For example, if you don't do any RF related work, you aren't very likely to need a spectrum analyzer, and especially not a high-end one. FFT mode on a scope will probably serve one well enough if they're not doing RF. Conversely, if you are doing RF work (where I mean, building or repairing radio transmitters and receivers) then you probably very much want a dedicated spectrum analyzer, and probably an RF power meter as well.
Similar arguments could be made around logic analyzers... if you aren't interested in doing a lot of digital logic work, and especially not high-speed digital logic, you probably don't need a dedicated logic analyzer. For retro-computing stuff and debugging simple stuff you can probably get by with one of those cheap little USB logic analyers[1] and Sigrok[2].
Now probably everyone wants at least one decent hand-held DMM, a lab power supply, and an at least halfway decent oscilloscope. But beyond that, you really can tailor your test equipment choices largely based on what kind of stuff you find yourself doing.
And just to throw one other random anecdote out there: when I was building my lab, I bought a decent dedicated bench multimeter relatively early on. It wasn't the first thing I bought, but it was maybe the third or fourth. In hindsight, I rarely turn it on, and it probably gets less use than anything else in my lab, including seemingly niche devices like a programmable load. It turns out, I prefer using one or more handheld DMM's over using the bench DMM. I can't even tell you exactly why that is, so take it for what it's worth. But a benchtop DMM has been just short of useless for me. shrug