| This 54 message thread spans 2 pages: < < 54 ( 1  ) || |
|Changes in Ranking Formula|
Were we notified of this?
| 6:03 pm on Apr 19, 2005 (gmt 0)|
from: [webmasterworld.com ]
AWA SAID :
|The difference is that now, the CTR of the ad copy itself is factored in, instead of it being solely the CTR of the keyword. Which only makes sense, IMO, given that it is the quality of the keyword and the particular ad it brings up that defines relevance, for a given search. |
Was there a notification on this sent to advertisers?
To what extent is it "factored-in"?
Here is my problem. We did not know of this change. We recently (a month and a half ago) changed all the wording of all of ours ads. Under the "old" formula this would not have had an effect on our average CPC unless, over time, the new ads performed less well than the old wording.
But because our new ads don't have a history of CTR then this must be why our CPC went up 20% overnight! We thought we had a bidding war on our hands but it looks like it is the new ranking formula which increased our buy by 20%.
I am certainly glad that AWA mentioned this here some time ago, but we consider emails from Google to be offical announcements.
Why was this fundamental change NOT disclosed to us?
When I called G, i cited these posts and still the rep said AD CTR ws NOT factored in.
I am not trying to pick a fight, just trying to understand. I am also trying NOT to get angry about spending all this money we shouldn't have spent.
If I had been informed about this change I would have been much more cautious in ad rewording.
| 12:51 am on Apr 30, 2005 (gmt 0)|
So basically what you're saying is that now in additon to everything else we have to have keyword spam in our ads too?
While it makes sense on the surface, I can compose an ad that is 100% relevant without using the actual bidded keywords in the ad itself. Am I being penalized for this?
On a related note, can you explain how it is possible that I bid on a new keyword, write a new targeted ad, and then the keyword is disabled with 0 impressions? This is a keyword that others have been bidding on for as long as I know, so it's not like it's a keyword that no one has ever been able to maintain a decent CTR for.
| 1:32 am on Apr 30, 2005 (gmt 0)|
Way back to Robert:
|My big question is this: |
If I change my ad every couple of days to reflect special offers, will I be at a disadvantage?
I change ads obsessively, often 2-3 times a week (offer changes, etc) 100 ads will change. Aside from having to await the review before the ad will again appear on the partner networks, I've noticed no negative effects.
In fact, I just checked and my overall average CPC has gone down about .01 each month since January.
|But remember that it is also the targeting between the ad copy and the keyword that matters. So be sure that you're creating a new ad that it is highly related to the keywords that cause the ad to appear. |
As determined by the algo again, AWA? The only issue I have with that is that about 20% of my campaigns are non-pornographic, but adult themed. And I can tell you, the computer and its algo do not understand how sex works! As it is, I get some strange, let's just say, unrelated queries as it is! It may start getting worse :)
| 12:59 pm on Apr 30, 2005 (gmt 0)|
|On a related note, can you explain how it is possible that I bid on a new keyword, write a new targeted ad, and then the keyword is disabled with 0 impressions? This is a keyword that others have been bidding on for as long as I know, so it's not like it's a keyword that no one has ever been able to maintain a decent CTR for. |
That point, combined with the AW changes in this thread, raises a question about something I have observed this week. That is, in a conceptual/intangible KW I follow, all but one of the many off-topic ads have suddenly disappeared.
To explain, I'll call the KW here as "purple." It is an intangible that you cannot "buy." I use "purple" to provide a similarity to the actual KW in that purple is made by the combination of two other concepts, "yellow" and "blue."
In this KW, there used to be eB*ay ads
Get all your purple today.
All purple for sale today!
There would also be irrelevant ads such as
When you want yellow,
we have what you need.
Blue this and blue that.
We have blue for you.
Now, obviously, the absurd "Buy Purple" ad is ridiculous and impossible because you simply cannot buy such an intangible concept. Yet those kind of eBa*ay ads have appeared, as have other similiar affiliate ads like that.
As well, just because purple is comprised of yellow and blue together, when those colors are by themselves, they are NOT PURPLE at all. In this same way, very off-topic ads have appeared in the KW. They only share that basic background (as yellow does to purple or blue does to purple), but they are completely irrelevant to the KW itself.
But now, I have observed this last week, that all but one of those off-topic ads has disapeared in the KW I watch. (The last remaining of-topic ad has been paying a high CPC to stay higher up in position than all the others, which may explain why it remains.)
I wonder if this new AW system has effectively shut out such off-topic ads.
Does anyone else have any input or ideas about this?
