Lenny2 - its something that has been going on for years - and I do agree with you as Ive done similar testing when the source of traffic stays the same (usually SE or LL,TGP)
The only variable Ive been able to come up with that would explain part of the problem is a change in processor scrubbing filtering - although if you are comparing two programs that use the same processor that should take that out of the equation.
Other than that I still keep what a good friend told me many years ago in the back of my mind - there isnt a program out there that doesnt shave in one way or another - its all in how you define it
When I find someone doing the "initial push" or the ratchet down the sales during higher payouts - I play them for as long as sales stay better than a certain level - then drop them and move on to the next sponsor that I can play for two or three weeks
