Compressing acoustic guitar is all about the end result and how the instrument will fit in the mix. Assuming you are getting a good source (good mic placement, instrument, etc.) you should decide why you want to compress. If the track doesn’t need it then don’t do it!
Strummed guitar in a thick mix
If it’s just a singer and a guitar the odds are you are going to want the track to breathe a little. However, if you are trying to fit a strummed guitar into a heavy mix then compression can work for you by creating a fattened percussive sound that will add texture. For things that I am going to compress heavier I like to use two compressors. This helps the track sound more natural and not pumping from the spikes in the signal. Also, you should EQ before compressing this so the low end spikes are not overloading the compressor.
The first compressor ratio should be set to about 10:1-12:1 to chop off any spikes and the compressor should only be hitting on the highest peaks. You could also use automation here.
The second compressor is to even everything out. I set the ratio at 3:1 or 4:1 and the attack relatively quick to bring out the percussive sound that will help the track cut through a mix.
Finger picked guitar
When you are trying to get an up-close intimate feeling you can compress to bring out some of the room (which should be treated). For this you could use the settings roughly of the second compressor except bring the attack up to about 15ms. If you want a grittier sound then bring the ratio up to 6 and you will get more finger squeak noise. Try to have no more than 4-6db of gain reduction or you will lose some of the nuances that make the acoustic guitar great.
Generic Compression Disclaimer
Compress for a reason. Some other ways to fatten your acoustic guitar parts are to double them, use delay, chorus or a micro shifter like the Schwa CMX.
Do you use compression on acoustic guitars? Let me know what works.