When I had the freedom to set my own schedule, I found my max productivity came by doing 2-3 days of 14-16 hours grinding, followed by 1-3 days of break where I didn't "work" at all. This corresponded to just how complex an idea I could fit in my mind at a sufficiently fine-grained level of detail to do real work on it.
Cal Newport's "Deep Work" is another approach to this, turning maximum productivity into a daily, scheduled system. I've also been influenced by "Four Disciplines of Execution", specifically the first discipline - pick no more than three Wildly Important Goals, and do not work on anything else. If it's not a Wildly Important Goal (and it's not an actual emergency), it deserves zero attention.
I've been lucky at my current gig to be able to speak this language, to talk about the cost of interruption. When you have one job, you can focus 100%. When you have two, you don't get 50/50, but more like 40/40, with the remaining 20 lost to context switching. And the more often you context switch (or "multitask", hahaha), the more focus and momentum you lose. I'm sure a lot of knowledge workers lose north of 50% of their available productivity to the thermal losses of context switching.