The first rule of observing is: you don’t leave the telescope until your data are reduced and analysed. If that seems like too much to ask then you’re using old analysis approaches and your competition isn’t.
When I was learning how to analyse high-resolution stellar spectra I wrote an intuitive, graphical software package for analysing spectra quickly and precisely. What used to take ~1 day per star now takes a couple of minutes, and it means I (and now, all my collaborators) follow the rules! It means we can vet candidates quickly, find the most interesting objects and return to them in the same night. Now the reduction takes more than an order of magnitude longer than the analysis! The code is described in Chapter 3 of my thesis, and a screenshot is below. There are more objective (read: better) ways to do stellar spectroscopy – and I will post about this in the future – but the code allows us to get a very good idea on what we’re looking at, very quickly. That’s important.
The last three nights I’ve been observing on Magellan (with Schlaufman) using the MIKE spectrograph, looking for extremely metal-poor stars using a novel technique devised by Schlaufman and me. The selection approach is as efficient (or more) than existing techniques, but the candidates are ~3 magnitudes brighter. That makes the requisite follow-up spectroscopy achievable for a large sample of stars. And our approach only uses global existing sky surveys, so targets are available throughout the year no matter where you’re observing from. The approach will appear in print later this year.