According to Prabhu's analysis the following sites receive different content depending on the user agent calling the site. We should eliminate these pages from our tests (and replace them with the next N pages in the spreadsheet), so that the data do not throw off our conclusions.

https://youtube.com
17186
58586
1903 insertions(+), 133 deletions(-)
http://espncricinfo.com
272017
19899
500 insertions(+), 5237 deletions(-)
http://waptrick.com
19851
8166
129 insertions(+), 92 deletions(-)
http://mercadolivre.com.br
192798
169973
145 insertions(+), 686 deletions(-)
https://google.com/?q=restaurants
223680
15540
485 insertions(+), 2 deletions(-)
http://flipkart.com