A topic not adequately covered in this topic, IMO, is corner-case testing. E.g. testing across temperature, supply voltage, and especially Si process variation, in the simplest example. EDA tools of ever-increasing sophistication definitely help economize on time-to-market, assuming you understand your tool chain adequately, but it is still actual unit testing, in temp chambers and similar apparatus, that is your most loyal ally in catching flaws before production. Indeed, a large video graphics company I interacted with recently would run their newest cores through corner-case testing across something like 20 different axes. (Imagine a chip with several voltage domains, e.g. core, GPIO, memory, and some sort of GHz-domain serialized interface, now multiply the number of domains by additional factors such as operating temp and Si process variation, and you get a hypercube representing all possible corner cases.)

Testing across corner cases is exhausting, and yes, proper design discipline coupled with EDA proficiency give you a sporting change of avoiding a reduction in performance goals (at best) or even a re-spin (at worst).

But sending a bad chip to market is often a company-ending act for anyone not in the Fortune 500.