Blog / Guides
Five structural signals that tell an AI who you are
Being readable gets your words into the model. Being resolvable is what gets you recommended, and it comes down to five things in your structured data.
Most advice about AI visibility stops at readability. Can the crawler fetch the page, can it parse the text without running JavaScript. That is necessary and it is not sufficient, because a model that can read every word on your site can still fail to work out what your company is, what category it belongs in, or whether the brand mentioned on your homepage is the same brand mentioned in a directory.
That second problem is entity resolution, and it is what Entity Confidence measures. It is free on every graded Lantad report, it never feeds the composite score, and it is built from five structural signals worth one hundred points between them.
- Identity declared 25 pts An Organization or LocalBusiness node in your JSON-LD. Without it there is nothing for the rest to attach to.
- Clear name 10 pts A name on that node. Cheap to add and surprisingly often missing.
- Category declared 20 pts What kind of thing you are, via a type such as SoftwareApplication or a description that says so.
- Verified profiles (sameAs) 25 pts Links from your entity to the same entity elsewhere. This is how a model connects your site to everything else about you.
- Named founder 20 pts A person attached to the organisation, which anchors the entity to something independently verifiable.
Readable is not the same as resolvable
Consider two pages with identical prose. Both server render, both are fetchable, both would score well on parity. One ships an Organization node with a name, a category, profile links and a founder. The other ships no structured data at all.
To a crawler these are the same page. To a model trying to decide which company to name in an answer, they are not. The first is an entity with edges leading out of it. The second is a body of text that mentions a word.
Sample Illustrative, not a measurement of any real site.
Readable only
- server rendered, parity high
- no Organization node
- no category, no profile links
- a model reads words, not an entity
Readable and resolvable
- server rendered, parity high
- Organization with a name
- category declared, sameAs present
- a model can connect you to yourself
Why the weights sit where they do
Identity and sameAs carry twenty five points each, and they are the two that most often decide whether the rest matters. Identity is load bearing because there is no entity to describe without it. Everything else in the list is an attribute of a node that has to exist first.
sameAs earns the same weight for a different reason. It is the only signal that reaches outside your own site. Anyone can claim a name and a category on their own domain; a set of links to the same organisation on other properties is what turns a claim into something a model can corroborate. Name is deliberately cheap at ten points because it is trivial to add and its absence is usually an oversight rather than a decision.
sameAs is the one most sites skip
In our own scanning the pattern we see repeatedly is a site that has done the work on identity and category and then stopped. The Organization node is there, the name is there, and there is nothing linking it to the company's presence anywhere else. That leaves twenty five points on the floor and, more importantly, leaves a model with no way to tell that your homepage and your entry in some industry listing are the same organisation.
Adding it is a few lines. The requirement is that the URLs are properties you actually control or are actually listed on, because the point is corroboration and a link to somewhere you are not does the opposite.
Sample Illustrative, not a measurement of any real site.
Common shape
- "@type": "Organization",
- "name": "Acme Analytics",
- "description": "Analytics software",
- score 55, grade F
Resolvable shape
- "@type": ["Organization", "SoftwareApplication"],
- "name": "Acme Analytics",
- "sameAs": ["https://...", "https://..."],
- "founder": { "@type": "Person", "name": "..." }
How to read your own score
Entity Confidence uses the same grade bands as the main score, so ninety and above is an A and anything under forty is an F. Because the signals are weighted rather than counted, the count and the grade can disagree in a way that is worth understanding: three of five signals present can be anywhere from thirty five to seventy points depending on which three.
It is diagnostic only. It does not move your composite score, exactly like retrievability, because it measures whether AI can resolve you rather than whether AI can read you. Run a scan and it appears on the report for free, with the specific missing signals named.
See what AI can read on your site
Run a free scan and get a graded report of exactly what AI crawlers can and cannot read, with ranked fixes.