Ahrefs data shows sites with author schema are nearly three times as likely to appear in AI answers. The standard advice follows: build named expert authority. It works, and almost nobody discusses what the resulting asset is attached to.
Entity relationships link a person to an organisation. They do not merge them. The schema property is literally called worksFor, which is a statement about a relationship rather than about ownership.
Two years of building someone into a cited authority produces something real. Some of it stays when they leave, and a meaningful portion walks out with them.
Key Takeaways
- Authority attaches to the person: the organisation link is a relation, not a merge.
- The split is asymmetric: they keep their entity immediately, your association fades slowly.
- Institutional artifacts transfer: named research, datasets, and tools carry the organisation.
- Concentration is the risk factor: one cited expert is a single point of failure.
- The answer is more people, not fewer: underinvesting in named experts costs more than the exposure.
Why the Authority Sits With the Person
Models build entity records for people the same way they build them for companies: from repeated appearances across independent sources. A named expert quoted in trade publications, bylined on articles, and cited in research accumulates a record that resolves to them.
Your organisation appears in that record as an attribute. It is how they are described, not what they are. When the attribute changes, the entity persists and the description updates.
This is not a flaw in how the systems work. It reflects something true: expertise does belong to people. The problem is that content programmes are frequently budgeted as if the resulting authority were a company asset, because that is how content usually behaves.
What Actually Happens After a Departure
The departing expert keeps their entity intact from day one. The company's association with their work decays gradually, over the months in which new sources describe them at their new organisation.
Nothing breaks immediately, which is why this rarely gets noticed. Their existing articles remain on your site and continue to be cited. Their quotes in past coverage still name your company.
The change happens through accumulation. As they publish and get quoted at the new organisation, the more recent sources describing them carry a different employer. Models weight recency for factual attributes, and the same citation freshness dynamics that govern product information govern employment details, so within a few quarters the association has moved.
What remains on your side is the archive: content they wrote, coverage that mentioned your company alongside them, and any material published under the organisation's name rather than theirs.
What Transfers and What Stays
| Asset | Goes with them | Stays with you |
|---|---|---|
| Their personal entity record | Entirely | Nothing |
| Bylined articles on your domain | Attribution | The content and its citations |
| Press quotes naming both | Their side | Historical association |
| Research published under the company name | Nothing | Entirely |
| Proprietary datasets and tools | Nothing | Entirely |
| Their audience and relationships | Entirely | Nothing |
| Category vocabulary you established | Partly | Partly |
The pattern is clear enough to plan around. Anything attributed to a named individual moves with them. Anything attributed to the organisation stays, which is an argument for a specific kind of output rather than for less named expertise.
How Exposed Are You
Concentration determines the size of the problem, and most companies have not measured it.
| Structure | Exposure | What to do |
|---|---|---|
| One cited founder or expert | High | Add named experts; publish under the org |
| Two or three named voices | Moderate | Ensure coverage does not concentrate on one |
| Several across functions | Low | Maintain schema and attribution hygiene |
| Institution-led, few named people | Low but capped | Add named experts; author entities lift citation rates |
The last row matters as much as the first. A company with no named experts is not safe; it is forgoing a substantial citation advantage. The objective is distribution, not avoidance.
To find where you sit, run your category questions and record which of your people are named, and how often. If one person appears in most answers that mention your company, that is your concentration figure, and it is the kind of finding that visibility tracking surfaces only if you are looking for it.
Building So the Organisation Retains Value
Four practices, in rough order of return.
Publish institutional artifacts
Original research, datasets, benchmarks, and tools published under the organisation's name accumulate citations that do not depend on any individual. These are the assets that survive departures intact, and they are the same assets that earn the third-party mentions and authority that drive citation in the first place.
Distribute named expertise
Three cited people is meaningfully more resilient than one, and the citation benefit compounds rather than divides. Coverage that consistently routes to a single spokesperson is a choice, usually made for convenience.
Co-attribute deliberately
Where an individual is the author, the organisation should appear in the same structured record. Person schema linked to Organization via worksFor, and the organisation referencing the person in return, builds the relational depth models use to resolve both.
Keep the archive functioning
Bylined content by former employees should stay published. Removing it destroys citations you still hold, for the sake of an attribution that has already moved. Where the content is materially out of date, update it and note the revision rather than deleting it.
When Someone Does Leave
The instinct is usually to remove or rewrite. Both are worse than the alternative.
Leave the content published and attributed accurately. Add editorial notes where content has been reviewed or updated by someone still with the company, which creates a current attribution alongside the original.
Update schema so the organisation's structured records reflect current staff, and let historical attributions stand. Contradicting the public record does not help, since third-party sources will continue to describe the past accurately regardless.
Then look at where the concentration was. A departure that materially changes what models say about your company is a diagnostic result, and the response is building the distribution that was missing rather than replacing one spokesperson with another.
The Argument Against Overcorrecting
This is not a case for avoiding named experts, and reading it that way would be expensive.
Author entities produce measurably better citation rates than generic bylines. Named experts quoted across publications create the cross-verifiable relationships models rely on. A company that responds to departure risk by publishing everything anonymously forfeits that entirely and gains very little.
The exposure is real and manageable, and the mitigation is structural rather than defensive. Our AI Recommendation Gap report covers how much of what models say depends on sources outside your control, and named individuals are simply the most mobile version of that dependency.
If You Want to Know Where You Stand
The audit is a short one. Run your core category and brand questions, record which people are named alongside your company and how often, and note which of your published research and tools appear without any individual attached.
The ratio between those two answers is your exposure. If everything routes through one name and nothing carries the organisation's, you have a concentration problem that a single resignation letter would make visible.
That measurement is part of what Lureon runs at the start of an engagement: which sources produce your answers, which entities they attach to, and where the dependencies sit.
FAQs
1. Does AI visibility built around a founder transfer to the company?
Partly. The organisation appears as an attribute of the person's entity record rather than absorbing it, so the individual keeps their entity while the company retains the archive: their published content, historical coverage naming both, and anything published under the organisation's name.
2. How quickly does the association change after someone leaves?
Gradually, over the months in which new sources describe them at a different organisation. Nothing breaks immediately, which is why the shift usually goes unnoticed until several quarters have passed.
3. Should we delete content written by former employees?
No. Removing it destroys citations you still hold, in exchange for an attribution that has already moved. Keep it published, update it where it is materially out of date, and add a review note attributing the update to someone currently with the company.
4. Does this mean we should avoid building named expert authority?
No, and doing so is more costly than the exposure. Author entities produce better citation rates than generic bylines, so the mitigation is distributing authority across several named people and publishing institutional artifacts, not withdrawing from named expertise.
5. What kind of content stays with the organisation?
Original research, datasets, benchmarks, and tools published under the company name rather than an individual's. These accumulate citations independently of any person and are also the assets most likely to earn third-party coverage.
References include Ahrefs data on author schema and AI answer appearance, and published guidance on entity schema relationships. How quickly associations shift varies by publication frequency, sector, and the individual's public profile.