tl:dr:
Some CCRs are really bad clinkers while others work reasonably well. That's the same situation as anything coming out of China, though. They'll make anything you ask them to make at whatever price point you want it to meet. So read reviews and test reports on this (or any) radio. If they work alright and seem well built then don't worry about the architecture.
Long answer:
Don't get wrapped around the axle about SoC or similarly the terms direct conversion, direct sampling, SDR, hybrid, double conversion, heterodyne, etc. The words are thrown around indiscriminately. Basically a good radio is a good radio no matter its architecture while a bad radio will be bad regardless. The RDA1846 chip that most of these radios use can perform adequately if the design is sound otherwise. It's one particular frequency agile wideband RF chip but not the only by far. There's tons of them now, some that cover HF to SHF. One I'm familiar with is an Analog Devices part that can TX and RX on 2x2 MIMO from 70 MHz to 6 GHz. It's not something intended for cheap consumer HTs though. The chip is $300 alone and a drop-in SoM (System on a Module) is $1,600...
You might generalize saying it's easier to make a straight superheterodyne radio that performs fine but is less expensive. Mostly it's reliability and build quality that suffers. After decades of engineering churn there's not much fat left to cut in the fundamental architecture to significantly improve performance so all you're left with is using cheaper components. So it won't last as long being tossed around but the actual RF performance is indistinguishable otherwise. But such a radio is inflexible and doesn't lend itself to adaption to something else.
DSP techniques are still evolving and proprietary. So there's a lot more ways to screw up a direct conversion radio. Algorithms and firmware are where established companies (Motorola, Harris, Yaesu, Kenwood, whomever) can really distance themselves from competitors. Motorola is probably using direct conversion in the XPR7000 and EVX models and it works well. But then again no one other than Motorola knows really what is inside their ASICs to know for sure.
But the guts don't matter, they aren't able to bend the laws of economics, physics or engineering. What matters is where the rubber meets the road - does it work or not? And if they can do it other companies can, too. In fact Harris builds most of the high performance, high reliability radios for the military and one portable example, the AN/PRC-152 (Falcon III), is a highly frequency nimble (covers 30-512 MHz and 762-870 MHz without gaps) 5 watt TX SDR that can do several analog and digital modes, some of which (like APCO-25, amongst other things) were added with a firmware upgrade after the radio was accepted and put into service.
That sort of flexibility is why SDR is where it's at. The reason these CCRs proliferate is once a basic design is done they can adapt it to just about anything. That comes with up- and downsides. They can just make them faster and cheaper or they can add interesting form factors or features, whatever. And even heterodyne architecture have DSP, so the line isn't really that distinct to say this-or-that.
If you're still interested after that diatribe you can start digging into the engineering behind various architectures and RF generally here: https://analog.intgckts.com/wireless-receiver-architectures/