I would say it's the title. Description is next. If it can attract, I don't think it matters if your rank is #1 or #3. But if the rank is 4 or lower, I guess the click rate will go down naturally because the positions of them are so low that people need to SCROOL DOWN to find you. (default is 10 links on a page) In my opinion, #3 is the best place because its usually the center of the screen.
I just don't get it... How can anybody think that there is a "one size fits all" rule for SERPs? That totally defies logic, and I have already given one simple example. This example was based on my own personal clicking habits. I really cannot believe that I am that much different than most other users. I would even suggest that there are more parameters that go into the importance of click. (such as whether or not it is a first time search for the keywords)
I"d agree that being #5 or #6 really slacks the traffic. You're thinking about 70% is probably not far off. What are your secrets to one upping your neighbor who is in a higher position?
If your neighbors got good titles and descriptions, you probably lose lots of clicks. Especially if they are above you!
Here is a study that was done to see where people look when they visit Google. I use this as part of my sales pitch when I tell clients the importance of ranking higher on search engines and how it can help their business. http://www.eyetools.com/inpage/research_google_eyetracking_heatmap.htm
I am not talking about bad versus good descriptions but good descriptions versus killing descriptions and I am talking about descriptions that not only bring visitors but targeted ones that want what you have. I had(now have #3) a page 2 ranking for a term that a friend had a 1st page ranking for. I got double the traffic he got with my second page ranking then he did with a first. I agree that as high as you can get is best. I try to get a #2-3 ranking and then that is good enough for me. After that I just work on fortifying it.
Self plug but I did do the math carefully on this: http://www.redcardinal.ie/search-en...08-2006/clickthrough-analysis-of-aol-datatgz/ Data was derived from the full AOL dataset (~= Google)