Since channels were first offered, I've used them to test colored ads, placement of ads, size of ads, ads on certain pages, etc.
I've made some interesting observations.
However, if push came to shove and someone asked me if I'd bet the farm on it, I wouldn't say that all that testing and information has necessarily resulted in more money in my pocket.
There are just too many variables from site visit to site visit, from visitor to visitor, from ad to ad, etc. to be able to say with certainty that this works best or that doesn't work well, speaking in a broad sense anyway.
After the recent channel discussions I began eliminating channels, both by putting new code on the pages and by deleting channels on the Channel Management page.
I'm almost to my goal of having no custom channels and only URL channels.
As a result, I'm spending less time analyzing reports, changing channel codes, naming channels, etc. I'm spending more time improving my sites and adding good quality content. The improvements and additional content SEEM to be gradually growing my income over time.
Ask me a year from now and I might have a different opinion, but as of right now my advice to a new AdSense publisher would be to use channels to test various layouts, color schemes, etc. to learn the basics and then scale back and spend more time on the sites and less on the channels & analysis.
Information is a valuable commodity. But only if you do something useful with that information and it was worth the time and effort expended to obtain the information.