I'm fully aware that it looks like I'm beating up a dead horse, but there's been a (relatively) new piece of evidence published by Fluke on the 15+ years old discussion. I think it's worth to look at if you are making this decision for yourself or the customer.
In a nutshell: you've got to play along with the manufacturer(s) of your choice: if they optimized their cables and connectivity for T568B, you will have a hard time making it pass RL if you installed it on T568A, and Fluke gives you some testing data to substantiate that claim.
In my mind, the tricky part is: how do you get them to tell you which pinout they were optimizing for? Alternatively, you could stay within one-manufacturer solution and hope that they worked that out for you and have gotten the cable factory to talk to the connectivity factory before they released the product.
Careful with the link, it's long and ugly, but once you get there, the page reads very easily, even has a nice little touch with that animation of A changing into B. Kudos to A. Young @ Fluke who put this together.