Assessments of a service ask whether it is secure and whether it is trustworthy. They rarely ask what becomes of the data if the company is sold, wound up, or simply stops. Those three endings have different mechanics and different consequences, and all three are ordinary.
The ending nobody assesses
Due diligence examines the present and assumes it continues.
It looks at the security posture, the certifications, the uptime, the funding. Every one of those is a statement about a company that exists. None of them is a statement about what happens when it does not.
Companies end routinely — bought, wound up, or simply switched off — and those endings receive a small fraction of the attention given to breaches, despite being a normal part of how the industry works rather than an exceptional failure of it.
Three endings, three mechanics
They are not the same event and they do not have the same consequences.
Acquisition. The data is an asset and transfers with the company. The terms you agreed were with a party that no longer exists in the same form, and the incentives that now apply are the new owner's. Nothing has been taken from you; the surrounding context has been replaced.
Insolvency. The data becomes an asset in a process whose purpose is to realise value for creditors. This is not misconduct — it is what the process is for — and that is exactly why it is worth anticipating rather than resenting. The obligations in force are to creditors, and you were not one of them.
Discontinuation. The service is switched off, usually with notice. Ownership does not change and no process is realising anything. The only question is whether you can get your material out, in what form, and before the interface stops answering.
People conflate these and then generalise from one to the others. Acquisition preserves the data and changes the rules. Insolvency turns it into an asset. Discontinuation threatens its availability. Preparing for one does not prepare you for the others.
Why the terms do not protect you as much as expected
Three structural points, each of which is a legal question in detail and a simple one in outline.
Successor clauses are standard, and they commonly permit transfer of accounts and data to a buyer or an affiliate. Agreeing to the terms usually included agreeing to that.
A promise made by one company is not binding on another in the way people assume. The commitments you relied on were made by an entity, and entities do not survive all three endings.
A wind-up process carries obligations to creditors that were never part of your agreement and take precedence over it.
I am describing a shape, not giving legal advice; the specifics turn on jurisdiction and on the documents. The shape is enough to plan around.
What determines whether you survive it
Three conditions, in increasing order of strength.
Can you export in a format something else can read? This is a different question from whether an export exists. A format nothing can open is a promise discharged, not a capability delivered.
Is the data usable without the service? This excludes anything whose value lives in the service's own index, graph or model rather than in the records themselves. Raw documents travel. A ranking does not.
Do you hold a key? This is the only condition under which an acquirer, an administrator or a buyer at auction inherits ciphertext rather than content. It is also the only one that does not depend on the company's cooperation at the moment the company has stopped cooperating.
The export that is not an export
An archive in a proprietary format. A dump with no schema. An interface that exports one record at a time, at a rate that would take weeks.
Each of these satisfies a stated commitment and meets no actual need. They are not necessarily designed to obstruct — an export nobody exercises gets built once and never tested.
The test is concrete and takes an afternoon: could you reconstruct your working state somewhere else, from this file, today? If you have not tried, you do not know the answer, and the moment you will need it is the moment support has stopped replying.
What to ask before depending on something
What happens to the data on wind-up.
Whether export includes everything or only the content you typed.
Whether there is a stated notice period before discontinuation.
Whether anything at all works locally, without them.
Four questions. Most of them have a published answer or a conspicuous absence of one, and the absence is itself informative.
What a responsible service publishes
A stated position on all three endings, not just the polite one.
A working export in an open format, which is to say one that somebody has actually round- tripped into a competing product.
A documented way to read or run the data without the service.
Publishing this costs something. It is an admission that the company is not permanent and a reduction in the friction of leaving. That cost is what makes it evidence rather than marketing.
Close
A service is not something you use. It is something you depend on, for as long as it exists, and it will not exist indefinitely.
The question worth asking early is not whether they are trustworthy now. It is what you are left holding on the day they are not there to be trusted.