There are a few ways to achieve your goal. But the obvious one is to structure your notebook as you describe which is to read the data, organise then analyse them.

I would think hard before resorting to mx files, but they may be required if you have memory issues.

Don't forget that you can organise your notebooks into sections and subsections, and that you can use loops and relative paths to access data in files.

As per my attached notebook, setup some global variables, read a bunch of data into arrays, plot them and write the data so that they can be analysed further.

It is difficult to be more specific without and idea of data sizes and what you want to do with them

Attachments:

Presentation.nb

Sections etc,. help you organise your thoughts. I typically use them to hide details in steps in an analysis so that when I execute the entire notebook, I can reproduce an output such as a plot. Such an organisation assists debugging and self-documents the workflow.