By DLVX · 12 min read · Updated 20 October 2026 · Part 3 of 10 in Get Found

Structured data and your entity, without the myths

  • Structured data earns rich results and removes doubt about who you are; it is not a lever for getting cited by AI.
  • Ship a small, honest set: Organization with a stable @id and sameAs, LocalBusiness if you have premises, Article and Product where they fit.
  • Your entity is mostly built off the page: one name, one address, one description, repeated the same way everywhere people and engines look.
On this page

Two jobs that get mixed up

When owners say "schema", they usually mean one of two different things, and the confusion costs money.

The first job is rich results: stars, prices, breadcrumbs and other extras Google may show next to a listing. That is what most structured data documentation is about, and it has real, checkable rules.

The second job is entity clarity: making sure engines know that the business on this page is the same one on Google Maps, LinkedIn, a directory and a news article. That is less about markup and more about consistency everywhere.

What structured data is not, in 2026, is a switch for AI answers. Google's AI features page says there is no special schema.org structured data you need to add for AI Overviews or AI Mode. Its AI optimization guide, last updated 10 July 2026, says structured data is not required for generative AI search, while still being useful for rich result eligibility. We plan with that sentence in mind.

An agency told us that adding FAQ and HowTo schema is the fastest way into AI Overviews. True?

No. Both of those rich results are gone, and markup was never the entry ticket to AI answers. We explain the history below so you can spot the pitch next time.

What the evidence says about schema and AI

We read the studies before we form an opinion, and this one is more settled than most.

+2.4%change in AI Mode citations after 1,885 pages added JSON-LD, not distinguishable from zero (Ahrefs, 11 May 2026)
3xhow much more likely AI cited pages are to carry JSON-LD. A correlation, not a cause
22 of 28top named Bangkok businesses with business schema on their site, against 57 of 82 others (our study, 3 October 2026)

The Ahrefs study from 11 May 2026 is the closest thing to a test we have. It tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 against about 4,000 matched pages that did not. AI Mode moved +2.4 percent and ChatGPT +2.2 percent, both not distinguishable from zero. AI Overviews moved minus 4.6 percent relative to the controls. The same study notes that AI cited pages are almost three times as likely to have JSON-LD. Read those two facts together: pages that get cited tend to be on well built sites that also use schema. Adding schema did not raise citations for pages that were already cited.

Our own study of who AI recommends in Bangkok, published 3 October 2026, points the same way. Among the businesses Google AI Mode named most often, 22 of 28 readable sites had business schema. Among the other businesses it named, 57 of 82 did. That is a small difference, and it is descriptive, not causal. The bigger gap was reviews, which we cover in the local visibility guide.

So our verdict is simple. Schema is worth doing because it is cheap, it supports rich results where they still exist, and it states your facts in a format machines parse cleanly. We never sell it as a way into AI answers.

What Google switched off, and when

A lot of schema advice online is stale. These are the dated changes we check every audit against.

DateChangeWhat it means for you
8 August 2023FAQ rich results limited to well known government and health sites; HowTo cut back, then dropped on desktop from 13 September 2023FAQ and HowTo markup stopped earning visible results for almost everyone
12 June 2025Google phases out Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement and Vehicle ListingNot a ranking change, per Google. Reporting for most of these left Search Console in September 2025
5 November 2025More lesser used features phased out; Search Console drops support for affected types from January 2026Check the changelog before adding a niche type
7 May 2026FAQ rich results stop appearing in Google SearchChanged in 2026. No visual benefit from FAQPage markup anywhere
15 June 2026FAQ rich result documentation removedChanged in 2026. The old doc URL now redirects to the changelog
24 July 2026Review snippet rules add: no fake or undisclosed incentivized reviews on the page or in markupChanged in 2026. Review markup now carries a clear honesty rule

Sources: Google, 8 August 2023, Google, 12 June 2025, Google, 5 November 2025, and the Search documentation changelog for the 2026 entries. Google's 2023 post also says there is no need to proactively remove markup that no longer shows anything. It does no harm. It just does nothing, so we stop adding it and we never flag its absence as a problem.

Keep the questions, drop the promise

A visible question and answer section is still good for readers, and it often matches how people phrase searches. Write it for them. Just do not pay anyone for "FAQ schema for AI".