If the new AW changes are behind these new observed results, it does give us something to at least applaud here. As I have been repeatedly saying at WebmasterWorld, it is important to have a way to effectively weed out the off-topic ads cluttering the AW adspace. I wonder if my many requests have been heard? :)
| 11:19 pm on Apr 30, 2005 (gmt 0)|
Evening everyone, I been sitting here listening to this and I think I found a basic summary to the issue without any solutions. if someone has a better summary please post it, since a clean summary helps everyone.
google needs to make money, so they are placing the ad that gernerates the best return for themselves ( that's the summary ).
pros- it makes them money
cons- highest bidder does not get the free branding effect
pro - makes other advertisers design better ads
con - makes the small guy that can not dedicate the resources to misout by using his buying power
net effect is that someone will design software that will optimize the ads.
how does it effect people placing ads... we will have to sit and think out newer ads or find a pro that does this for us.
| 2:00 am on May 1, 2005 (gmt 0)|
Overall I welcome changes like this and will do what it takes to generate the highest level or ROI possible, given the criteria.
I do have one point. The art of writing advertising and compelling as copy does not always go hand in hand with kw/kp/relevancy stuffing.
| 2:44 am on May 1, 2005 (gmt 0)|
The ad is only 25, 35, 35. Keywords that are only relevnt to that... well, it's gonna be a reeeely short list. Wouldn't it make more sense if the keyword list had to be related more to the landing page? I know, that's coming next. I think I would like that better, but I don't make the rules.
btw, where's the answer to Nyet's original question?
|Was there a notification on this sent to advertisers? |
AWA, as good as he is about helping others here, avoided this question. Not only that but then claimed only 2 people cared about the issue. This thread has moved about 3 times now and it's easy to loose participants, but I counted more than that. Plus, I think I can count for my accounting department. That's 7 more people. hows that? ;-)
The change has cost my company THOUSANDS. And I'm not even a big player. I'll bet it cost others LOTS more. Google isn't gonna give me a refund.
Don't take it personal AWA. I've been hoping around reading other threads and I can say that Ive learned quite a lot from you. Thank you _very_ much.
Anyway, the whole thing has made google look quite sinister at our company. I've been told to pause all my ads until google calls us back. That's cool, I guess. I could stand not to spend a few extra thousand on basicly nothing.
| 3:43 am on May 1, 2005 (gmt 0)|
|The art of writing advertising and compelling as copy does not always go hand in hand with kw/kp/relevancy stuffing. |
Agreed 100%. Very short-sighted of G to reward ads just for keywords in the ad, if that's what they are doing. The only thing they should care about is CTR as that clearly defines how relevant users think an ad is.
| 12:06 pm on May 1, 2005 (gmt 0)|
|The art of writing advertising and compelling as copy does not always go hand in hand with kw/kp/relevancy stuffing. |
Agreed 100%. Very short-sighted of G to reward ads just for keywords in the ad, if that's what they are doing.
I am not sure that's what is happening. Read the Msg#33 here I wrote.
Continuing with the same example from my earlier post (msg#33), I am no longer seeing this kind of very KW-stuffed but ridiculously irrelevant ad in the KW that I watch.
Get all your purple today.
All purple for sale today!
The only thing they should care about is CTR as that clearly defines how relevant users think an ad is.
I understand that G$ has put out that untrue propaganda in the past, but it really is not correct. Actually, G$'s CTR-theory has been completely debunked. While it had some good possibility, it is just too flawed to define RELEVANCE.
Up until this new situation, G$ allowed false advertisers to generate (and therefore maintain) high CTR by deceiving users with false ads. Without such false ads, those advertisers would never have generated enough CTR if their ad were required to be true. And they would not even appear in the first place if they were required to be on-topic. So, G$'s old CTR-theory has been utterly disproved by the existence of such false advertisers. Worse, G$ mostly refused to stop it -- until perhaps now.
In the end, off-topic CTR obtained by false advertising proves nothing about RELEVANCE except that users can be deceived. On top of that, adspace that is cluttered with deep-pocket off-topic ads (even if not false advertising) is useless for G$'s users and it forces legitimate on-topic advertisers to unnecessarily pay more in competing with deep-pocket off-topic advertisers. When that adspace becomes too cluttered with so many off-topic ads, the USEFUL on-topic advertisers will no longer want to remain. Why bother? So, when the on-topic sites can't or won't compete with cluttering off-topic ads destroying the adspace, there is definitely a problem -- especially for a SE that wants to offer users RELEVANCE!
