Forum Moderators: martinibuster
I came up with a theory that smart pricing was not really all that smart and tried to figure out what the bot does to determine your "smart pricing"
I think that there is an algo that ties in your poor performing pages and uses them to weight your site for worthiness.
The more pages you track the lower your quality (to the bot).
On the 2nd I deleted all custom channels and saw a 400% jump in income. Went through the 4th of July weekend with better than average earnings. Daily income going up by 2 to 8 dollars a day.
Ctr and ecpm are doing better but slowly, with ecpm rocking down and up.
Payout per click went up from 3 to 9 cents and had 5 .19 cent clicks on search...unheard of in the past. (my average payout per click was 8 cents for my niche)
Noticed better ads began showing up that had not been there before and no crap ads that I could find.
Things slowed down and started rocking so I thought, "hummm, mr bot has found another way". So I started going though my site page by page and changing out the channel ads for regular ones and adding them to pages that did not have them...more page views...:), and have seen the ctr and epcm renew their upward creep.
This weekend was higher than the weekends I've been having for several months now.
For me it is working, I still have over 800 pages left to examine and check for adsense channel ads, (doing this in alphabetical order.)
I am not saying anyone else should do this, that is your choice. All I am saying is it is working for me and judging by the payout and the better ads I can tell the worth of my sites have gone up.
Seems My little idea turned out good for someone other than myself. I don't even believe this weekend....Friday and saturday has been real crappy for the past few months but this time it is holding a high PPC and income. Now to see how Sunday goes.
Ann
Keep up the good work, I think you got hit like I did.
Figuring I'd nothing to lose, I replicated what she did, and junked all my custom channels on 18th July, leaving only regular URL channels - I think I got that right, Ann ;) At the same time, I decided to go the whole hog and ditched the entire contents my competitive ad filter.
Since then (11 and a bit days), there's been a small, but noticeable increase in earnings. I can't say whether this is as a result of Ann's idea, emptying the filter or a combination of both, but any upwards increase in earnings is welcome. CTR hasn't changed much, only the average amount earned per click. I've also put only two URLs back into the competitive ad filter during this time.
So thanks Ann, I'll happily continue this experiment a while longer :)
(Its doubled! Its doubled! Its doubled! Yay! ;))
On the 17th after reading this thread I decided to remove all my unused custom channels (some 30 or so channels) and leave only the 3 remaining active channels. Since the 17th however there has been a sharp drop in eCPM, down to a third of what it once was. My site is actually reaching record traffic levels these day YET I am earning LESS than before. To say I am a mite angry would be an understatement.
Why then, if removing channels may have resulted in a drop in eCPM, am I going to do a crazy thing like remove the remaining custom channels. Well the answer is simple: if Google is doing such a "two-bit operation" thing as relying on custom channel information to determine EPC then they don't deserve my prime real estate advertising space. I want to prove to myself that if I decide to ditch them because of low earnings then it would be entirely justified and I can sleep easy. So the remaining custom channels will be removed shortly.
Alex_Miles said:
Perhaps I just did even with an EPC penalty, and the lack of historical data for the new domain meant Adsense now has to rely on previous account data?
This was a very interesting part of Alex's post. I think when I removed the inactive custom channels from my account Google was then forced to factor in all of my past, unchanneled performance into my current performance. And since my past unchanneled performance was very poor (I started using AdSense on my sites when I was still very much inexperienced hence low CTR/performance due to positioning/colour/content) this adversely affected my current performance.
So what are my predictions when I conduct this experiment. I predict that my eCPM will tank even further and it may just happen that I cannot afford to run AdSense any longer. Of course I would love to be proven wrong but I don't have the time or the patience to go chasing after a flawed algorithm.
Will keep you guys updated.
glad to hear how all are fairing. I hope the little wins turn into big wins for all.
As a short side test to myself just to see if a couple or so ads were really gone, I dumped the four ads from my filter. Within hours the PPC had fallen by 3 points. I went searching and that one ad, good landing page etc, was still there under all three of its URL's. I filtered them and since it is now up to MORE than triple payout today. That particular ad has been a thorn for a long time...been on my site for a year and everyone was probably tired of seeing it. The other ad was a ringtone and it did not show again.
