Is your schema accurate and supported, or only valid?

This free schema evidence audit reads your JSON-LD, Microdata or RDFa and sorts what it finds into five lanes that most tools blur together: syntax, entity relationships, factual consistency, Google feature support and AI citation effects. It says what it could check, what needs the visible page, and what is not established, and it never gives a score. It also lists markup you could remove and a checklist to work through. Nothing is uploaded.

Google requirements copied from Google's documentation on 23 October 2026, the same rule set as the schema markup validator. Vocabulary: schema.org version 30.1.

Part 1

Audit your markup

Paste markup or a page's source, or send a live page with a bookmark.

1. Is your schema accurate, connected and supported, or only valid?

Paste JSON-LD, a script block or a page's full source. The audit separates what can be checked from markup alone from what needs your visible page, and ends with a remediation checklist you can export. Only need a quick syntax and vocabulary check? Use the schema markup validator.

How do I copy a page's source?
  1. Open the page in Chrome, Edge or Firefox.
  2. Press Ctrl+U (Cmd+Option+U on a Mac) to open View Source.
  3. Select all, copy, and paste it into the box below.

Confirmed View Source shows the HTML the server sent, before any script runs. To include markup that JavaScript adds, use the bookmark in question 2

Directional The rules copy Google's documentation as of 23 October 2026 and cannot see Google's own decisions or your visible page; confirm with Google's Rich Results Test before you rely on a feature

2. How do you audit a live page, including markup that JavaScript adds?

Use a bookmark. It runs on the page you are viewing and sends the rendered page to this tool in your own browser, so the audit sees what a browser sees. The tool never fetches an address itself.

  1. Drag Audit this page to your bookmarks bar, or and paste it as the address of a new bookmark.
  2. Open the page you want to audit and click the bookmark. This tool opens with the page filled in.
  3. Press Run audit.
What it sendsThe page's HTML to this tab only, by browser messaging; nothing passes through a server
Why not paste an address?A browser page cannot read another site's HTML, and fetching addresses on a server needs safeguards against misuse that this free tool does not have
Where it failsMobile browsers, and sites whose security policy blocks bookmarks; paste View Source instead
Comparing server HTML with JavaScript changesThe schema markup validator has a bookmark that does this

Part 2

Read the results

What each lane can and cannot tell you, and how this differs from other schema tools.

3. What do the five lanes mean?

Each lane answers one question with one kind of evidence. A pass in one lane never stands in for a missing answer in another.

LaneA good result meansIt does not mean
1. Syntax and vocabularyEvery block parses as JSON, and type and property names exist in schema.orgFull Schema.org validation, Google eligibility or that the content is true
2. Entity relationshipsReferences resolve, each @id names one thing, and no duplicate entities or identity clashes were foundThat the entities are the right ones, or that a search engine will join them
3. Factual consistencyNever "pass". It lists the claims to compare with your visible page and flags placeholders, bad dates and self-serving reviewsThat any claim is true. Only the visible page, and you, can show that
4. Google feature supportThe required properties for a current Google feature are present, and retired features are namedThat a rich result will appear. Google decides each time
5. AI citation effectsAlways "Not established", with the evidence behind thatA score, a probability or a readiness grade. None is offered, because none is supported

Confirmed Google requires structured data to describe content that visitors can see, and says no special markup is needed for its generative AI Search features

4. How is this different from a schema generator or the schema markup validator?

A generator writes markup and a validator checks that it is well formed. This audit starts from markup you already have and asks whether it is accurate, connected and worth keeping.

Compared onSchema generatorSchema markup validatorSchema evidence audit
Starts fromA formMarkup or a pageMarkup or a page
Main questionWhat should I write?Is it valid, and what can Google do with it?Is it accurate, connected and supported?
Syntax and vocabularyBuilds valid outputFull schema.org check, three formatsReuses the validator engine
Entity relationshipsLinks what it buildsEntity map and cross-page checksIdentity clashes, orphan entities, sameAs, author and publisher links
ClaimsNoNoEvidence table of claims to compare with the page, plus flags
Removal adviceNoNames retired featuresRemoval candidates with reasons
OutputJSON-LDFindings and CSVFive lanes, checklist, Markdown, CSV and JSON

Directional The audit shares its parser and Google rules with the validator, so the two always agree on syntax and feature requirements

5. When should you remove schema markup?

Remove it when it is wrong, misleading or serves nothing. Leave it when it is accurate and harmless. The audit lists candidates and reasons; the decision stays with you and the page owner.

