01 / Name the publisher
Begin with responsibility, not appearance
Open the result and identify who publishes the page, what kind of page it is, and whether it names an underlying source. A polished card, prominent ranking, or confident snippet does not establish editorial responsibility.
Distinguish a primary publisher, a directory, an index, and a republisher. That label determines where a correction belongs and how much independent weight the page deserves.
02 / Follow the trail
Look for the earliest supportable record
Compare distinctive phrasing, dates, field order, and outbound links. When several pages repeat the same structure, they may descend from one shared feed rather than corroborate one another.
Stop when the trail reaches a responsible source or becomes untraceable. An unknown origin should lower confidence; it should not invite a guess about where the information began.
Working principle
Repeated text can show distribution without showing independent confirmation.
03 / Mark transformations
Note what changed between source and preview
A search engine may shorten a title, select a query-responsive passage, or display structured data supplied elsewhere on the page. Record those transformations so the preview is not mistaken for a quotation written by the publisher.
If the underlying page has changed, capture only the date and the category of difference. A privacy-safe provenance record does not need to preserve the obsolete personal detail itself.
04 / Act at the right layer
Send corrections to the surface that can resolve them
Ask the publisher to correct live source material. Ask the search provider to refresh a stale preview only after the source is accurate. Ask a downstream directory to update its copy when it operates an independent record.
One carefully mapped supply chain can replace many unfocused requests. It also shows which residual pages are likely to update automatically after the originating record changes.