I am not sure about them being forced to use earlier data. I rather think they are working strictly on the merits of your sites abilities to send converting traffic....don't quote me on that as I have no way to know anything for sure. :) Do any of us when it comes to Google.
Fraklik, By not replacing the ads pointing to those channels Googlebot could still "read" the data coming in and use it whether you could or not...hence the part about replacing the ads.
Good luck,
Ann
All we know about smartpricing is a) it's designed to work out what clicks do and don't convert, and b) that Google take lots of factors into account in determining a conversion. A lot of advertisers don't use conversion tracking, and even if they did it's a possibility that Google would ignore that in working out a conversion anyway.
So to add more speculation to the field, are we looking at the possibility of Google considering how we track adsense performance as a factor in determining a conversion, therefore how smart pricing rates the click? Is that such an insane idea anyway?
If (and it's clearly speculation here) that is the case, then the algorithm might be placing undue emphasis on our tracking. Makes me wonder what data they use instead, and if they will simply come back and hit us in some other way in a couple of weeks time once they get wise to this.
I am not sure how this would work on a forums site or a niche that lacks popularity...just mulling over things out loud so to speak. Those kind of sites may not see an increase, but who knows, they could I suppose.
Yesterday ended fantastic for a Saturday and today has made a good start...
Ann
David, It is possible that could happen but I would hope to see them fix it instead of using some other method since this amounts to site suppression of some really high quality sites and that makes for less money for Google or sales for the advertisers.
Here's hoping anyway!
I went to look at the ads showing on the new domain and thats the crappiest excuse for targeting ever. It was fine yesterday! Today it sucks donkeys.
These stats are way outside any range I ever saw and I've been with Adsense since it started.
Anyone else have any data to confirm or refute my suspicion that channels have something to do with ad targeting?
It could also be possible that targeting is done on the basis of what the url is called until the bot visits, after that you get any old rubbish until your customers express a preference for certain ads.
Shame really. My urls are 100% targetted.
I'm going to stick with the experiment for now and see if results improve.
Or if I feel like being evil, I might ban the bot from seeing anything between header and footer and change the urls a teensy bit.
[edited by: Alex_Miles at 8:03 pm (utc) on July 30, 2006]
Ann
It might have happened as my usually stable CTR dropped slightly yesterday but is within normal ranges today.
That could explain why some people have reported fairly sudden changes, whereas smart pricing is thought to be calculated on a weekly basis.
It could also explain why Anne's improvement was almost completely on the CTR side. Based on her most recent posted numbers, her EPC only improved about 17% (i.e. from 6c to 7c a click, or whatever). I'm sure most people will agree that a change of this size isn't proof that smart pricing was affected in any way.
It could also explain how her CTR jumped 200% (i.e. tripled). It's pretty hard to imagine ANY single change resulting in this kind of jump, but going from terrible ads to highly targeted ads might just do it.
...BTW these numbers are based on her July 28 post. But it's a little difficult doing any kind of analysis, because every time she posts results the numbers are wildly different. For example, in post 1 the CTR was only slightly higher but the EPC had tripled. 11 days later, the CTR had tripled but the EPC was only slightly higher.
I do wish I'd listened more in maths class though! A working knowledge of simultaneous equations would be real useful right now.
By tomorrow a lot more traffic should be coming in and each keyphrase they arrive on should cause a slightly different url.
If I like the results, I'm going to tell the bot to ignore everything else so all it has to go on in future is that and the teeny-weeny phrases I feed it. See how it likes a starvation diet.
My thoughts about this test:
It can be just one of your pages AND/OR can be the increase in your traffic quality(AKA buying power and intellectual level of your visitors) which is connected to many other things(number and quality of incoming links, a sudden, surprizing, unexpected media coverage, etc.)
It's a result of many factors forming up a positive boolean. Glad you've cought by it.
Forget about cracking Google algo, well at least by disabling channels. Google doesn't need channels to calculate anything at it's end.
Finally and clearly, I don't think enabling/disabling channels affects earnings. Your good experiments are result of something else, which, God knows what... and, whatever it is, I wish it keeps whatever it does. :)
[edited by: Alioc at 10:37 pm (utc) on July 30, 2006]
Perhaps you should read my post #:3026929 before making such sweeping statements. AND the eCPM has increased a small amount from then.