The small set we ship on every site

Google's general structured data guidelines, last updated 10 July 2026, carry the rules that matter most: use JSON-LD, Microdata or RDFa, mark up only content that is visible on the page, never mark up misleading or hidden content, and remember that valid markup does not guarantee a rich result. Within those rules we ship a short list.

TypeWhere it goesWhy we use it
OrganizationHome page, onceGoogle uses it to understand administrative details and disambiguate the organization, per its Organization docs
WebSiteHome pageHelps Google pick your site name in results
LocalBusiness or a subtypeLocation pages, if you have real premisesAddress, hours and phone in a parseable form; see Google's LocalBusiness docs
ArticleGuides and newsAuthor, dates and headline stated once, the same as the visible page
Product with OfferProduct pages that sellPrice, currency and availability for merchant features
BreadcrumbListDeeper pagesClean path display and a clear site hierarchy

Should we put our Google review stars in our own schema?

Not for your own business. Google's review snippet rules, last updated 8 September 2026, say LocalBusiness and Organization review snippets are only for sites that capture reviews about other businesses, that self serving review rules apply, and that you must not aggregate ratings from other websites. If your rating comes from Google Maps, the markup can never earn you stars. Showing real reviews as text on the page is fine and useful for readers.

Your entity: one business, described one way

An entity is the thing engines think your business is: a name, a place, a set of services, and the pages and profiles that confirm it. Engines build it from everything they can read, not from your schema alone.

The markup side is short. Give your Organization a stable @id (we use the home page URL plus #organization) and reuse it whenever another block refers to your business. Add sameAs links to profiles you control. Schema.org defines sameAs as the URL of a reference page that unambiguously indicates the item's identity, such as a Wikipedia page, a Wikidata entry or an official site. Google's Organization docs give social and review profiles as examples.

{ "@context": "https://schema.org", "@type": "Organization", "@id": "https://example.com/#organization", "name": "Example Studio", "url": "https://example.com/", "logo": "https://example.com/logo.png", "description": "Website and AI integration studio in Bangkok.", "address": { "@type": "PostalAddress", "addressLocality": "Bangkok", "addressCountry": "TH" }, "sameAs": [ "https://www.linkedin.com/company/example-studio", "https://www.instagram.com/examplestudio" ] }

The off page side does more work. We list every place the business appears and make the facts identical: the same legal or trading name, the same address format, the same phone, the same one line description of what you do. A name that is "Example Studio" on the site, "Example Co., Ltd." on Maps and "ExampleStudio BKK" on Instagram is three weak signals instead of one strong one.

  1. Pick the canonical facts. Name, address or service area, phone, one sentence description, founding year, main services.
  2. Update every profile you control to match, starting with Google Business Profile, Bing Places, LinkedIn and the directories that rank for your category.
  3. Mirror them in Organization markup with the stable @id and sameAs.
  4. Use one site name consistently. Google's site names doc says it supports one per domain or subdomain and asks for a concise, commonly recognized name.
  5. Earn independent mentions. Third party pages that describe you correctly are what make the entity solid. That is the work in the content guide and the cornerstone guide.

Can we just create a Knowledge Panel or a Wikidata item for ourselves?

You cannot create a Knowledge Panel. Google's Knowledge Panel help says panels are generated automatically from public information, that Google does not create or delete them by hand, and that verified owners can suggest edits. As for Wikidata, our rule is to add an item only when independent, published references about the business exist. A self written entry with only your own site as a source does not make you more notable, and it tends to get removed.

How we audit the markup a site already has

Most sites we inherit already carry schema, usually from a theme or a plugin, often from three plugins that each think they are in charge. We do not start by adding more. We start by reading what is there, the way a crawler would.

