LucidSW - 12:09 am on Dec 3, 2011 (gmt 0)
Well I like playing with numbers and have lots of data so I checked my own after reading that link provided by RhinoFish. It's an interesting article but I totally disagree.
At first, I simply compiled data from all clients that track conversions. I did not take into account anything else hence my statement in my previous post.
That article is two years old. I did the same calculation by year. In 2009, there is great variations in conversions by position. I don't have as much data as Google has - enough however to be significant - but the percentages from first to fifth position were 3.06, 3.62, 3.12, 2.62 and 3.18.
In 2010, conversions were 3.48, 4.47, 3.20. 2.57 and 2.17 and so far this year there's a drastic change although explainable: 1.64, 0.72, 0.75, 0.89 and a hair under 1% in fifth. It's also for a much wider variety of clients as I didn't include the archive of past client for 2009-10.
>> the real question is how the conversion rate for the same ad would change if it were displayed in a different position.
Yes, that's the real question. So I did the same calculation for ads with at least 30 clicks in each of the top five positions. There were a total of 48 ads meeting the criteria in many different verticals. The conversion rates from first to fifth: 1.96, 1.45, 1.07, 0.91 and 1.11. The best rate was 2.06 for those same ads in the seventh position (total of 72 conversions at that position).
So I don't know how they came up with their conclusion that rates don't vary much with position. I don't see it. There is more than a 5% difference across positions, something mentioned in that article.