# Five structural signals that tell an AI who you are

> We re-measured the five structured-data signals behind Entity Confidence against 30 pages fetched in one day, removed two of them, and added two nobody was checking.

- Canonical page: https://lantad.co/blog/five-signals-that-tell-ai-who-you-are
- This file: https://lantad.co/blog/five-signals-that-tell-ai-who-you-are.md
- Last substantive update: 2026-07-26

## Key facts

- **Published:** 2026-07-25
- **Category:** Guides
- **Author:** Lantad
- **Length:** 1740 words
- **Takeaway 1:** Lantad scores Entity Confidence from five structured-data signals weighted 35, 25, 15, 15 and 10, all read from one identity node in the HTML a no-JavaScript crawler receives.
- **Takeaway 2:** Across 30 pages Lantad fetched as raw HTML on 26 July 2026, 24 returned JSON-LD: a logo appeared on 19 of those identity nodes, a sameAs on 16, and an absolute @id on 11.
- **Takeaway 3:** A named founder appeared on 5 of those 24 pages and appears nowhere in Google's Organization structured-data documentation, updated 15 April 2026, so Lantad removed the 20-point founder signal on 26 July 2026.
- **Takeaway 4:** Six of the 30 pages returned no JSON-LD at all, including developer.mozilla.org and theguardian.com, so whatever makes those pages worth quoting is not structured data on the page.
- **Takeaway 5:** Entity Confidence carries no letter grade from 26 July 2026: Lantad's stored capture of astro.build renders composite 88.8 grade B beside entity confidence 0, so three tiers computed from the signals replaced the bands.

## Summary

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 fetch that captures every word on your site can still leave a machine with no way to work out which company published it, what category it belongs to, or whether the brand named on your homepage is the brand named in a directory.

That second property is entity resolution, and it is what Entity Confidence measures, read from the [structured data](https://lantad.co/glossary/structured-data) already in your crawler-visible HTML. It is free on every graded report and it never feeds the composite score. This post is about the re-measurement we ran on 26 July 2026, which moved two signals out of the score and two in, and about the one number we deleted rather than replaced.

## 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, an absolute url, an @id and a set of sameAs links, the shape the [get-cited guides](https://lantad.co/how-to-get-cited) ask for. The other ships no structured data at all.

To a crawler these are the same page. To anything trying to work out which company is being described, they are not: the first declares one identity that other documents can point at, and the second is a body of text that mentions a word. That is the whole distinction Entity Confidence measures, and it is worth being precise about what it does not measure. Nobody outside an AI system can observe what that system does with a given property, so neither this post nor the report claims a signal earns you a mention.

## What 30 real pages actually declare

On 26 July 2026 we fetched 30 pages, one per site, as raw HTML with no JavaScript executed. They were picked by hand as widely referenced technical sources, publishers and well-known products, not sampled at random. 24 returned at least one JSON-LD block and are the denominator below. Six returned none at all, among them developer.mozilla.org and theguardian.com. Nine more sites refused the fetch with a 403, 404, 429 or a TLS reset and were excluded rather than counted as absent, because a refused fetch is evidence about the refusal and nothing else.

Two results reordered the whole score. A logo turned up on 19 of the 24 identity nodes and an absolute @id on 11, and Lantad was checking neither. A named founder turned up on 5, and Lantad was paying 20 of 100 points for it. Thirty pages picked by hand is enough to rank one property against another and nowhere near enough to support a sentence beginning "most sites", which is why you will not find one here.

## Why the weights sit where they do

Identity takes the largest single weight at 35 because it is the gate: every other signal is a statement about an identity node, and none of them can be true without one. It stays under half deliberately, so that the mere existence of a node is never most of the score. It now requires a type, a name and an absolute url together, which means a bare typed node with nothing on it no longer earns anything.

