We live in a "golden age" for radio/mm studies of high-z galaxies, thanks to major upgrades in the capabilities of existing facilities (e.g., JVLA, PdBI) and to the advent of ALMA. These developments allow us to perform measurements that were simply inconceivable until a few years ago. Nevertheless, studies of high-z galaxies at radio/mm wavelengths are still sensitivity limited. Even in the ALMA era, radio/mm astronomers more often than not struggle against modest signal-to-noise detections or upper limits. In this context, two common problems are: 1) how can we push the depth of radio/mm interferometric observations via stacks? and 2) how do we blindly search for emission lines in data cubes? Both problems require a good understanding of the properties of the noise in the data, and of the targeted sources. In my talk I will present and compare some state-of-the-art approaches to deal with these two questions.