Every business makes mistakes. Ours are not your problem to absorb
Any provider can promise nothing will go wrong. The more useful question is what has been built for the day something does, because that is a design decision made long before your case existed rather than a favour granted afterwards.
What can go wrong, and what happens then
Why the refund path is engineered rather than administrative
Returning money sounds trivial and is the place these systems most often fail. A refund involves a database, a payment provider and a network, and a process that dies between any two of those can leave you refunded twice, or refunded never with a record claiming otherwise.
So ours does not hold a database lock while it talks to the payment provider. It records the intention, makes the call with a deterministic key, then records the result separately. If the process is interrupted at the worst possible moment, running it again reuses that key and receives the same single refund rather than creating a second one.
None of that is visible to you when it works, which is the point. The reason to describe it at all is that it is the difference between a refund policy and a refund mechanism, and only one of those survives a bad afternoon.
The same discipline runs through the payment side. Charges are created with keys that make a repeated attempt harmless, so a flaky connection at checkout does not become two payments you have to chase us about.
How to go over somebody's head
Five steps, and you are entitled to start further up if the situation warrants it.
What we cannot fix, and will not pretend to
Being honest about the boundary is part of being useful inside it.
Why this page exists at all
Almost nobody in this field publishes what happens when they fail. The omission is deliberate, because describing your own failure modes invites the reader to imagine them, and a conversion-optimised page never invites that.
We take the opposite view for a practical reason. The people who need this service most are often dealing with a housing situation that is already going badly, and the last thing they can absorb is a provider who becomes unreachable the moment something slips.
So the escalation path is written down, the refund path is engineered rather than promised, and the limits are stated where you can read them before you buy rather than discover them afterwards. If that costs us a few sales to competitors with cleaner-sounding pages, it is a trade we are content with.