This tool vs. checking vendor pages by hand
Names no specific company or product — this compares against the manual habit nearly every practitioner already has: opening a vendor's terms, privacy, or subprocessor page from time to time and eyeballing whether it looks different from what you remember.
What checking by hand actually does
Revisiting a vendor page by hand is free and needs no tool, which is exactly why it is the default. You open the page, skim it, and compare it against your memory of the last time you read it. If nothing jumps out, you move on assuming nothing changed.
That method has three failure modes this tool exists to close. First, memory is not a diff: a single clause rewritten in place — a retention window stretched, a notice period shortened, a subprocessor swapped for another with the same category label — reads the same at a glance as the version you remember, because nothing preserved the exact wording to compare against. Second, a failed check looks identical to a clean one: a page that timed out, redirected, or was blocked by the vendor's own bot protection shows you nothing, and "I saw nothing wrong" and "I saw nothing at all" produce the same blank result in your head. Third, there is no record afterward — no source URL, no timestamp, no before-and-after text — so six months later neither you nor a reviewer can reconstruct what the page said the last time anyone actually looked.
| Question | Where to answer it |
|---|---|
| I want to know if a vendor's page has changed at all, and I am willing to re-read the whole thing from memory. | Checking by hand — free, and works if the change is large enough to notice. |
| I want to know exactly which clause changed, word for word, without relying on memory. | This tool — a section-by-section, word-level diff between two captures. |
| I want to tell a genuinely unchanged page apart from a check that silently failed. | This tool — a failed or blocked capture is reported as inconclusive, never as "no change." |
| I want evidence I can put in a ticket: a source URL, both timestamps, and the exact before/after text. | This tool — the report is built from what was actually captured, not from what someone recalls. |
| I want to run scheduled checks without remembering to revisit the page myself. | This tool's CLI — fetches on your own machine, on whatever schedule you already run. |
Checking by hand is not wrong so much as unverifiable: a practitioner doing it carefully, every single time, would eventually catch most large changes. What it structurally cannot do is prove a check happened, prove what it found, or catch a change smaller than what memory retains — and this tool's whole design, on both the browser page and the CLI, is built around closing exactly those three gaps for free.
This tool needs no account, no hosting, and no server to try: the comparison page runs entirely in your browser, and the CLI's stored data never leaves your own machine.
Common questions
What does checking a vendor's page by hand actually catch?
It catches a change large enough to jump out against your memory of the page — a rewritten heading, a new paragraph, a page that looks obviously different in layout or length. It does not catch a single clause edited in place inside a page that otherwise reads the same, because nothing about a by-hand look preserves what the page actually said last time you read it.
Why is “I checked and it looked the same” not proof of anything?
Because a page that failed to load, redirected somewhere else, or was blocked by the vendor's own bot protection can look exactly like a page that never changed — you scroll, you do not see the change you were looking for, and you move on having actually looked at nothing. A by-hand check has no way to distinguish those two outcomes from each other, and no record afterward of which one happened.
Can I put a by-hand check in a ticket as evidence?
Only as a note that you looked, on a given date, and did not personally notice anything. That is weaker evidence than a reviewer, an auditor, or your own future self usually needs — it carries no source URL, no before/after text, and no independent way to confirm which version of the page you were actually looking at.
Open Vendor Change Watch to try your own vendor page.