘visible QS’ (to use a term you coined rnrnSo longrnrnMartin (@bloomarty)

By: Chad Summerhill

Chad Summerhill — Fri, 12 Nov 2010 16:07:00 +0000

Hi Craig,rnrnHow would you explain the following scenario that’s happening in my account?rnrnI have my ad-groups split-out by match type and have for quite sometime now and in order to get AdWords to serve my exact match ad for the exact search query I use negative keywords in my broad match ad-group to force the ad to serve properly and prevent ad-poaching. Here is what Iu2019m seeing in my account for one of these high-volume terms represented as u2018dog foodu2019 below. I only have one high-volume term in each ad-group.rnrnAdGroup = DOG FOOD_ExactrnKeyword = [dog food] rnCTR = 3.96%, QS = 4/10 (this low QS is OK because I get a 54% conversion rate)rnrnAdGroup = DOG FOOD_BroadrnKeyword = dog food rn[-dog food] (exact match negative)rnCTR = 1.16%, QS = 6/10rnrnSo, in my broad match ad-group I have the exact search query set as an exact match negative keyword. Meaning that the broad match term is never matched to the exact search query. I checked my SQR to confirm this. The search query dog food never shows up in the broad match ad-group.rnrnHow are they calculating my QS? And why would my broad match keyword have a higher QS?rn

By: Craig Danuloff

Craig Danuloff — Fri, 12 Nov 2010 13:32:00 +0000

Richard Fergie adds an interesting question and report of Google info below, and points to a very worthwhile thread at RimmKaufman. Pending further investigation, I’ll say that I do not believe that these ‘phrase match’ queries are used in any calculations. I’ve talked about this subject enough times with senior folks on the ads quality team to think the chances are slim that this subtlety was missed – but I’ll check it out.nnAlso, if you read the RimmKaufman post, I still am not convinced that the CTR impact of non-identical queries does not impact lifetime CTR of the account or of display URLS. I should have a definitive answer on that very soon.nnI guess it’s worth disclosing that I’m working on (and have for some time) a book on Quality Score and Google has agreed to cooperate in terms of providing access to ‘people who know’ and some of the top related folks to answer questions and confirm facts. This is why I say with some confidence that I can get definitive answers to these items. They’ve been great about it and claim they want the truth to come out – all this discussion really helps me to ask the right questions and get the full story. – Thanks everyone.