This is something I've heard over the years to do to improve performance. That when you get a new machine to "set it up as new computer then spend the weekend re downloading all your apps, getting preferences setup, moving over files one by one" and I parroted this myself to many people. The idea was that stuff can get garbled along the way and your system will work better because... reasons.
But is there actually any concrete evidence that this does anything? I'm sitting here looking at a new laptop and with my sketchy internet speed, and all the little tweaks I run in various programs, I'm not thrilled about spending what I estimate to be 10+ hours of my time re-configuring everything even after the downloads are done. Is there any actual A/B testing showing if this matters or if that's one of those hold over myths from the old days?
Setting a new machine up as new will clear out any clutter that has accumulated under the hood of the older machine. However, it's impossible to quantify this because everyone will have different amounts of clutter under the hood. The other consideration (which the the most important one in my mind) relates to applications themselves. Assuming you would be moving from an Intel-based Mac to an Apple Silicon based Mac, I would set up as new anyways because Universal or Apple Silicon apps will have better overall performance than Intel-native apps. As far as documents and other files go, those can just be copied over from the old PC to the new in bulk and only takes as long as the actual transfer between machines. File types are platform and architecture agnostic, which is why you can open a .docx file in Microsoft Word under Windows, Intel Mac, and AS Mac without needing to convert the file, even though all three builds of Word will be different based on the OS and hardware architecture being used.