When using compression you always have to be aware of the final destination for that audio.
A lot of my work is for my own internet radio station where I obviously control the final effects that the audio passes through before being streamed. I can then set my processing so it compliments that. I also work for an FM station, anything I produce for that I apply much less compression because I know they compress their audio.
If you’re producing for radio there will be some form of compression in the chain of their output. Probably a complex formula their engineers have worked out to achieve the sound they’re after that involves multiband compression, gain control and possibly hard limiting so the sound has the desired “punchiness” but also maintains a consistent level. There’s nothing worse for a listener to have to keep adjusting the volume so radio stations often process heavily to ensure the sound level is uniform all day long.
It sounds like the files you’re producing sound good to you locally because you’re hearing them with just the compression settings you’ve applied but when that compression is added to the station processing that covers all their output it sounds bad? Compression is a great tool but remember there’s nothing worse than audio that’s been over compressed, usually by being double compressed.
The best option would be to ask the radio station how they’d like the output but if that’s not an option you want to try then I would suggest that when you produce your work only apply light compression if it’s going to radio. Do your eq settings, normalise then apply light compression. Whether you use dynamics, dynamics processing, single band compressor or the multi-band compressor make sure it’s a light touch rather than a finishing. Remember the finishing will happen at their end when the sound leaves their desk and heads to the transmitter so you just need to make the audio ready for that finishing rather than finish it.