Truly, that is a very serious problem. So I am hopeful that maybe this new situation finally puts an end to it. It would be nice to see that G$ has heard me here finally in this matter (not for my sake, but for the sake of legit adspace).
But I don't know if that's what has happened yet. We'll see.
| 2:34 pm on May 1, 2005 (gmt 0)|
I agree, 100%. When I said what I said, I was referring in general to on-topic ads. As far as I'm concerned those stupid "buy maggot larvea at ebay" ads should never be approved in the first place. They know who is doing it and can easily shut those down - if they wanted to.
But Multi, I'm not exactly sure what you're saying. Those keyword stuffed ads you're referring to are done automatically via keyword insertion. On the other hand it sounds like AWA is saying we need to stuff our ads with keywords. Not sure if using dynamic keyword insertion would count for that or not. If it's that easy, soon everyone will do it. But what exactly is your position on this? Sorry, I must not have had enough coffee yet ...
| 1:59 am on May 2, 2005 (gmt 0)|
If the newly announced AW situation is sill preventing the ads like that absurd KW-stuffed eB*ay type ad, then I do not think the dynamic word insertion is the issue in such a case. Perhaps more is at play in determining ON-topic relevance than simple KW alone?
As for my position? I am not sure I know about what specifically your asking for my perspective. Could you please me help and clarify your question for me? Thanks!
| 3:57 pm on May 2, 2005 (gmt 0)|
The positioning/pricing of new ads is effected by what amounts to a predicted CTR of the combination of the ad copy and the keyword that brings it up. This means that 'ad CTR' is effectively different per keyword, and it's a mistake to draw blanket conclusions about an automatic and instant change to a higher CPC if one simply changes ad copy.
I would say that any blanket conclusion would therefore be unfounded. Is it possible that avg CPC could instantly go up? I don't think (but I dunno) answering that would compromise G's secrets.
This change seems to stress CTR over ROI and as such favors G more than the advertiser. (certainly G's right to do so.) Because the vagaries of the language in our industry we want to show for a keyword but only draw clicks if the user determines we are relevant. This means we might actually avoid high CTR wording in favor of more relevant wording.
It is all about balance.
But the fact remains. How ever it is couched, G changed fundamental pricing criteria and I have yet to see where this was disclosed prior.
(again, can anyone help me find it?)
Again (the central point which has yet to be addressed): Had we known the fundamental calculation was going to change we would not have changed 20+ ads in one fell swoop!
We would have experimented a little bit on our most expensive keywords.
Perhaps CPC went up or down but the day after we changed our ads our spend vastly increased for the same words and placement.
| 4:50 pm on May 2, 2005 (gmt 0)|
(This reply is based on AWA's reply on April 30, haven't yet read beyond that).
|I'd like to first say that I have heard and understood all the feedback given on this topic to date. And, more importantly, I have routinely passed it on to the right people here at AdWords. So please rest assured that your voices have been heard. |
AWA, I really appreciate your efforts! Not only in bringing our concerns to the right people, but also explaining most issues that come up to us.
Just wanted to let you know, especially because I tend to post complaints and wishes :) I hope our continuing discussions may help further improve AdWords.
|The positioning/pricing of new ads is effected by what amounts to a predicted CTR of the combination of the ad copy and the keyword that brings it up. This means that 'ad CTR' is effectively different per keyword |
! This is the answer I've been looking for in the thread I started, from which this one is split off. One may have to read it twice or more, but it's very clear once you understand what it says.
|Bottom line, what occurs when you create a new ad depends in part on how well the ad is composed, and particularly on how well it is composed and targeted in relationship to the keyword that brings it up. |
Now it's interesting to know how that prediction is made. Especially: is the performace of the specific account, campaign and adgroup taken into account? Or is the prediction based on Google's overall idea of how ads perform?
It is very well possible (see also the in trial/on hold stuff and higher CPCs) that new ads CTRs are based on some general system wide average. This is nice for average and bad performing campaigns but no good for very well running and fine tuned accounts with high average CTRs.
Very often you want to tweak the best running ads, to see if they can do even better. It would be nice if the CTR of the most similar ad would be taken into account for the predicted CTR of the new one.
That's a better measure than ad copy match. For some types of searches there is no need to repeat the keywords in the ad copy, a special offer can work a lot better. Changing such an ad or adding a similar one should have a similar predicted CTR, even if the words don't match. I understand this ad may proove itself over time, but in the meantime the advertiser probably pays too much based on the lower predicted CTR.
| 5:07 pm on May 2, 2005 (gmt 0)|
|Wouldn't it make more sense if the keyword list had to be related more to the landing page? |
Very good point. The relation keyword - landing page is very important for the final results (=conversions). Especially if a word has several meanings, a lower CTR is not always an indication for bad performance. The "other meaning" of the word may simply be more popular among searchers. Those who are searching for "your" meaning will only be too pleased when they find your lower CTR ad.
| 5:25 pm on May 2, 2005 (gmt 0)|
I ask this in all seriousness and honesty. But why stop there? Can't Google just determine for us the relative keywords, placement and budget?
