The efficiencies supposedly gained on WDM is lost when using just the (Standard)1550 and 1310nm frequencies vs. using DWDM (Dense wave division multiplexing) i.e., the utilization of the sample frequencies such as: 1547nm, 1550nm, 1553nm, 1555nm, 1557nm, and 1559nm. The slice and dicing of this frequency lends itself to the economies of scale when outfitting the equipment at the end users domain. Sparing both a 1550nm and 1310nm transceiver is very costly especially if it's not generating any revenue. The one caveat is the revenue stream of it being planned in a redundant path circuit. Another big point to ponder is that only ONE fiber is being used for multiple forward and reverse applications, not two.
Remember that the link loss budget is inherently hindered when introducing both 1310 and 1550 onto any given circuit. As an example, if the outside range of a given design is a 29dB link loss @1310 but the 1550 is spec'd out to 35db, one of two things must happen. First, either you have to regenerate the 1310 signal ahead of the upper link loss drop off or you'll have to underutilize the full potential of the 1550 gear. Either way, you'll end of having duplicated equipment plus a regen. integrated somewhere in between your two points.
blagh blagh blagh......
