@troux, good idea. I do tend to see similar feedback in a lot of the "rack feedback" requests.
In addition to the ideas above, how about adding in some of the following:
a) write out your vision for your modular synth. What do you want it to be great at? It's okay if this is vague to begin with, but the clearer the vision, the more coherent the modular design will be (e.g. more fun, more good sounds, less frustration and waste).
b) consider how big a modular case you would likely get. Next, consider a larger case! Ideally, plan to leave 25-50% open space in your first case so you can add additional modules as you learn your needs and interests better.
c) consider your $s budget for modular. Is modular a better option for you versus other music items (e.g. software or hardware synths, grooveboxes, etc.)? While modular is great at many things, it can be an inferior alternative in some situations.
d) plan your first (or next) modular rig. Start by thinking about proportions of modules: a useful guideline (courtesy of JimHowell1970) is "sound sources < sound modifiers < modulation sources < utilities." A rig with good proportions of modules will tend to seem balanced and deep (for its size); a rig with skewed proportions will tend to seem unbalanced and limited (for its size).
e) pick some modules! A balance to keep in mind is "inspiration / working instrument / going slow". Something has led you to modular -- make sure at least some modules embodying that "inspiration" appear in your rig. "Complete working instrument" -- if you buy a hardware or software synth, most of the time it arrives as a fulling working instrument, essentially complete on its own for some musical task. With modular, there is a risk of missing an important module type, and severely limiting the rig so that it is not as much a complete working instrument as a VST or hardware counterpart may be. Consider all the modular functions you need to accomplish your main musical tasks, and include those in your draft rig. Lastly, there is the longstanding advice among modular synthesists to "go slow," namely get modules one or few at a time and learn them very very well before adding new modules. Keeping inspiration in your rig, designing a rig that is a working instrument, and going slow enough to learn your incremental new module are somewhat contrary guidelines -- with some consideration, you can find the balance right for your rig.
f) consider some "no regrets" or "tried and true" modules if you're a beginner. These can include [to insert list from across thread above].
g) use tools like Modular Grid and/or VCVrack to test your rack designs. There is a large modular community and other modular users routinely help each other start or extend their modular journey!
h) dive in, take time to learn and explore your rig, and enjoy!
IMO maybe some of the above is helpful broader advice to include in a primer for new folks. Feel free to keep / toss / update / recombine any of it as you see fit.