When an Expert Leaves, Fix the Authorship Evidence Before AI Systems Learn the Wrong Version

When a named expert leaves a company, the content problem is not whether to erase the person's work. The problem is whether the page still tells the truth about who wrote, reviewed, or supplied expertise for the claim, and whether the machine-readable identity evidence still matches the visible record.

The safe operating rule is simple: preserve truthful historical authorship, update current-role claims, date the correction, and keep one owner accountable for the affected author page, bylines, schema, and external corroboration trail.

This is a maintenance checklist, not a claim that role changes cause or lose AI citations. AuthorityTech's demand signal is narrower. In the August 11-September 8 Google Search Console window captured in the September 11 daily brief, the existing authorship-credentials page received 6,394 query-page impressions and 0 clicks for the exact authorship and credentials best-practices question. Those are Search Console query-page impressions, not search volume, readers, turnover demand, citation events, or proof of buyer intent; Google's own Search Console Search Analytics API defines rows around requested dimensions and metrics such as clicks and impressions. They do show that the broader authorship-credentials question is exposed enough to deserve a more specific operator artifact.

The existing AuthorityTech brief on authorship credentials for AI visibility owns the general best-practices answer: named expert bylines, dedicated author pages, truthful Person schema, independent corroboration, and topic history. This page owns a different lifecycle moment: what to do after the author, reviewer, or quoted expert's current affiliation changes.

Authorship evidence has two dates: the work date and the current-state date

A byline is historical evidence. A job title is current-state evidence unless the page clearly says otherwise. Most authorship errors after a role change come from mixing those two dates.

A page published in 2025 may truthfully say that an expert wrote it while employed by the company in 2025. The same page may become misleading in 2026 if it still implies the person currently holds that role. The fix is not to delete the author. The fix is to make the time boundary visible.

Use three labels internally:

Evidence element	What it means	Role-change action
Original author	The person responsible for the original work at publication time	Preserve unless the original attribution was wrong.
Current reviewer	The person or team responsible for the current version	Update when the page is materially reviewed after the role change.
Current affiliation	The person's present relationship to the company	Remove, revise, or date-limit if no longer true.

That distinction matters because Google's helpful-content guidance asks whether content shows clear sourcing and background about the author or publishing site, including links to author or About pages. The guidance also warns against changing dates to make content seem fresh when the content has not substantially changed. In practical terms: a role correction should clarify the record; it should not manufacture freshness.

A useful correction note looks like this:

Authorship note: This article was originally written by [Name] while [role] at [Company]. It was reviewed and updated by [Current reviewer] on [date] to clarify current affiliation and evidence references.

That note preserves provenance. It also gives a machine and a human a clean way to separate original authorship from current editorial custody.

The role-change audit checklist

Run the audit against every page where the departing expert's identity is used as evidence. Do not limit the review to biography pages. Authorship evidence often appears in bylines, JSON-LD, quote blocks, contributor pages, speaker pages, case studies, press pages, and syndicated copies.

Start with this worksheet:

Surface	Question to answer	Safe correction
Article byline	Does the byline name the original author truthfully?	Preserve historical author if accurate; add reviewer/update note if current custody changed.
Author page	Does the page imply current employment, current responsibilities, or active availability?	Replace current-tense employment language with dated historical language or remove unsupported current claims.
Person schema	Do jobTitle, worksFor, affiliation, sameAs, and knowsAbout match the visible page?	Keep only claims the visible page can support; date-limit former affiliation in visible copy.
ProfilePage markup	Is the profile page still mainly about one person affiliated with the website?	Keep if the page remains a legitimate author/employee/alumni profile; otherwise revise the purpose or retire the page with a redirect.
Review blocks	Does a reviewer box imply the expert reviewed the current version?	Update reviewer identity or add a dated historical review note.
Quotes and expert commentary	Was the person quoted while in role, or is the quote independent of role?	Preserve the quote if accurate; clarify the capacity in which it was given.
Internal links	Do article pages link to a profile that now says something different?	Keep links only where the destination resolves the same person and explains the relationship.
External corroboration	Do LinkedIn, speaker pages, contributed articles, and outlet bios still support the same expertise claim?	Record which external sources corroborate expertise and which only prove past employment.
Syndicated copies	Do external reposts or partner pages repeat stale current-role language?	Request corrections only where the copy materially misleads; do not rewrite third-party history.
Measurement notes	Did the correction change tracked entities, URLs, or author identifiers?	Annotate the date so future visibility changes are not misread as performance movement.

The point is not to make every page say less. It is to make every page say the right thing at the right date.

What to update in structured data

Structured data should repeat the visible truth; it should not preserve a stale employment claim because the old markup looked clean. Google's Article structured-data documentation describes author markup as a way to identify the author, including with a URL or sameAs reference. Google's ProfilePage documentation says author pages, About Me pages, and employee pages can be valid profile-page uses when the page's primary focus is a single person or organization affiliated with the overall website. Schema.org's Person type includes properties such as affiliation, alumniOf, and sameAs that can describe the person, but those fields are only useful when they match the human-readable page.

Use this practical sequence:

1. Resolve the stable person identity. Keep the same person entity where the author is the same person. Do not create a new identity just because the job changed.
2. Separate former and current affiliation in visible copy. If the person is no longer employed by the company, do not let worksFor or current-tense biography copy imply they are.
3. Keep sameAs links only when they identify the person, not merely an old employer page. A LinkedIn profile, personal site, ORCID profile, author archive, or speaker page may help resolve the person. An outdated employer bio may prove history but should not be the only current identity link.
4. Use reviewer markup only for real current review. Do not mark a former expert as the current reviewer unless they actually reviewed the updated version.
5. Validate after publishing. Test the page, inspect the rendered JSON-LD, and make sure visible byline language and schema tell the same story.