  1. Pull the raw HTML, not the rendered page. Fetch the home page, one service or product page and one article as plain HTML and list every JSON-LD block. If the markup only appears after scripts run, we note it: Google renders JavaScript, but plain markup in the served HTML removes a layer of doubt for every other reader.
  2. Count the organizations. Two or three different Organization blocks with different names or logos is the most common entity problem we find. One survives, with one @id, and everything else points to it.
  3. Compare markup to the visible page. Prices, hours, author names and dates in the markup must match what a visitor sees. Google's general guidelines require it, and a mismatch is the fastest way to lose a rich result you had.
  4. Check dates. datePublished and dateModified should reflect real publishing and real edits. A dateModified that changes on every deploy is noise.
  5. Delete what can never pay. Self serving ratings and markup for features Google has switched off go. FAQ blocks that do no harm can stay until the template is touched anyway, as Google's 2023 post allows.
  6. Run the Rich Results Test and URL Inspection on the pages that matter, and keep a short note of what each template emits so the next redesign does not undo it.
What we findHow often it mattersFix
Several Organization blocks with different namesHigh: it blurs the entityOne block, one @id, one name
Ratings copied from Google MapsHigh: against Google's review rulesRemove from markup, keep real reviews as visible text
Markup that disagrees with the pageHigh: against general guidelinesGenerate markup from the same data the page shows
Missing sameAsMedium: a cheap clarity gainAdd profiles you control
No FAQPage or HowTo blockNoneLeave it. Never a finding

Worked example: cleaning up a confused entity

A typical audit, kept generic. A clinic group has three branches. The site footer shows the old address. Google Maps has one profile with the trading name and two profiles with the legal company name. The schema on the home page lists the founder as the organization name, and every service page carries an AggregateRating copied from Google Maps.

  1. Decide the facts. One trading name, the three real branch addresses, one main phone and one per branch.
  2. Fix the markup. One Organization block on the home page with a stable @id; one LocalBusiness subtype per branch page, each pointing back to the Organization through parentOrganization with that @id; remove the copied ratings, because they can never earn stars and break Google's review rules.
  3. Fix the profiles. Rename the Maps profiles to the trading name where that is how the business presents itself, align hours and categories, and update the footer.
  4. Validate. Rich Results Test and URL Inspection, as Google's guidelines suggest, then recheck a few weeks later.

Nothing in that list is clever. It just removes doubt, which is the whole job. Schema.org's own type pages are current to release 30.1 of 16 September 2026, so check the LocalBusiness and Organization definitions there when you pick a subtype.

Hype and waste

ClaimVerdictWhy
"Add FAQ schema to win AI Overviews"HypeFAQ rich results ended 7 May 2026, and Google says AI features need no special markup
"Schema gets you cited by ChatGPT"HypeNo measurable lift in the Ahrefs matched test of 11 May 2026
Organization with @id and sameAsWorksCheap, states your identity cleanly, Google documents it
Product and Offer on real product pagesWorksFeeds merchant features where you sell
Self serving review starsAgainst the rulesNot eligible for LocalBusiness or Organization; no aggregating ratings from other sites
Marking up content that is not on the pageAgainst the rulesGoogle's general guidelines require markup to match visible content
Twenty schema types "for completeness"Skip itMore to maintain, more to drift out of date, no extra benefit
Paying for a "GEO schema score"Skip itA composite score nobody has tied to outcomes

Mistakes and a checklist

The mistakes we find most: schema generated by a plugin that disagrees with the visible page; a different organization name in markup than on the site; ratings copied from Google Maps; dateModified bumped on every deploy; markup on pages that are noindex; and an agency invoice for FAQ schema in 2026.

  1. Canonical facts written down: name, address or area, phone, description.
  2. Every profile you control matches them, word for word where it can.
  3. Organization on the home page with a stable @id, logo, and sameAs to real profiles.
  4. WebSite markup with the same site name you use everywhere.
  5. LocalBusiness per real location, Article on guides, Product and Offer on product pages.
  6. Nothing marked up that a visitor cannot see, and no self serving review stars.
  7. Validated in the Rich Results Test, rechecked after every template change.
  8. No promises to anyone that schema will get them into AI answers.

Next in the track: Local and map visibility that lasts.

Sources and dates

Every external fact in this guide was read on 20 October 2026. Each source shows its publisher and the date it was published or last updated, and we flag what changed in 2026. Studies are their authors' measurements, usually correlations, not promises of results. Our own numbers come from our own Search Console and tools and say so where they appear.

Want this run on your own site? Start with the AI visibility diagnostic, or see the whole Get Found track.