No-Logs and Non-Custodial: What They Mean and How to Check
"No-logs" and "non-custodial" are the two phrases every privacy-oriented swap puts front and center. They are not the same promise, they are not equally verifiable, and understanding the difference tells you a lot about how much to trust any given service. Here is how to read both claims critically.
Non-custodial: verifiable by design
Non-custodial means the service never holds your funds as a balance you have to withdraw. You send coins in, the conversion happens, and the output goes straight to an address you control. There is no account wallet sitting in the middle.
The important thing is that this claim is structurally checkable, because it changes how the swap behaves:
Custody is where the biggest risk lives — frozen balances, exit scams, and honeypots all require the service to hold your funds first. A genuinely non-custodial flow removes that entire category of risk, and you can recognize it by the absence of a stored balance.
No-logs: a trust claim, not a proof
No-logs means the service says it does not retain records tying deposits, withdrawals, addresses, and IPs together over time. This is valuable — but unlike non-custodial, it is not something you can directly verify from the outside. You cannot inspect a server's retention policy from your side of the connection. Treat it as a statement of intent, and weigh the surrounding signals:
The honest framing: non-custodial protects you structurally; no-logs protects you only as far as the operator is truthful and competent. Both matter, but you should lean on the one you can check.
How to reduce your dependence on the claim
Rather than trusting "no-logs" absolutely, shrink what a log could even reveal:
The bottom line
When you evaluate a swap, verify the part you can: is it truly non-custodial, with no account balance and no withdrawal step? Then judge the no-logs claim by data minimization, onion support, and track record — and, either way, protect yourself with Tor and minimal disclosure so the claim matters less. A service worth using should make the verifiable promise cleanly and give you the tools to not have to take the unverifiable one on faith. You can see how this works in practice on the XMRS onion.
