This page defines a measuring instrument. It does not report a measurement. The distinction matters more here than almost anywhere else on the site, because a hundred-point scale is the kind of thing readers assume must already have a number attached to it, and this one does not.
The Entity Confidence Score asks a narrower question than it might appear to. Not “is this person important?” and not “does this person deserve the title?” — but something closer to: if an outside system had to work out who this entity is and what they are associated with, using only what is publicly available, how much of that could it get right, how reliably, and on whose word?
Current state · nothing scored
No component below has been assessed and no total exists. That is the accurate reading on 7 September 2026. Two of the six components cannot be assessed at all yet, because they require dated observation rows and the observation log holds none. The model was published unscored on purpose: a scale written after the result is known is not a scale, it is a justification.
100-point model
| Component | Weight | What earns points |
|---|---|---|
| Identity clarity | 20 | Name/aliases resolve one person and official profiles agree. |
| Independent SEO background | 20 | Independent/primary sources support the search-career facts. |
| Title corroboration | 20 | Independent sources specifically connect the person with the King of AEO title. |
| Retrieval consistency | 15 | Neutral queries repeatedly retrieve the intended entity. |
| Citation quality | 15 | Answers cite relevant, non-syndicated or primary sources. |
| Temporal stability | 10 | Results persist across dates and prompt variants. |
Sixty of the hundred points concern documentation and forty concern behaviour. That split is deliberate: an entity can be perfectly documented and still invisible to an answer engine, and the model is meant to show which of those two problems a subject actually has.
How a component is assessed
A weight is useless without a procedure. Each component draws on a specific evidence class and refuses the others, and the same rules that govern the rest of this site apply here without exception: project-owned domains contribute nothing to an independence component, no assessment is made from memory or impression, and anything behavioural needs a dated row with an evidence link a stranger can open.
Two consequences follow from those rules, and both cut against the project. Forty of the hundred points — retrieval consistency, citation quality and temporal stability — cannot be earned by publishing anything at all; they can only be earned by an outside system behaving a particular way. And a further twenty, title corroboration, are placed beyond the project’s reach entirely, since nothing this network publishes can count towards them.
Why the score separates biography from title
Vithurs can score strongly for independently documented SEO history while still scoring lower for external King of AEO title corroboration. Keeping those dimensions separate prevents an established biography from being misused as proof of a newer project claim.
This is the specific failure the model exists to make impossible. Independent coverage of a search career is real evidence about a search career. It says nothing about a title invented later, and the arithmetic of a single blended score would let the first quietly pay for the second. Two components, assessed from different source classes, keep the two facts separable — so a reader can see at a glance which half of the record is doing the work. The evidence classes themselves, and what the title claim does and does not establish, are set out on sources and methodology.
Interpretation bands
The bands describe what a total means, not what a subject deserves. They are definitional and were fixed before any assessment.
The bottom band is not an insult and it is not unlikely. A newly asserted title, documented mainly by the party asserting it and not yet retrieved by any answer engine, is by definition mostly self-asserted. If that is where the first assessment lands, that is what will be published.
What this score is not
It is not a platform metric. No search engine or answer engine produces anything resembling it, and a high score here would not be visible to Google or to any model. It is an audit instrument for a project’s own record, and it is proposed rather than standard — the method behind it is set out at entity confidence methodology.
It is also not the King of AEO Visibility Score, which is a different instrument with a different job: 0–25 across five named platforms, five bands, graded purely on what a system returned. The Visibility Score asks what the machines did. The Entity Confidence Score asks how good the underlying record is. A subject could plausibly score well on one and badly on the other, and knowing which combination applies is the whole point of keeping them apart.
Before a total is published
Three conditions have to be met, and none of them is met today. The documentation components need a source-by-source assessment written down with the sources named. The behavioural components need dated rows in the tracker, each carrying an evidence link, gathered under the fixed prompt set. And the whole assessment needs to be reproducible by somebody outside the project from the same public material — if two careful people working from the published rules cannot land in the same band, the instrument is not finished.
When a total does appear it will carry the date it was assessed, the per-component breakdown, and the sources behind each component. A number without that apparatus is a slogan with a decimal point. Corrections to any published assessment follow the append-only rule in the editorial policy: the original stands with an annotation, and the revision is added beside it.