Real quick and one of my favorites is to take a client let them sit at two identical systems and do like they normally do, open up some browser windows, pop open maybe a CAD or GIS program, run a virus check, check email, basically do what they do in a normal session, then put them on the other rig and let them do the same, not telling them which has say 2133 or which has 2400....if they are on a single system build , very similar, have them sit and do what they do, thhen tell them I want to try it with some tweaks (not telling them I'm changing DRAM) and send them to get coffee or a sandwich or something, they come back and get back into it, the whole spiel, windows open, virus scanner, CAD, images, whatever.....then I'll ask which they like better the one rig or the other, or the system w/ or without tweaks, appr 85% of the time they pick the system with the faster DRAM, and if it's a couple steps different say 1600 vs 2133 or 1866 vs 2400 it's even higher they pick the one with the faster sticks...and blindly, they have no idea what the difference between the two even is.....generally many are amazed it's the DRAM because they have read the same junk as everybody else - and then I'll often pull out some reviews showing that these 'expert' writers don't know what they are talking about...i.e. have seen a few where a writer complains 'the RAM didn't set the proper timings under XMP', well, duh, the sticks don't set the timings, maybe that's why. The sticks have the proper info in their SPD, it's the responsibility of the BIOS (actually the programmers) to pull that info and implement it, which it often doesn't so and why about 50-75% of most BIOS updates are actually RAM fixes/updates....It would help if during their QVL testing the mobo makers would test under XMP, but they don't, they 'test' with whatever they have on hand (you'll often see sets that have been out of production for a couple years, and they test at the mobos default of 1333 or 1600