I was thiinking along your lines Scott.  Attempting to poll examiners and abstractors will probably make them clam up if you're a competitor and getting an idea of how many items are recorded (from the county) monthly isn't indicative of how many searches are actually done as Robert said.  I couldn't come up with any idea how to accurately answer the question without knowing more about the why.  The number of searches done in a county isn't an indication of how many any of us would get there (how we wish it were). 

More tech friendly counties with better computer systems may have some tracking implemented but I don't know whether it would differentiate between all the single doc retrievals and true searches that you're trying to pin down.  The random poll Scott suggested would be nice but its accuracy would still depend on who participated and their honesty - rather like the Census.  It would be a pretty big campaign compiling input from a nationwide poll too.