Didn't Tony Northrup do a survey on card usage across thousands of people, hoping to pick out which brands might be reliable/unreliable? Or was that just SD cards? Just in my head? That's the type of data that would be useful in this sort of discussion.
Aside: I was going to write that YouTube's search function - and video search generally - is pretty poor, but really I'm complaining about most of the industry's content moving to video from the written word. Dagnabbit.
Found it:
It was just SD cards. Key conclusions:
1 - SSD drive formats appear to be more reliable than older memory forms.
2 - Aside from Transcend being an outlier with far more failure likelihood, the other brands were fairly close in terms of cards failing at some point during ownership (around 1/5th to 1/4th of people experienced failure in any given brand).
3 - The assumption that Sandisk and Lexar (now ProGrade, sort of) give better performance over other card brands was misplaced with SD cards. They both offered at least one failure experience to about 28 percent of users, which was beaten by Samsung, Sony and PNY. We don't have the data yet, but I see nothing so far indicating that brand is a good indicator of CFexpress card reliability.
A couple additional points:
4 - We do have anecdotal evidence of some card companies reacting better to user needs, such as service, communication and reacting with firmware updates. Speaking as someone who runs a small software company as a day job, I find that more indicative of likely future reliability than how loyal the brunt of photographers is to particular brands based on their experiences in the 2000s.
5 - Being involved in these reviews over the past couple of years, I've had the opportunity to speak personally with some interesting people inside these companies. I was able to communicate directly (connected via LinkedIn stalking) to the guy who led the small team who wrote the firmware for the drives first put in Western Digital chips that went into CFexpress cards. It was terribly enlightening, but even he didn't know which CFexpress cards had which chips in them afterward. SanDisk was obviously using them, but at the time there were only 2 or 3 parts makers churning out components specific to the market, so that same code was likely in others. The biggest thing I took away from him was an expanded view of the variables affecting speed, heat and reliability. There were many.
6 - I could be wrong on this, but a signal to me about the willingness to iterate the product is that the product line gets updated with some frequency. That's a cultural and organizational issue as much as a prioritization. My personal sense is that people avoiding some brands due to the small size of the companies is actually avoiding those firms staying most on top of the standard. But I've been talking to people inside those companies, asking questions, so I have some inputs telling me about those cultures that I don't think I expressed much or well in the reviews.