Can anyone tell me the best way to estimate the groundwave coverage
I'm likely to get on 75 meters?

I tried R. J. Edwards (G4FGQ] GRNDWAV3.EXE, answered all the questions
about type of terrain, frequency, power output, and antenna
efficiency, and got an answer of around 50 miles for an S-5 signal at
the other end.

(With the noise level around here, S-5 is about the minimum to shoot
for around here...)

And yet, in the real world, I am lucky to be able to establish
reliable contact with another station that's a bit less than 10 miles
away.

We are both using horizontal NVIS antennas, which I guess aren't as
good for groundwave as vertical antennas. In any case I have no idea
what to put in to G4FGQ's program for "antenna efficiency". I tried
25 percent and got the 50 miles for S-5, then reduced antenna
efficiency to 10 percent and got 40 miles for S-5... still well below
what I am seeing in the real world.

I switched to a terrain type of "City blocks, roads, streets,
industrial areas, rivers" and got 32 miles for S-5, then switched to
"Mountainous regions, bare rock, vegetation in valleys" (we do have a
few molehills around here that people call mountains) and got 28
miles... still beyond what I'm really seeing.

So, how can I estimate the longest reliable groundwave distance on 75?

Thanks...