A recent audit of B2B companies surfaced a number that is traveling fast through marketing circles: when an AI answer engine quotes a named person rather than a company or a forum thread, that person had published on LinkedIn 74% of the time. It is already being read as a story about a platform, and the reflex takeaway is the predictable one, post more on LinkedIn. That reading misses what the number is actually telling you. The real signal is bigger and more useful: citability has gone personal. AI engines do not just cite brands and domains. They cite people, by name, and a person can be built to be cited the same way a brand can.
What the audit found
The audit, run by the B2B platform Oktopost, is their own snapshot rather than a peer-reviewed study, and worth reading as such. Across 63 companies they tracked more than 27,000 citations from AI answer engines. Social sources were a minority of the total, around 10%, a reminder that most of what these engines cite still comes from owned content, review sites, and press. But inside that social layer, one split is striking. Posts by named individuals accounted for 32.8% of social citations. Company pages accounted for 6.7%. The model reached for the attributable named person nearly five times more often than for the faceless brand page. And when it quoted a named person specifically, that person was on LinkedIn 74% of the time, far ahead of Medium, Substack, Quora, or X.
It is not the platform. It is attributability.
Ask why a model would prefer a named source in the first place, and the platform explanation falls away. An answer engine assembling a response wants to attribute a claim to something accountable for it. "According to Maria Rossi, head of product at [company]" is a clean, defensible attribution: a name, a title, a person who can be held to the statement. A company page speaks for the brand in aggregate, and aggregate is harder to attribute and easier to distrust. The model is not rewarding LinkedIn. It is rewarding the property LinkedIn happens to carry in B2B, a resolvable individual with a title, a topic, and a history of saying related things under the same name.
That property is what makes a named voice citable, and it is structural, not social. A person the model can resolve to a clear identity, connect to a consistent topic, and attribute a claim to with confidence is a citable entity. A person publishing under an inconsistent identity, on scattered topics, with nothing to corroborate the claim, is not, no matter how often they post. LinkedIn wins the 74% because it is where that resolvable, attributable expertise lives for business audiences, not because the feed has magic in it.
A model does not cite a platform. It cites an entity it can name, resolve, and hold accountable. LinkedIn is just where that entity currently lives.
Citability goes personal
Everything we have argued about brand citability applies, without modification, to people. The Citability Score measures whether an entity is structurally built to be cited: whether a model can resolve it cleanly, trust the signals around it, and find a reason to attribute an answer to it. That entity has always been able to be a brand. It can just as easily be a person. A named expert is an entity in exactly the sense the model cares about, and the same structural questions decide whether that expert gets cited: is the identity resolvable, is the expertise consistent and corroborated, is the claim specific enough to be worth attributing.
This is the same lesson we saw when LinkedIn's own Hiring Assistant was shown to rank people by inferred, structured signal rather than by keyword repetition, a Citability Score for profiles. The system that ranks a person and the system that cites a person are asking the same underlying question: can I resolve this individual to a clear, trustworthy entity. One decides whether they surface in a search. The other decides whether they get quoted in an answer. Both reward the same structural clarity.
You cannot out-post your way in
Here is where the easy takeaway goes wrong. The volume reading, get a dozen executives posting consistently, treats citability as an output of frequency. The data does not support that. What earns the citation is not how much a person posts but whether they are a resolvable, topically coherent, credible entity on the question being asked. Ten posts a week from someone with no consistent topic and no corroborated authority do not create a citable entity. They create noise under a name. Frequency without structure is the personal-scale version of flooding a domain with generic content, and it fails for the same reason: it gives the model more to ignore, not more to trust.
The move that works is the structural one. A named expert becomes citable by owning a specific topic consistently enough that a model associates the person with it, by being corroborated across surfaces rather than asserting authority in a vacuum, and by making claims specific enough to be worth attributing. That is personal citability, and it is measurable with the same instruments as brand citability.
What this changes
For a decade, visibility work optimized brands and domains. AI answer engines are now quoting individuals by name in the answers your buyers read before they ever reach you, and that extends the entire discipline down to the level of the person. Your experts are not just contributors to a content calendar. Each one is either a resolvable, attributable entity the model trusts enough to quote, or a name the model cannot place. The question is no longer only whether your brand is citable. It is whether the people who speak for it are. On the queries that matter in your category, those named voices are being pulled into the answer three times out of four when a person gets quoted, and whether yours is among them is a structural property you can measure and build, not a matter of how loudly anyone posts.