Isn't that the logical endpoint?
| 6:33 pm on May 2, 2005 (gmt 0)|
I hope not!
| 7:47 pm on May 2, 2005 (gmt 0)|
I read through this thread today and am still confused.
Since the keyword's CTR was already measured - and since the person searching never sees the actual account keyword(s) that gets triggered (they only know what searchword they typed in) - and since they just see the ad copy that comes up - didn't the CTR for that keyword already inherently measure the Ad's copy CTR?
If someone typed in "compare blue widget services" and the triggered keyword was "blue widgets" (in broad match) and the ad that was shown for that keyword was not well targeted for the compare and service angles, didn't the CTR for that keyword reflect that? Meaning it went down cuz they were casting too widely.
And didn't those who had a deeper keyword list, more groups and more time spent developing ads that match their keywords (and not using broad match as much) already benefit with a higher CTR based on the keyword?
What is the compelling need for the change? Seems that a high clicking ad, will now move higher - even if it's not well targeted to the keyword - because we're now adding a new factor for ad ctr and I assume overweighting the ad's performance itself. Like if I ran an ad for the keyword "blue widgets" and my copy said "win a free car" or "win a date with Pamela"... What happened to relevance as king? Sounds like we just juiced the formula to favor clicking, not relevance.
| 8:41 pm on May 2, 2005 (gmt 0)|
This topic and click fraud has got me thinking. What about "reverse click fraud". With a change in CTR factor, your necessary cpc rate (to maintain a certain average position) and cost can change dramatically for a keyword. Lets take a high priced keyword that doesn't receive a ton of traffic. Lets go with one my favorite industry, "mesothelioma lawsuit" 9k searches last month - top cpc $12 (I was going to use "Mesothelioma" but somehow Overture is showing over 1 mil searches last month, wonder what % of that is fraud - what an increase! AND top cpc is down to $27 from $100..).
Now what if someone did this:
A "meso" advertiser lowers his bid to the minimum for "mesothelioma lawsuit". Then ,using proxy servers or something anonymous (so as not to get caught in a fraud filter). This advertiser begins to search for "mesothelioma lawsuit" and begins to click on his own ad. Slowly, but surely driving up his CTR. Then, as he climbs the rankings, he increases his searches and clicks (trying to make things look natural, of course). Finally, if and when necessary, he raises his cpc and gets himself prime real estate at a discount (suspending his "reverse click fraud campaign"). As he sees his CTR fall, he lowers cpc and repeats process... An added bonus, is that while he's increasing his CTR, all of his "opponents" CTRs take a little hit..
| 1:52 am on May 3, 2005 (gmt 0)|
To respond to MultiMan's question, while I believe there may be a decrease in off-topic ads, I'm still seeing them for the keywords I watch. I did see a spike in avg. CPC from Feb. to March but it's come down again since then. Our overall CTR is up. We have made changes to all ads in the last two months but learning about how ad CTR, etc. is now affecting pricing, I'd have to do a little more research to confirm if this has been good or bad for us. Just too much data right now that it's not a simple or straightforward analysis.
| 7:45 am on May 3, 2005 (gmt 0)|
A suggestion for AdWords. If you want to assign an initial CTR to a new ad, why not just give it a temp CTR equal to that of the average of existing ads in that ad group, until it has built up enough history to be given its own CTR.
| 7:30 am on May 4, 2005 (gmt 0)|
Has anyone else noticed that Google like to change the algorithm about 2 weeks in the start of a new quarter?
I thinking these regular algorithm updates should be given a name.
| 10:40 pm on May 5, 2005 (gmt 0)|
|I thinking these regular algorithm updates should be given a name. |
Black Plague? ;-)
| 3:13 am on May 6, 2005 (gmt 0)|
I'd call this one BS.
| 6:25 am on Jun 6, 2005 (gmt 0)|
this is a horrible thing google does. its also currently costing me that top banner which si going unused on 'my' keyword
| 9:56 pm on Jun 6, 2005 (gmt 0)|
The algorithm changes always seem to be about 2 weeks into the start of any new quarter.
| This 54 message thread spans 2 pages: < < 54 ( 1  ) |