Personally I dont care how you feel about me or my test but I wouldn't want others stopped from helping themselves by your sneers. Excuse me if I am wrong but you seem to have an attitude problem that seems to be aimed at me at the moment.
Ann
We all respect the test and trying to figure out what's really going on.
Regards.
Which was introduced earlier?
B) Smart Pricing
A) Channel Reporting
Simple?
I actually based my comments on your post 3026929:
"eCPM up by 250%, CTR up by 200%"
That statement means your CTR is up 3X and eCPM is up 3.5X. Thus your EPC must be slightly up as well, by about 1/6 (17%). What part of this am I getting wrong?
So what are my predictions when I conduct this experiment. I predict that my eCPM will tank even further and it may just happen that I cannot afford to run AdSense any longer. Of course I would love to be proven wrong but I don't have the time or the patience to go chasing after a flawed algorithm.
Well Day 1 is living up to my predictions. eCPM has tanked even further. If it weren't so ridiculous I'd be angry as hell.
Good luck to all who tried this and if It didn't work for you I haven't the foggest idea why not.
Goodnight :)
ANN
One thing I just thought of is that I only swapped out the code containing the channel ID on my main site. I have left the other sites alone although I did delete their channels. I'm not sure if that may or may not be a problem, but those sites don't get much traffic or clicks and when they do the click value has usually been .25 or more.
Anyone else have any data to confirm or refute my suspicion that channels have something to do with ad targeting?
I'd say that it hasn't affected the ads that show one iota - nor do the Google updates seem to make any difference. I see the same ads virtually all the time in my best paying blocks. In fact, I wish they WOULD rotate the ads!
My test has been going only since Satuday lunchtime GMT, so it's too early to say if there has been any changes. CTR's been low, but that may be due to ads I can't see showing or more likely that it was the weekend when traffic is slow anyway. EPC is up quite a bit, but not at it's highest this month and eCPM is slightly under average because of low ctr. I suspect that what I saw over the weekend was purely normal variations.
Yesterday did improve and showed a 100% return before I did this, but I'm still cheesed off at the crappy ads and wanted to give it a go anyway. I did eventually get my EPC back but the average for the day was 20c not the 36c I want.
Over the next few days there should be traffic to new pages that never saw the light of day before, so I won't have to wait for re-spidering to see what the effect of strangling the Adbot is. I'm expecting an initial good EPC in keeping with the url, then I expect the bot to to come round and eat the search phrases and titles. Then we'll see.
Still not using channel data except domain channels. Fewer variables that way.
[edited by: Alex_Miles at 6:43 am (utc) on July 31, 2006]
Jomaxx said:
It's been known for some time that if you change a page's subject or are getting mistargeted ads, changing the ad block will trigger a re-spidering and a re-evaluation of the page. It's possible that simply removing the channel ID could have the same effect
This is not my experience. In periods of mistargeted ads, changing/removing/adding new ad blocks did NOT cause my page to be respidered.
In addition, I can also attest to the fact that I was in a period of mistargeting when I coincidentally instituted Ann's plan, and my mistargeting was NOT corrected immediately as a result.
You guys are driving me crazy with this theory.
I just got rid of all channels.
I guess the least that will happen is the bot will take a fresh look at my pages - good thing since I've been adding content like crazy for the past 2 weeks.
I need $100 more a day to go back into retirement... fingers crossed.
This is not my experience...
i didn't delete the channels on my site but disabled all of them.
and ...i found that my earnings have nearly doubled...!
thanx ann... :)
I'm filling up my competitive filter and I'm going to add URL channels to see what's going on...
How long did it take and how long will it last?
I did the same and almost had a heart attack when I saw that my earnings plummeted after a day of substantial increase.. They went down 70%!
So all I can say , that Ann's theory has SOME merit. BUT it can work both ways. The lucky ones will gain and the unlucky ones......
It may have something to do with the average ECPM or average EPC. If channels are removed and the erstwhile low paying channels dominate the high paying ones, you could see a disaster happen. In my case I had 20/30% high paying channels and the balance where low paying. When I disabled all, the low paying ones seemed to dominate! I maybe wrong totally but this experiment is not for me seeing that I am basically faint hearted (with little staying power ;-) )
