As someone super into retro computing now, who also lived through the history getting my first computer in 1981, participating in users groups, reading all the zines, going to regional and national trade shows for 8-bit home computers, etc, this was (and still is) a significant recurring theme. Owners of 8-bit and 16-bit computers in the 80s/early 90s (and retro hobbyists/collectors today) obsess about the mistakes made by their respective manufacturers. Often this armchair analysis results in frustrated conclusions like "If only (Atari/Commodore/Tandy) had done X" the computer would have "made it" (ie survived against the PC). I engaged in this prevalent pastime myself for years and even today there are forums full of questions seeking to post mortem the "the fatal mistake(s)".
The fascinating part for me is that my early hobby computing led to programming, product-making and tech startup entrepreneurship. By the mid and late 90s I'd evolved from sneaking into Comdex and CES circa 1983 to now launching products in my startup's huge Comdex booth literally next to Microsoft's, with our own private hospitality suites and VIP parties. I'd somehow gone from fanboy to insider. During this era I had multiple private business meetings with industry luminaries I'd idolized including Bill Gates, Steve Jobs, Andy Grove and many others. I'd even met, had a meal with or otherwise gotten to personally know 8-bit era luminaries including Nolan Bushnell, Jay Miner (designer of the Amiga), a couple of Tramiels and so many others. I got to ask my fanboy era questions directly to the players - and the best part was that I was then a peer and the meetings were private. I got a lot more color and perspective than the answers they gave to the press or even at retro-computing conference panels. It was fascinating! But overall I also learned that, aside from a couple notable exceptions, there was no single "Big Mistake". The reality was far more complicated and kind of boring: there were complex business, financial, manufacturing and distribution challenges far beyond users group hobby analysis. The future was unknowable and they were living through the first iteration of what only later became obvious patterns - and they were just doing the best they could to figure it out enough to survive.
By the late 2000s I'd graduated to selling my third startup to a Fortune 500 valley tech giant and became a senior exec driving strategy and identifying new opportunities and markets, as well as leading large teams of MBAs, bankers and lawyers to identify, analyze and negotiate multi-billion dollar acquisitions of other tech companies. And this perspective is the one that most informed my prior obsession with identifying "the fatal mistake" which doomed Atari, Commodore, Tandy, Sinclair and so many others. Having recently retired early, just for fun I've delved into the history and using my post-2000s expertise done rigorous strategic analyses.
In every case the answer is the same and, frankly, a little disappointing: Nothing they could have done would have "saved" them for more than a couple years. While Atari, Commodore, Tandy et al did make many mistakes, none of them were the root cause of their eventual demise. Even if those mistakes had been avoided, it would only have delayed the inevitable. In each case, macro factors beyond their control that were baked into the market, the technology or their own DNA, doomed them. For example, one of the Amiga's greatest advantages in 1985 was the brilliant custom chipset designed to exploit every quirk of analog video timing. And by the early 90s that great advantage was one of its biggest weaknesses. Also, the much-beloved 68000 series processors at the heart of the Amiga and Atari ST were ultimately doomed by the combination of being made by Motorola and being, perhaps, the ultimate expression of CISC architecture (which made them fun to program by hand in assembly language). But RISC was ultimately the only way forward when Moore's Law scaling kept delivering ever more gates into the 90s. But bridging over from CISC to a RISC ISA while maintaining backward compatibility was enormously complex. Only Intel eventually managed it and very nearly died in the attempt. Intel's lead in process fabrication helped them over the hump but Motorola was too far behind in fab tech because they hadn't invested as deeply. For Intel it their lifeblood and, ultimately, existential. For Motorola's board of directors, CPUs were just another business in their portfolio of businesses. Motorola was a decades old conglomerate that made prudent financial calculations. Intel was born as a chip startup and would either live or die as one. Andy Grove had to bet the company and find a way to make it work. Motorola didn't.