The required receiver gain will depend on sample concentration, nucleus, experiment, probe, and probe tuning. For a given nucleus and experiment the higher the concentration, the lower the gain value.

I'm not quite sure I understand your question. Why do you think that the gain should be the gain should always be the same for every sample? You can set the gain manually using the "rg" command, but on Bruker systems the best way to set the gain is by using the "rga" command (receiver gain automatic) before you start your acquisition. If you are running in automation (ICONNMR) this will be done automatically. Even with the same sample and experiment, "rga" might select a different value at different times if the probe has been tuned slightly differently or the required gain is just on the threshold between two steps.

I think that you may be misunderstanding the meaning of the term "gain" here. It is not a measure of instrument performance (like sensitivity or resolution), but rather a user adjustable parameter that adjusts the gain of the instrument to the appropriate level for the current experiment and sample.

Having a gain value that is too high will cause clipping in the receiver and/or digitizer and result in a distorted spectrum. That is the situation that you want to avoid. Having a receiver gain that is too low can cause loss of sensitivity and dynamic range, but modern systems have such high dynamic range (effectively > 20 bits) that the gain can probably be a factor of 10 too low and you wouldn't see the difference.