Avoid the common shortcut: changing the visible biography while leaving the JSON-LD untouched. That creates two competing records. Machines may extract the stale one because it is cleaner than the prose.

Keep the expert's historical credit unless the credit itself was wrong

A role change is not plagiarism cleanup. If the person wrote the article, contributed original analysis, or reviewed the work at the time, historical credit should normally remain. Removing the name can damage provenance and make the page less trustworthy.

Use narrower corrections instead:

• Change "Chief Scientist at Company" to "former Chief Scientist at Company" only if that was the relevant historical role.
• Change "leads our AI visibility research" to "led AI visibility research at the time of publication" if that is the truth.
• Add "current review by" when another expert takes custody of the page.
• Add an update note when statistics, examples, or vendor references changed after the original author left.
• Remove unsupported credentials entirely when the company cannot substantiate them.

This distinction is especially important for pages that answer expert-sensitive questions. Google's helpful-content documentation emphasizes expertise, sourcing, and easily verified factual accuracy. An author page that overstates current affiliation is an easily verified factual error. An article that quietly removes the original author may create a different problem: it obscures who was responsible for the original work.

How to decide whether to redirect, preserve, or retire an author page

Do not delete an author page just because the person left. Decide based on the page's ongoing evidence role. There are three common outcomes.

Outcome	When to choose it	Implementation note
Preserve and revise	The person authored or reviewed durable work, and the profile still helps readers resolve responsibility.	Keep the URL, revise current-tense claims, add a dated note, and maintain links from the person's article archive.
Convert to alumni/former-contributor profile	The person no longer represents the company but has a meaningful historical body of work.	Make the former relationship explicit; avoid implying current employment or endorsement.
Retire and redirect	The page existed only as an employee profile, has no public authorship value, or creates a privacy/compliance problem.	Redirect to an author archive, team page, or relevant article only when the destination preserves the user's intent.

For AI visibility work, the first option is often strongest when the author has a real body of work. Machines and human reviewers benefit from a stable author archive. But stability is not a license to freeze the employment claim. The page should remain stable as an identity record, not as an outdated business card.

Measurement: annotate the correction so future visibility changes are interpretable

An authorship correction changes the evidence layer, so measurement teams should annotate it like a migration. If AI visibility, search impressions, or citation presence changes after the update, the team needs to know what changed in the source evidence.

Record at least:

• affected URLs;
• old and new author-page URLs if they changed;
• whether the byline changed, reviewer changed, or only affiliation language changed;
• whether Person or ProfilePage markup changed;
• whether external corroboration links changed;
• whether the page's dateModified changed for a substantive content update;
• when the evidence event happened, using an explicit timestamp format such as RFC 3339 where the system supports it;
• who owns the next review date.

For the metadata layer, use ordinary provenance discipline rather than a marketing note. The W3C PROV overview frames provenance as information about the entities, activities, and people involved in producing something, and DCMI Metadata Terms provides a stable vocabulary for common descriptive metadata. You do not need to expose an academic provenance graph to readers. You do need enough source, date, actor, and change information for a later reviewer to reconstruct what changed.

This is where Machine Relations discipline matters. The current Machine Relations Index release shows why answer-layer measurement needs visible denominators, source dates, and engine boundaries. That release does not prove an authorship correction will change citation behavior. It does model the habit operators need: preserve the observed object, the date, and the limitation so later interpretation does not overreach.

Apply the same posture here. A role-change update is an evidence-maintenance event. It may make the page more accurate and easier to resolve. It is not a guaranteed visibility lever.

The operating memo to send when a subject-matter expert leaves

The content owner should send a short, specific memo within the same week as the role change. The memo should not ask every stakeholder to "review the website." It should name the affected evidence surfaces and decisions.

Use this template:

Subject: Authorship evidence update for [Name]

[Name]'s role with [Company] changed on [date]. We are preserving truthful historical authorship while updating current-affiliation claims. Please review the affected surfaces below by [date].

1. Article bylines and reviewer notes: [owner]
2. Author/profile page visible copy: [owner]
3. Person/ProfilePage JSON-LD: [owner]
4. Internal links and author archive: [owner]
5. External profiles or syndicated copies needing correction requests: [owner]
6. Measurement annotation for AI visibility/search reporting: [owner]

Do not remove historical credit unless the original attribution was inaccurate. Do not leave current-tense employment or reviewer claims in place unless they remain true.

That last sentence is the quality floor. It protects the former expert, the company, the reader, and the measurement record.

FAQ

Should we remove an author's name when they leave the company?

Usually no. If the person actually wrote or reviewed the work, preserve truthful historical credit. Update current-role language, add a reviewer note when someone else owns the current version, and remove unsupported employment claims.

Should the author page stay live after an expert leaves?

Keep it live when it helps readers resolve responsibility for durable work. Revise current-tense employment language, make the former relationship clear, and keep links to the person's article archive. Retire or redirect only when the page has no authorship value or creates a genuine privacy, compliance, or user-intent problem.

What should change in Person schema after a role change?

Person schema should match the visible page. Review jobTitle, worksFor, affiliation, sameAs, knowsAbout, and reviewer fields. Do not let structured data imply current employment or current review when the visible record no longer supports it.

Does fixing stale author credentials improve AI citations?

Not by itself, and no public engine has documented a simple role-change citation rule. The responsible claim is narrower: accurate authorship evidence makes the page easier to evaluate and reduces factual inconsistency. Citation impact must be measured separately.

How soon should role-change authorship evidence be reviewed?

Review high-traffic, expert-sensitive, and actively cited pages within the same week as the role change. Lower-risk archive pages can follow in a scheduled cleanup, but any page that states current employment, current review, or current expertise ownership should be corrected promptly.