  • Retired Google features. FAQ rich results ended on 7 May 2026, How-to in September 2023 and the sitelinks search box on 21 November 2024. The markup stays valid schema.org and does no harm, so remove it when you next edit the page if nothing else reads it.
  • Reviews or ratings a business writes about itself. Google treats them as ineligible for star results, and fake reviews break its guidelines.
  • Placeholder or expired values. "Your Name", example addresses, past priceValidUntil dates.
  • Duplicates. Identical nodes written by both a theme and a plugin, or two WebSite nodes.
  • Entities nothing connects to. A node that no author, publisher or page points at.

Confirmed Google's review snippet guidelines and its documentation updates log; the dates are the ones the validator uses

Part 3

Why there is no AI score

The evidence behind lane 5, and what it does and does not show.

6. Why is the AI citation lane always "not established"?

Because no published test has shown that adding schema reliably changes which sources an AI answer cites, and Google says its AI features need no special markup. A score would imply precision nobody has.

SourceWhat it foundLimits
Google: optimizing for generative AI featuresStructured data is not required for generative AI search, and no new machine-readable files are neededDescribes what Google requires, not what influences selection
Ahrefs, May 20261,885 pages that added JSON-LD, tracked 30 days before and after against control pages: AI Overviews -4.6% (statistically significant), AI Mode +2.4% and ChatGPT +2.2% (neither distinguishable from zero)Pages were already cited; both groups' citations fell; implementation dates were estimated from crawls
OtterlyAI, December 2025 to March 2026On one company's site, 319 prompts across seven AI search experiences: only Gemini reproduced the JSON-LD when asked directlyOne site and one prompt set; an assistant failing to repeat markup does not prove its retrieval ignores it

Neither study shows that schema hurts, and neither shows that it helps. Schema still describes entities, supports documented Google features and makes authorship explicit. Measure AI visibility separately, with first-party reports and a fixed prompt set, and treat any change after a markup edit as a hypothesis.

Directional Two studies, each with limits; absence of proof is not proof of no effect

Part 4

Questions answered

Short answers to the questions that come up most.

7. What do people ask about auditing schema markup?

Does a clean audit mean my schema helps AI citations?

No. A clean audit means the markup parses, links up and, where it matches a Google feature, has the required properties. It says nothing about AI citations, which the audit always reports as not established.

Confirmed Google's AI features guidance and the studies in part 3

Is this a replacement for Google's Rich Results Test?

No. It copies Google's published requirements, but only Google's tool shows how Google reads a live page. Use this audit first, then confirm any feature you want in the Rich Results Test and Search Console.

Confirmed Google documents the Rich Results Test and URL Inspection as its own checks

Why does it list claims instead of saying whether they are true?

Because it cannot see your page. Google requires markup to describe visible content, so each claim becomes a line you compare with the page, with what to look for.

Confirmed Google's general structured data guidelines

Why is there no score?

A single number would mix evidence of different strengths. Keeping five lanes separate shows which conclusions rest on a test and which rest on nothing yet.

Directional A design choice, not a Google rule

Can it audit a URL by address?

Not directly. Browsers stop one site's page reading another's HTML, and a server that fetches addresses needs safeguards against abuse. Paste the source, or use the bookmark.

Confirmed Standard browser cross-origin rules

Should I delete FAQPage and HowTo markup?

You do not have to. Both remain valid schema.org, and Google says leftover markup of retired features causes no problem. Remove it when you next edit the page if nothing else uses it.

Confirmed Google's documentation updates and its 2024 sitelinks search box post

Is what I paste sent anywhere?

No. The audit runs in your browser and does not send your markup to a server. The bookmark passes a page to this tab inside your browser. The only thing a tool can send is the business email you choose to give when you unlock more runs.

Directional Stated by the builder; you can confirm it in your browser's network panel

How many times can you use the free tools?

You get 3 free runs across all the free tools. After that, joining the Search Signals newsletter with a business email unlocks 10 more.

Loading an example, copying and downloading do not use a run. The count is kept in your browser, so clearing site data resets it.

Directional A fair-use limit that may change; the count is a convenience, not a security control

Part 5

Limits and sources

What this audit cannot do and where the facts come from.

8. What can this audit not do?

It cannot see your visible page, fetch an address, or know what Google will show.

Truth of claimsIt lists them and flags what it can spot from the markup alone, such as placeholders and impossible dates
Google's viewThe rules are a dated copy of Google's documentation; only Google's own tools show what Google sees
FetchingIt reads what you paste or what the bookmark sends
Vocabulary checksNames and value types against schema.org 30.1; not every constraint a strict validator applies
IdentityIt cannot tell whether two entities with one name are really one
AI outcomesNot measured, not estimated

Directional Prototype tool; heuristics such as placeholder and promotional wording are prompts to check, not findings of fact

9. Where do the facts on this page come from?

From Google Search Central, schema.org and two published studies, read on 28 October 2026. The Google feature rules were last copied on 23 October 2026.