Forum Moderators: martinibuster
Publisher has a site selling quality consumer, prosumer and professional digital cameras alongside adsense advertisments. The adsense ads displayed on said publisher's site are largely comprised of the following:
*Review sites about digital cameras
*Review sites with affiliate ads about digital cameras
*Generic Ads pointing to sites such as Ebay, Target or Megasearch etc.
*The occasional off-target ads about industrial products.
*Shopping sites whos landing page may carry everything from keychains, keychain cameras to banana lounges.
*MFAs with nothing but ads: Consider that MFAs ads may or may not be targeted well or may contain more MFA's, general advertisers etc.
I think that if Google can’t always get their advertiser quality control or targeting right, then samrtpricing really has no way of working effectively as in many cases it doesn’t really know what actually is the desired outcome in many cases.
[edited by: Scurramunga at 11:57 pm (utc) on May 13, 2008]
purplecape:
…so you've given an example of a situation in which the AdSense targeting algorithm is flawed. Does that prove that smartpricing--which may be based on a completely different and perhaps much simpler algorithm--is also flawed?
Obviously, a flaw in one algo doesn't conclusively prove a flaw in another algo, but I'm willing to bet that predicting the likelihood of a conversion is harder to achieve than the task of targeting ads to content.
<Changing The Assumption Slightly>
Working an assumption that smartpricing is superior to other Google algos:
How effective or useful would smartpricing be when Google applies it to our e-commerce digital camera site if it is sending traffic via Adsense ads to websites such as generic shopping websites, fake search engines, unrelated websites or MFAs?
[edited by: Scurramunga at 3:23 am (utc) on May 14, 2008]
Yup, I have seen ..poor content leads to poor quality of ads in many cases.
However sometimes, the MFA publisher is smarter than the algo..
Things like Ho$ and Se%* babes, get rich scams..
It is a mix.
Yes..I have seen premium publishers in my country getting all kinds of MFA's..
And I wonder if premium publishers experience low payouts.
I prefer to focus on what I can do with the situation and to me that usually leads to more or less ignoring the algo.
If indeed the smart pricing algo is flawed, we could learn the nature of that flaw by trial and error. If we know how it works (in general of course, the Gods from the Plex will never give us the details) we can decide how to act on that.
Some sites I have built solely for the fun of it and for the sake of my readers. In that case the algo can take a hike. I might even remove the code from that site if needed.
Some other sites I have built with cash in mind. In that case, the algo can also take a hike. Because if the algo says it will serve the best paying ads, but my filter says banning advertiser X will give me higher ECPM, then I know what to do.
So yes, the algo might be flawed. But in the end I care about the fun or the cash it brings me. Regardless of advertisers conversions.
[edited by: BrandNewDay at 1:35 pm (utc) on May 14, 2008]
Publisher has a site selling quality consumer, prosumer and professional digital cameras alongside adsense advertisments. The adsense ads displayed on said publisher's site are largely comprised of the following:
If the publisher has a site that's selling the same products I am, in addition to running AdSense, wouldn't I, as an advertiser, choose to block my ads from appearing on his site, as he is in competition with me?
Next thought - do we not think that Google is tracking and somehow using the information on which sites and types of sites that advertisers block in a predictive sense, for the smartpricing algorithm?
If the publisher has a site that's selling the same products I am, in addition to running AdSense, wouldn't I, as an advertiser, choose to block my ads from appearing on his site, as he is in competition with me?
No, I love to intrude my ads on sites of competitors. Perhaps I can snatch a piece of their market! I just hope they don't block my ads.
but I'm willing to bet that predicting the likelihood of a conversion is harder to achieve than the task of targeting ads to content.
Well, that's true if you assume that what the smart pricing algo. is doing is making predictions about clicks made on ads on a particular site. But what if it doesn't look at that? What if it's based only a few key characteristics of the publisher's site--type of content, type of traffic, nature of incoming links, etc.?
We don't know. We can only speculate. As others have said, I believe that the best approach, at least for me, is to create and develop a site with the information that my visitors want.
