How the work is actually done.

Every engagement moves through the same three steps, in the same order. No step is optional, and nothing reaches a publisher until all three are complete.

Identify. We read the article or resource page the way an editor would: what it claims, what it links to, and where it has drifted since publication. We look for four specific kinds of problems: statistics and figures that have been superseded; references that are broken, dead, or missing, including a work that is mentioned with no reference a reader could follow; statements of fact that are wrong; and material gaps that leave the piece misleading as it stands. Each candidate problem is logged with the exact passage, the URL, and the date we reviewed it. Every engagement starts from a genuine problem on a real, existing page. We do not manufacture pretexts, and we do not contact a publication simply because a client wishes an article read differently.

Verify. Before we suggest anything, we substantiate it independently. For every candidate correction we confirm the current fact against the sourcing hierarchy below, record precisely where the fact comes from, and, where more than one source exists, check that the sources agree. We do not take our client’s word for a fact; a client’s claim is a starting point for verification, not a substitute for it. Anything we cannot substantiate is set aside. It is not softened into a hedge and sent anyway; it simply does not go out.

Suggest. We write to the publisher with a specific, sourced suggestion, addressed to the desk responsible for the piece where we can determine it. Every message carries our positioning statement, which discloses that we partner with the organizations whose work is covered. From that point the publisher holds full editorial control. Whether the editor applies our wording, rewrites the fix in the publication’s own voice, or declines entirely is their decision, and we accept it either way. Every message is reviewed and approved by a person before it goes out, and a person reads the replies.

When we say a fact is current, we mean we can show an editor where it comes from. We accept evidence in a strict order of preference.

• Primary documents and official records first: government and court records, regulatory and corporate filings, official registries and databases, and statements published directly by the organization or agency that owns the fact.
• Established reference works and peer-reviewed material: recognized reference publications, academic journals, and the publications of standards bodies.
• Reputable published reporting: established news organizations with named reporters and editorial oversight. We use reporting when primary confirmation is not available, and we identify it as reporting rather than presenting it as a record.

Some things we will not rely on, at any position in that hierarchy:

• Unverifiable claims, including claims made by our own clients that we cannot independently confirm.
• User-generated content: forums, openly editable wikis, social media posts, and review sites.
• Machine-generated summaries or AI-produced answers. We use software to find sources; a tool’s summary of a source is never itself a source.
• Anything we cannot check ourselves. If the document behind a citation is paywalled, archived, or offline, we obtain and read it before we cite it, or we do not cite it.

When sources conflict, we prefer the more primary and the more recent, and we say so rather than papering over the disagreement. If the honest answer is that a fact is unsettled, we do not send a correction about it.

When we ask a publisher for a factual correction, the request is built the same way every time. The editor gets six things:

• The exact passage at issue, quoted verbatim, with the URL of the page it appears on.
• What is outdated or incorrect, stated specifically: not “this section has problems,” but which figure, date, name, or claim, and why it no longer holds.
• The current fact, stated plainly.
• The source or sources, cited so the editor can check them in minutes: a working link, a document reference, or a case or docket number, never “our research shows.”
• A suggested wording, offered for the editor’s judgment. It is a starting point, and we expect editors to rewrite it or discard it.
• A plain statement that we partner with the organizations whose work is covered, so the editor can weigh our note with our interest fully in view.

A correction request that cannot be completed to this standard does not go out. There is no shorter version, no teaser, and no “we found several issues, reply to learn more.”

How we contact publishers matters as much as what we send. These are the rules every message follows.

• One clear message rather than volume. We write to a publication about a specific problem on a specific page. We do not send mass form emails with a name filled in, and we would rather send one well-documented note than twenty vague ones.
• Every email carries our physical mailing address (99 Wall Street, Suite 5515, New York, NY 10005) and a working opt-out.
• Opt-outs are honored promptly (within ten business days at the very latest) and permanently. An editor or publication that opts out goes on a permanent do-not-contact list that applies across every domain and campaign we operate, and is not contacted again.
• If a message goes unanswered, we may follow up at most twice, spaced days apart, and then we stop. Any reply ends the sequence entirely; a no is a complete answer.
• Honest subject lines and a reply-to address that a person actually reads.
• We never pressure, never demand, and never pay for changes or placements.

A firm that asks editors to correct the record has to accept correction under the same rules. If we get something wrong, if a fact we sent was inaccurate, a source did not support the claim, or an update we suggested turns out to have been mistaken, we say so plainly and fix it promptly. That means contacting the editor we wrote to, identifying the error specifically, and either supplying the correct, sourced fact or withdrawing the suggestion outright.

We call a correction a correction. Not a clarification, not an update, not a revision for context. If an editor, a reader, or a client believes we have sent something wrong, our contact page reaches us, and a note pointing out an error is treated as urgent.

If we cannot source it, we do not send it.

Everything on this page reduces to that sentence, and it is the standing promise behind every message we send. Not “we send it with a caveat.” Not “we send it anyway and let the editor sort out whether it holds.” A claim we cannot substantiate to the standards above never leaves this firm. That rule means some suggestions we are fairly confident about never go out, and it is the reason an editor can open an email from Content Integrity knowing that whatever is inside can be checked, in minutes, against a source that holds.