That product decision was very probably driven in a big part by Sinclair's (both the company and the man himself) fascination for aggressively simplified technical solutions and shifting the task from expensive hardware to cheap software solutions.
Following that principle, the company had achieved to come up with the first $100 "usable" computer and its descendant and megaseller, the ZX Spectrum. So, that approach was apparently working.
The very same principle was applied to try and solve the problem of the storage medium for the ZX Spectrum and planned later Sinclair computers (the QL): On the one hand, there was the cheap, but quirky to handle cassette tape, on the other hand, the elegant, but expensive random access possible with floppy drives - Sinclair apparently wanted something in-between and built it. They used relatively modern ULA chip technology to handle the direct-to-tape and sector locating logic and miniaturized the tapes and drives (I don't know of any evidence that Sinclair engineers were looking at, or even inspired by the Exatron Stringy Floppy, but it's likely).
The approach had, however, two major flaws: Mastering the technical challenges took too long while floppy drives and their media became cheaper and cheaper through volume uptake, and the cost of the media itself (while its storage capacity/cartridge could maybe be tolerated at least in an 8-bit computer) was already far from competitive with floppy disks. The storage capacity on the QL - with 100kBytes/cartridge on a computer that could be expanded to 640k of memory - was not really adequate.
By the time the Microdrives were somewhat "ready", third-party single floppy drives from Japan were reportedly already cheaper to have in quantity than the self-developed two-drive Microdrive system, but Sinclair (the person) stubbornly followed that path and nobody in the company dared to steer around or even tell him (Allegedly, there were secret contact attempts between Sinclair engineering staff and Japanese drive manufacturers that were hidden from Clive Sinclair). The QL might have turned out to be a totally different computer with a single floppy drive.
As I've already written elsewhere, the Microdrive reliability was much better than purported, at least after the system was fixed through a number of relative costly factory upgrades: I have some cartridges and drives for my QL that have survived nearly forty years of usage and are still doing great.