Given that field and signal density decrease occurs more quickly as wavelength shortens, it’s highly relevant to effective range at a given EIRP/ERP. So, Sig/field density drops quicker over distance at UHF range frequencies vs VHF just VHF has shorter effective range than HF has generally.
If range was frequency/wavelength independent, we’d never have ventured into multiple band usage and still using HF/MF/LF for most radio usage. It had more to do with practically and more stable propagation factor that we use typically VHF/UHF for LMR and other mobile inclusive radio use.
But addressing the posted problem, I’d be inclined to believe an installation issue is responsible for low performance, or terrain/geography or localised interference effectively reducing the usable range.
Installation, if the case, resolution depends on who fitted it for liability and resolving in the case it’s an installation issue will require looking and testing. But it could be any of those three combining to a reduced effective range - which is why when you tread into supplying VHF/UHF/SHF systems you really need to survey or at least assess the intended use area in terms of topography and geography.
The shorter the wavelength, the less latitude there is for non-LOS communication to be maintained effectively at long/extremes of range.
So just as wavelength and frequency have a notable factor to effective range as much as EIRP/ERP, so does transmission mode/modulation type. At extreme ranges, weak AM modulated voice is more readable than FM at the same effective Sig level received because AM just gets quieter until it disappears, FM gets less readable because as the Sig level falls towards low or below threshold levels, noise overtakes the modulated content or simply put, SNR deteriorates and readability is much worse at extremes. That’s why weak FM and the resulting balance between noise and weak transmission has a very harsh sound at extreme ranges, vs the soft ghostly whisper tone/softness of AM/SSB at extremes where it just fades away.
So unless there’s a fundamental fault in the setups, damage or poorly planned and fitted install, you’re looking at a range/geography at topography levels potential for why what range you’d expect 50W say to work out on at just isn’t the case. But given 50W around 130-150mhz is effectively usable for approximately 1 mile per watt EIRP under best case low lying or fairly flat terrain on a 1/4 vertical, but that ratio drops rapidly in hilly environments (I live in a very hilly environment at a low point myself) then I’d not write off geographical factors nor the potential that on FM, localised interference causing the receivers to be cycling between localised EM with harmonics near the operating frequencies and weak reception of compromised (for whatever reason) transmissions adds up to an even worse effective SNR in terms of the desired transmission as heard by receiver.
But it could be as simple as damaged coax on installations or unusually high feeder loss aggravating what’s probably a compromise antenna mount.
If the antennas are radial-less half wave or colinear types, with subsequent low to medium angles, comprised range shouldn’t really be much of an issue on very open land (I’m assuming agricultural type industrial use or similar outdoor use may be the case here).
But everything has a contributing factor to range and readability of received signals, from localised noise through adjacent signals off and on frequency, no matter what mode or wavelength is used, there is no such combination where a given mode/band or specific band is immune to the collective of causes or spot instances of a mixed bag of contributing factors potentially causing issues.
Nothing is mathematically or science lab perfect to modelled test case scenarios because mathematical and scientific modelling can’t account for non predicted or non-site specific measured operating conditions, which is an essential part of why best case modelled or estimated performance usually exceeds reality as end users experience it, so you need to employ a fudge factor to over engineer the setup to exceed requirements at estimation or modeling stage to create some breathing room and latitude for the unforeseen and prevailing on site conditions.
50W TX has phenomical potential even at VHF/UHF is the prevailing conditions favour a good LOS or very near LOS path. I mean 10 miles is achievable with low to moderate readability on UHF if the transmission path is LOS or very near LOS at 1W into a 1/4. But what you call low to moderate readability probably differs to my tolerance of readability as it’s quite a subjective thing as much as your sense of low to moderate readability will differ to that of your customer and/or users of his gear.