Publisher has a site selling quality consumer, prosumer and professional digital cameras alongside adsense advertisments.
From what I've seen, most e-commerce sites have terrible content. It doesn't surprise me that they get low quality ads.
As iridiax and Green Grass said, those ads might be a result of the site being judged as a lower quality site, so they get a larger percentage of the junk.
I think that the smartpricing algo may be frequently getting things wrong as it is often deciding what a click worth is worth or how much a click should be discounted based on successful conversions for fleas. How indeed would a quality publisher site with quality visitors on the verge of buying, convert successfully for an MFA or a fake search engine? Some publishers do claim to be flea free, but I think that it is fair to say that the fleas make up a substantial proportion of the total inventory.
Ok, you might counter my argument by saying that we do not know how the samartpricing algo works, or that ad targeting and smartpricing are independent. But doesn't smartpricing work on the likelihood of a conversion for advertisers? If the answer to this were "yes" then I would pose the question: which advertisers are being asked to convert for? And with such diverse bussiness models to convert for (eg b/n mfas to totally off topic advertisers) how we have faith in the algo getting it right when targeting and ad placement frequently gets it so wrong?
I am posing a question to an issue that I fully know is beyond our control. It may or may not be something to consider, especially for those of us who know we are attracting quality visitors yet seeing unexplainable periods of low EPC in what should otherwise be heavy demand periods.
You may think we didn't get the point, but I think we got it better than you are assuming.
If your content is hard to classify for ads, then your site is sure to seem "risky" to the algo.
If you get bad ads, it will probably predict your conversion rate according to your unknown (by them) type of site, for whatever type of ads they are serving up. It's probably poor, so smartpricing is appropriate.
If you have a standard e-commerce site, you probably also use a lot of the same stock descriptions as hundreds of other sites. That is certainly a quality indicator for the search algo, so I also suspect that it is a quality indicator for both the quality of ads to serve up and the odds of smartpricing.
Lots of ecommerce sites are also just plain spartan when it comes to the descriptions compared to the size of their templates. Can you think of another niche in adsense that would meet this definition?
When your targeting is off, instead of complaining about the algos, you should be asking yourself what you could do that might improve the results.
As Big Dave said, as I said before, look at site factors, not the impossible-to-predict likelihood of a click converting...
I am simply putting forward the proposition that many a publisher has at one time or will in the future be incorrectly smartpriced, whether they are high risk or not. Further to that, IMO it is factors that are beyond our control that contribute to smartpricing despite our best attempts to control the quality of content appearing on our sites.
Ayway, I wanted to take the discussion on step further and say that if my theory of smartpricing were to hold true, then the large amount fleas and off target ads that appear on publishers' sites in countries (where inventory is weak) would further contribute.
As I said before, what if smartpricing ISN'T determined the way you think it is.that's why I was hoping on alternative points of view.
When your targeting is off, instead of complaining about the algos, you should be asking yourself what you could do that might improve the results.
My thread is not just about targeting alone, besides the targeting in my niche countries is fairly good, so i am NOT complaining, It's other countries where ad inventory is low where my targeting becomes a problem and I am sure I have read that others have observed similar.
As I said before, this isn't a diagnostic thread for me or my site.
Perhaps I should have titled this thread
Who are we supposed to convert for and how? - in light of all the generic and junk advertisers within the inventoryand instead of using the digital camera site example, I should have just come out and stated that I think smartpricing may be flawed because the algo cannot possibly know who some/many publishers should be converting for in light of the fact that so many publishers complain about poor quality, generic or irrelevant ads being served. As with most disscussion on WebmasterWorld there is no evidence to support the claim, but looking at reported incidents of Adsense targeting flaws, Adwords QS not weeding out fleas effectively and various Google filters that botch up in the SERPS every now and then may not nail anything, but certainly adds food for thought IMO I would have thought.
[edited by: Scurramunga at 3:30 am (utc) on May 15, 2008]
