A vendor adds a subprocessor. The page updates. No email arrives, or one arrives in a monthly digest nobody opens. Nine months later an auditor asks how you review changes to your processors, and the honest answer is that you look when you remember to.
The control that closes this is unglamorous: dated snapshots, compared on a cadence, kept as a record.
Open the free vendor change comparison
Notification depends on the vendor's list being current, on the address they hold, and on the mail arriving somewhere a person reads. All three fail routinely, and none of them fail loudly. A snapshot you took is a fact you hold; a notification you did not receive is indistinguishable from nothing having changed.
The failure mode that makes a monitoring control worse than none is a clean "no changes" produced by a fetch that never happened. When a snapshot records a non-ok status — timeout, blocked, or anything else — the comparison reports inconclusive rather than stable, and says which side failed. A gap you can see is recoverable; a gap that reads as reassurance is not.
The tool reports evidence only. It does not decide whether a contract, a security posture, or a compliance status has changed — those are legal and commercial questions about your agreement.
Capture a dated snapshot of the pages you care about — privacy notice, terms, subprocessor list, security page — on a regular cadence, then compare each new snapshot against the previous one and keep the resulting diff as your review record. What matters for an audit is not only that you noticed a change but that you can show the two versions, the dates, and the date you reviewed it. Relying on the vendor's own notification is not a control, because notification is exactly what tends to fail.
Data processing agreements commonly give the customer a right to be informed of new subprocessors, and sometimes a window to object, because a new subprocessor can mean personal data reaching a new company or a new country. That right is only usable if you notice in time. A page that gains a line without an announcement is the ordinary case, so a dated diff is the practical way to keep the clock meaningful.
The report is inconclusive rather than stable. A snapshot that timed out, was blocked, or failed for any other reason is recorded with its non-ok status and the comparison says so explicitly. This matters because the dangerous failure mode for a monitoring control is a clean 'no changes' produced by a fetch that never happened.