(to my knowledge)
You'll remember some OneNote discussions, people complaining about the stall of the desktop version (which remains available, for the time being).
There's a general risk that any software developer out there might rise their update prices, at any given time, tenfold; the same is true with them switching to a cloud and/or subscription model, and the real risk here - it was me who mentioned this some months ago on the outlinersoftware forum (or additionally even here, I don't remember); nobody else seems to care, or then feared in private - is that the function of the software in question could cease, by real technical reasons or just since the developer will have decided so, with any (minor or more probably) major Windows 10 update.
As for going from desktop to cloud, a blatant example is Correlate, the ONLY (then) really smart (available) file management system, (then and) now correlate.com, and, the fact which is utmost appalling: They don't give awy ANY info anymore if you don't "sign up with your corporate [or MS or similarly pre-checked, "genuine"] mail address".
At the time (version 6.6, in 2013, for example), it was 149$ plus VAT, i.e. near 200 euro for continuously-buggered European Union customers, and now, can't even know; I don't know either if people who had bought in time, have been able to continue their software beyond them "going cloud" or not, but whatever: They could "kill" their defunct desktop application anytime, or then MS (by pure technical means) could do it for them, again anytime.
Since we're speaking of "serious users" here - who else would buy a file manager at 200 bucks? -, this situation seems to be inadmissible to me.
It goes without saying that users could only fall into such a possible trap whenever the tool in question either changes your data format or then does add real added value to your data, in form of invaluable but scarcely exportable meta-data (as Correlate does; well, I don't know anything of the latter's exportability there).
(To give a counter-example: If you edit your video with one tool, and then there's something in their policy you're really unhappy with, you continue your cutting work, on this movie or with any other coming along, with any other editing tool on the market.)
What do people / prospects speak about indeed, it's the fallacy - "fact" for them - that you should not confide your data to some "little developers" - "who knows if they still will be in business next year?"; at the same time, even the biggest software developers in the world simply just kill their software ad libitum (ie just as it pleases them); examples abound, for that last 30 years or so.
Thus, it becomes obvious that the risk does not lie in your developer's being a 1-man shop or some multi-national conglomerate (which will certainly outlive the particular software they will have sold you, but that'll be no consolation for you), but in the fact that whenever you accept to comply with some dedicated file format, or some very special meta-data management, even if that will rise your productivity to some unheard-before level, you'll be
- at the mercy of the respective developer re possible price increases ("backed-up" by automatic W10 updates killing what you've got before), or
- overhaul of other factors (the tool going desktop > cloud or any other nuisance you are not going to accept), and that's not even speaking of
- stealing your data (which is even possible for desktop applications, but more probable for cloud "solutions").
Hence, what to do? I think that every such application which treats your data in a way that it'll be NOT possible to switch to any competitor, hassle-free and without losing (core or meta) data, should provide, up-front, transition tools (and which you could try up-front, too, i.e. onto some sample data) in order for you to verify that if there is any problem with that tool, you'll be free to leave, in some way or some other.
Of course, if that tool is spectacularly superior to its competition, that will be difficult to achieve, but at least technically, there should be some chance for doing so, and/or for competitors to "overtake" then, without too much trouble.
Some weeks ago, I said, in the outlinersoftware forum, upon the question "Exporting Ultra Recall database (any experiences, pls?)",
"Hint: UR doesn’t have a principal parent, and then secondary parents, but then, the respective table mentions them in a given, chronological order (of time of creation of the parentage, which for the very first parent is identical to the time of creation of the item), so the first occurrence there could be treated as the parentage, further occurrences as links, even in tools without cloning (like RightNote)." - which was a little bit simplified, of course (the outlinersoftware forum not being a technical one, far from it), since
- not in all cases, the very first occurrence of some info item will be it's "natural" one; sometimes, it'll be some (later-created) "clone" which really will be the "core" occurrence (among other examples, you first write down some "ToDo", then only create the "real", "original" subject within its ontology)
- I (on purpose) didn't address the fact over there that such "original" items then could have become renamed and, in particular, moved within their tree in-between (but these are just minor, technical, "problems"; the necessary data, for analyzing all this correctly, in that case, will be all there).
Whatever: My point being: Especially the "exotic" data formats, or (meta-) data re-arrangements, should be perfectly "open", and then, it wouldn't matter that much if you confide your data to some - reliable - "lesser" developer, in view of the fact that the "big shots" don't treat you any better in the end:
- You should be confident that your data isn't at risk (in both respects: you not losing your data, and any third-party not getting access to it), and
- You should be confident that for your data, there will be some manageable way out.
If I have left out valid considerations / aspects, please add them.