External identifiers take the largest non-gate weight at 25, because sameAs is the only property in the set whose own definition is an identity statement: [schema.org](https://schema.org/Organization) defines it as a reference page that "unambiguously indicates the item's identity". The stable identifier and the logo take 15 each. The @id earned its place on discrimination, appearing on 11 of the 24 identity nodes, closer to half the sample than anything else measured. The logo earned its place differently: 19 of 24 carry one, and it is the only property in the whole set for which any platform owner documents a consumer, since [Google's Organization documentation](https://developers.google.com/search/docs/appearance/structured-data/organization), updated 15 April 2026, lists it among 23 recommended properties and says the markup can influence which logo is shown in Search results and the knowledge panel. They are equal rather than ranked because ranking a commoner-but-documented property against a rarer-but-structural one would mean inventing a basis we do not have.

## What we deleted, and why sameAs kept its 25 points

The founder signal is gone. Three things had to be true at once for that, and all three were: it is rare among the pages we sampled at 5 of 24, below logo at 19, sameAs at 16 and @id at 11; the word appears nowhere in Google's Organization documentation and no platform owner documents anything that consumes it; and our own check was wrong. It credited any standalone Person node carrying a name, so an article byline in a Yoast-style graph scored as a founder. css-tricks.com reached 100 out of 100 on the strength of its post author, under a sentence claiming real people stood behind the entity. That is the kind of confidently wrong sentence this product exists not to write.

The Clear name signal is gone too, for a duller reason: across all 12 of our stored captures, name and identity had byte-identical presence. Name never once fired on a page where identity did not, because it was defined as a name on an identity node. Ten points that discriminated nothing, so it folded into the gate. sameAs kept its 25 and lost its old label. It used to be called Verified profiles, and nothing is fetched and no ownership is checked, so verified was a word the measurement had not earned. It is now read from the identity node only, and a repeated URL counts once.

## How to read your own score

There is no letter grade any more. Grade bands calibrated for a continuous composite were being applied to a five-item weighted checklist, which manufactures precision, and it put two grades on different scales on one screen: our stored capture of astro.build renders 88.8 out of 100, grade B, for the composite, next to 0 out of 100 for entity confidence, because that page ships no JSON-LD whatsoever. Both numbers are right and the pair reads as the report arguing with itself.

Three tiers replaced them, and each is computed from the signals rather than from the score. Not declared means no identity node in the HTML a crawler receives. Declared means one is present and complete, with nothing outside the page pointing at it. Anchored means it is present and carries at least one identifier that resolves outside the page. The tier has to come from the signals because 50 points is reachable in both the declared and the anchored configuration, so the number alone cannot name the state. It remains diagnostic: it does not move your [composite AI visibility score](https://lantad.co/glossary/ai-visibility), the report generates a JSON-LD block from the properties your page is actually missing, and the full derivation lives on the [methodology page](https://lantad.co/methodology).

## Questions and answers

**Does Entity Confidence affect my AI Visibility Score?**

No. It is diagnostic and never feeds the composite, exactly like retrievability. It is scored from the same crawler-view fetch the composite already used, so it costs nothing extra to produce and it is free on every graded report.

**Why did you remove the founder signal?**

Three reasons at once. It is rare among the pages we sampled on 26 July 2026, at 5 of the 24 that returned JSON-LD. The word appears nowhere in Google's Organization structured-data documentation, updated 15 April 2026, and no platform owner documents anything that consumes it. And our own check accepted any standalone Person node with a name, so an article byline scored as a founder: css-tricks.com reached 100 out of 100 on the strength of its post author.

**Are my old and new Entity Confidence scores comparable?**

No, and the report says so rather than letting you assume otherwise. A report scanned before 26 July 2026 renders the signal set it was actually measured with, its own weights, and a note that the number is on the old scale. Compare the signal lists rather than the numbers, or re-scan the page.

**Why does a description no longer count as a category?**

A description is prose, and prose is the thing structured data exists to replace. Counting any non-empty description made a property that 9 of 24 sampled pages declare in a machine-readable form look common, and it printed a sentence claiming the markup stated what the site was over what was actually a marketing line. A subtype, an additionalType URL, an industry code, knowsAbout, or a SoftwareApplication, Product or Service node all still count.

---

Lantad measures whether AI crawlers can actually read a page: it fetches as a non-rendering
crawler, renders as a browser, and reports the gap. Free scan, one URL, no signup.

Method and weights: https://lantad.co/methodology | All pages as markdown: https://lantad.co/md | Crawler policy: https://lantad.co/bot