Forum Moderators: martinibuster
The latest thing that happend was I had a small 10 page website that only a trickle of traffic but each click was worth over $x on the main keyword search term.
Well this past week on Yahoo's update this main page moved up to rank 3 on Yahoo (and a surge of traffic and clicks) but still low on Google (where what traffic the page was getting was coming from).
Same advertisors , the page found for the same keyword search ... and poof EPC drops to pennies ..
Could it be that adsense algo pays higher for referrals from G and lower for oter engines?
Or have I just caused some brain damage from bangingmy head against the wall? ;)
Yes, it can be done, but is impossible as Google has written the script.
I will not claim my $50 :-) but I am sorry but this is false.
It is possible, and Google does it.
start.htm has a link to www.a.com/page.htm
www.a.com/page.htm calls a javascript in www.b.com (just calls it, without any previous manipulation)
The javascript's document.referrer will b start.htm, not page.htm
I have PM-ed you the URL of a page that demonstrates this.
Feel free to link to www.a.com/page.htm from any page, this will still work.
The javascript sees the page that linked to your page.
The additional trick that Google does is that the first JS then calls a second JS passing the referrer as a parameter on the URL.
On domaina.com: test.js:
document.write("URL:"+document.referer);
on domainb.com/test.html
<script type="text/javascript" src="http://www.domaina.com/test.js">
</script>
Go to domainb.com/test.html from a link on another website, it load the javascript from domaina.com and print URL:(the referering url)
The javascript can then if it wants send that information off to domainc.com through another javascript request.
e.g. document.write('<img src=http://domainc.com/reftrack/?ref='+escape(document.referer)+'>');
Google DOES track the refering URL. I bet this is useful for smart pricing and fraud prevention.
So it IS possible for google to pay less for referals form yahoo, though I doubt it is clear cut as that. Maybe referals from yahoo convert worse, or they can look at the keyword from yahoo.
In fact this has rather big implications for google's knowledge of the web, for every site that has adsense on it, they know where all of your traffic comes from.
Frox it's not $50 but $100.
I won't go in details here, but google is collecting a lot of information about you, and your configuration.
The first use I would do of all these information is "fingerprinting" against simple click-fraud
But also very useful for say comparing referers of yahoo, msn in terms of keywords and traffic levels :)
Ok, who will be the first to complain with Google about this? :-)
Thinking about this, this behaviour of Google COULD (could) make us break our own privacy policies, as we are giving to Google some information about our visitors' browsing history. I have some sites where I explicitely say I would not do such a thing, and will have to change that...
I have PM-ed you the URL of a page that demonstrates this.
Kudos to you guys for proving me wrong. It's a topic that has come up in the past & has always ended up being that Google was not collecting that info. I guess that means either they weren't then, or that no one noticed the document.referrer passed along your page referer instead of your page.
Anyone know of an old posting of show_ads.js to see if was being used back then, or if it's only been used recently?
Thinking about this, this behaviour of Google COULD (could) make us break our own privacy policies, as we are giving to Google some information about our visitors' browsing history. I have some sites where I explicitely say I would not do such a thing, and will have to change that...
Javascript is client-side... so one could argue that it is the visitor submitting that info to Google, not us.
PM your paypal addy
Javascript is client-side... so one could argue that it is the visitor submitting that info to Google, not us.
Of cource on could argue so, but frankly I don't feel it would be true: this javascript is code that gets run on the user PC as an effect of visiting my web site.
Or should I also feel ok to install a spyware on the users's PC? That's client-side too! :-)
PM your paypal addy
-----------------
Wow, I appreciate it. PM on the way!
Javascript is client-side
-----------------
Of cource on could argue so, but frankly I don't feel it would be true: this javascript is code that gets run on the user PC as an effect of visiting my web site.
OK , Now lets look at the second part of my question.. that you could get EPC of differing amounts based on referral (In My case lower EPC on Yahoo traffic) .. could it be that G is doing just that .
I seen clear evidence in my case that that happened .. it was easy to detect because the site page had so few visitors and they came from G ..
then last weeks Y update my ranking moved first page for main keyword and traffic multiplied X 10 or so and EPC dropped like a rock from x$ to pennnies.
Now 95% of traffic from Y and 5% from G
But I dont know what they would gain from lowering EPC based on a referral other than G?
Probably has nothing to do with it..but it seems to be much more than just a coincidence
And Kudo's MrAnchovy for standing by your bet!
To test this properly is to use channels and assign google a channel, and yahoo a channel then you will know for sure.
dauction, it is quite frequently reported that an increase in traffic results in a drecrease in EPC.
Call it smart pricing, call it a conspiracy (I don't anyway see any logic in this) but it seems to be real.
I also had similar feelings, although it's of cource VERY difficoult to show wether this is truew or not.
As per the yahoo vs google, if your pages are for example php, as already suggested you could:
1) prepare two (better three) custom channels
2) when your page is called:
if referrer=yahoo then include adsense_ yahoo elseif referrer=googe then include adsense_google else then include adsense_other
after a few days this shoud be quite clearly visible.
I can't do this because the percentage of my non-google referrers is bell below 1%!
The problem is, it requires that you minimally play with the Google javascript, which per TOS you should not.
All you need is just inserting something like
if (...yahoo...)
{
block for first channel
}
else
{
block for second channel
}
So, you should perhaps ask Google if they find it legitimate..
