You have to start by looking at where the issue starts? You said a WAN location. First rule out that the network is good from remote site to your core resources. Then... Look at it from the users perspective that is making the complaint. It could be a specific application and not network related at all. Work your way back from remote user to the resources they are trying to access. More times than I can count it wasn't the network but some poorly designed client server application or fat client doing a bunch of database transactions accross the WAN. This can sometimes be remedied with Citrix. Also remember how subjective your question was. If your pings are between 0 and 500 ms from edge to core you can probably rule that out if it's consistent. Usually if it's not it could be real errors or as you might say over utilized which would mean something like a circuit reaching capacity and packets actually being dropped. NPM can help rule these things out... but keep in mind... it's often not the network.
Absolute times are difficult to say, but the old "I hit the keyboard and it hasn't responded in 2 seconds" is a good rule of thumb. There are dramatically different implications when you compare workstations running, say, telnet v thin Citrix and that is where the absolute time problem really arises. It's all so subjective and heavily reliant on the actual application. If you have sites that perform an application well and a similar site, running the same application, is unacceptable, then the line characteristics may be the cause, but be very careful to compare like with like.