I generally keep my nose out of the pay-per-click arena, but PPC does present some great opportunities to do statistical analysis of user behavior. I recently got my hands on a list of about 6,000 adwords ads complete with impression and click through numbers, so I started writing scripts to see if I could find some patterns and the first place I looked was line length.
In an adwords ad you have 4 lines, a title, a first line, a second line and a display URL, the title can only be 25 characters long, but the 2 following lines can be up to 35. Below are graphs of the clickthrough rates for ads where a specific line was a specific length (with a linear trend line plotted as a bold black line), first up is the title:
Here we see that as an ad’s title gets longer, the average CTR in our data set got lower (as long as its more than 8 characters long).
The first line under the title however shows an opposite trend, the longer this line the higher the CTR rates in our data set. There does appear to be a sweet spot between 25 and 31 characters where the ad performs the best.
In our data set however the length of the second line appears to have very little correlation with the performance of the ad.
The length of the display URL also seems to have a very low correlation to CTR.