Welcome to WebmasterWorld Guest from 18.104.22.168
Do not overwhelm your visitors with ads all over your website. Often less ads is more money + more user satisfaction.
Or put lots of ads on your pages in a discreet and attractive manner.
I do both across various sites and can't say one method works better than the other. It depends on the situation, topic, etc.
What worked for mine could ruin yours
What X says works could be a smoke screen
What Y says could be just for bragging rights
What Z says could be to get others to do the testing
There is almost ALWAYS something left unsaid
Trust nothing but your own extensive testing over long periods, drop the word advice and replace it with hints.
Those are the things that have worked best on all our news sites. Have tried hundreds of colours, layouts, blending etc over millions of impressions. Though of course you have to see what suits your own site
All that is taking for granted you have traffic and are producing good content.
[edited by: FattyB at 5:52 pm (utc) on June 10, 2008]
P.S. It's probably against the TOS now
What's your best Adsense Advice
Well, there are many good tips (ads near site content to reduce ad blindness but with enough differentiation to avoid confusing users; no ads on non-performing pages, that is with very low eCPM; etc.).
However, I think the #1 advice comes from Google itself:
"1. Focus on the user and all else will follow."
(Ten things Google has found to be true [google.com])
WORK ON YOUR SITES.
I went through a long desert - where i tried to obsess about the value effect of every work on my site - and how it was affecting assense ads. This got me nowhere - and my earnings were stagnant.
Since i started focussing on my sites and my users - and less time on Adsense - i have grown to point where i make my primary income from Adsense. This is one reason why CPA has not worked well for me... because it requres to much time and effort on the offers and ad code - and you find very little time to work on your own sites - development and marketing.