While monitoring activity from Covenant Eyes on one of my servers it became obvious that many of the pages being accessed were fairly unique, not as popular, and easily allowed me to figure out the actual customer Covenant Eyes was watching.
To test my theory I checked the log file for one unique page Covenant Eyes requested and sure enough only a single IP had accessed that file during the course of the day.
Then I got a list of all files that this visitor's IP had viewed and compared it to all the files that Covenant Eyes requested and it was an exact match in the exact same order of access, without any obfuscation, so it was a 100% match without a doubt.
I've been monitoring this situation for several days now and it's always the same.
The visitor comes and views some pages and about 90-120 minutes later Covenant Eyes comes and asks for the exact same pages in the exact same order.
Here's a sample of a visitor's access:
127.0.0.1 "justapage.html"A while later Convenant Eye's asks for the same pages in the same order:
69.41.14.x "justapage.html"Same pages, same order, definite match with a unique page like "veryspecialpage.html" that nobody else visits on the same day. Additionally, they appear to do each customer's files they monitor very quickly in a batch so it's pretty easy to see that those files are related to a single visitor making identification even simpler.
Now with a simple script I can find out who they were monitoring with extreme accuracy as long as the visitor looked at more than one page unless that one page was unique and nobody else looked at that page during the day.
Making it harder to identify which visitor they're monitoring wouldn't be that difficult just by staggering and randomizing their page requests over the course of the day. However, I still don't see how you could protect the identity of your customer if that was the only customer of the day that accessed that web site unless you throw in a few bogus page requests to throw a webmaster off the trail. Even with randomization and fake page requests you would still have a problem if that customer was the only one to access a specific page as mentioned above, but at least it would be a start in making the monitoring activity just a little more covert and possibly less traceable.
The site of mine where I did this experiment, which isn't this blog, gets from 20K-40K visitors daily, so if I can easily find a needle in that big haystack then it would be trivial on a low traffic site.