Hmm, that should be interesting to evaluate physically...
My experience with external TB3 adapters for my NVMe drives wasn't too great.
I mostly used them to create backups of operating systems that were either important, needed to migrated to newer storage or actually were used as a template for other machines: cloning Windows has become so extremely easy, even Microsoft is using it for every installation.
But the first thing I noticed was that there was no USB fallback, just in case your host wasn't 100% certified as some of my AMD notebooks were.
The second thing was that the controllers only used two lanes at PCIe 3.0, which had things top out at 1GB/sec or quite the same speed that 10GBit USB sticks deliver the cost of a 2TB stick for that TB3 NVMe enclosure... economically TBx to NVMe is dead for bandwidth and capacity at this point.
And then my TB3-to-NVMe thingies just died within three of four usages, with absolutely no fuzz, which meant it took me a long time to figure out it was the bridge chip that was just plain dead and not some interoperability issue!
Native SATA to USB 10Gbit sticks deliver 1GB/sec bandwidth today and almost the same economy as a Samsung 970 Evo+ in terms of capacity: that makes it very hard to take a leap of faith into a tech that has failed so quickly and completely.
I'm rattled enough not to trust a real USB failback, so please do your very best and test these things within an inch of their life, so we don't have to suffer the slings and arrows of an outragous ...overdesign without reliability or economy?