This - potential to tie in much information eventually - is one argument for pretty dense and open coding, at least in the early stages.
However, sometimes the purpose is pretty big: 'how do X described their experience with Y?,' or pretty vague: 'what are the characteristics of x?' (this was one of my prior studies), or emergent, 'what is the process of xyz?' or not all that directed: 'what was it like living during Y?' Arguably just about everything in a transcript - which is just about everything someone said in response to questions about the purpose - could at least tangentially relate, or might relate to something else that is directly relevant, or might begin to describe a relevant experience, etc. (For those people who chop up transcripts with scissors and spread them on the floor, perhaps you need a dangling pennant or flag with your questions/purpose.) I think this is great advice and have certainly repeated it elsewhere in this blog, and dozens of times to students. Ron Chenail, my favorite (overall/generalist) qualitative research expert, suggested writing your research question or purpose on a sticky note and attaching it to your computer monitor - or wherever you will see it while you code. The amount of coding shown in the examples seemed just about right to me - but I wondered why I though so.
HyperRESEARCH makes a free trial version available and has some nice coding examples. I also like the transcription software, HyperTRANSCRIBE, and know that at least one of my former students purchased this. I am not quite 1/3 of the way through the class/semester and I realized that I only barely touched on the idea of 'what should be coded?' This question came to me as I was reviewing a tutorial for HyperRESEARCH, a program I used in the past and think is among the particularly good alternatives for student researchers who are considering purchase of a user license for something.
One of the things I am particularly enjoying about this is that I am able to focus on the mechanics and style sof coding - by default most QDAS analysis is coding - this is what these programs were mostly set up to do and remains what they are most comfortable doing.
This semester I am teaching a qualitative data analysis course with a focus on use of software or QDAS (Qualitative Data Analysis Software).