USDM is strict on MPM these days... even a few miles over will result in an invalid itinerary.
I wonder how US DM (or rather, agents) "check" the MPM rule.
I just finished a recent booking, the end result was BNE-HKG-NRT (stop)-HEL-ZRH (dest)-LHR-HKG-BNE
So originally, using EF, I checked out the MPM for BNE-ZRH and it was incredibly generous (IMO), especially after taking into account the 25M number. The original routing was BNE-SIN-HND,NRT-HKG-LHR-ZRH-LHR-HKG-BNE. Now this original outbound routing was in violation of the MPM (inbound was fine), so - alright - I managed to find an alternative for NRT to ZRH, which went via HEL instead. That shaves of a couple of thousand miles off the outbound, but the agent still contends that the outbound exceeds the MPM. Apparently it is the routing to TYO for the stopover which is doing it. Luckily, whilst the agent was on hold with another person trying to work out what I could do to save the stopover, I found some availability via HKG (this only happened to open up / reappear in recent days!), so when the agent came back I changed the routing to NRT, shaving even more miles off the outbound and
now I was fine.
Using EF's MPM calculator, if I punch in the full routing into the main field, it will give you a coughulative MPM calculation, which compares the coughulative mileage travelled against the MPM for the origin to progressively each destination in the routing. If I put in the routings I had before the final one that was ticketed, at some point the calculator will show that the MPM is exceeded along the way, even though once I got to the end (i.e. ZRH), I was overall within the 25M MPM.
This leads me to believe that
at least the routing to the stopover point (if you choose to have a stopover) must also comply with the MPM (or 25M as it were). That said, as with anything US DM, how consistently this is applied could be anyone's guess. Furthermore, my proposal may be incorrect anyway, albeit MPM is being taken rather seriously at the moment, if not from here on in.