FAQ

How much does structured data (JSON-LD) matter for AI search?

Structured data is the 'fact sheet' machines read: it tells crawlers unambiguously which company an entity is, what a page is (an FAQ, a service, a report), and what the facts inside are. Its importance is asymmetrical — it does not buy you citations, but its absence causes a class of real failures: AI reads your facts wrong, merges you with a namesake company, or simply cannot parse your pages. In Negency's sample of 30 Chinese companies going global, 17 lacked JSON-LD entirely — the most common major gap in the whole machine-readable group.

Figure 01 · JSON-LD skeleton: Organization + FAQPage

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "name": "Negency",
      "legalName": "上海熵宜信息技术有限公司",
      "alternateName": "熵宜",
      "url": "https://www.negency.com/",
      "description": "Negency is an AI-native enterprise systems engineering and intelligent solutions company.",
      "foundingDate": "2014",
      "contactPoint": {
        "@type": "ContactPoint",
        "contactType": "sales",
        "email": "[email protected]"
      }
    },
    {
      "@type": "FAQPage",
      "mainEntity": [
        {
          "@type": "Question",
          "name": "什么是 GEO?",
          "acceptedAnswer": {
            "@type": "Answer",
            "text": "GEO 是让品牌在 AI 生成的答案中被提及、被引用为来源的工作。"
          }
        },
        {
          "@type": "Question",
          "name": "What is GEO?",
          "acceptedAnswer": {
            "@type": "Answer",
            "text": "GEO is the work of getting a brand cited in AI answers."
          }
        }
      ]
    }
  ]
}
</script>
  1. Organization pins down who you areLegal name, website, founding year and contact — so engines stop confusing you with namesakes
  2. Only mark up facts visible on the pageMarkup that contradicts the visible page is worse than no markup
  3. FAQPage Q&A must match the visible FAQNever invent questions the page does not show
  4. Every field comes from schema.orgDo not invent property names
Example facts are taken from Negency's published llms.txt and official website. Note: Google stopped showing FAQ rich results in May 2026; FAQPage here is valued as machine-readable structure, not as a Google Search display.

What it does in practice

Entity grounding: Organisation markup is how a machine knows, without guessing, that this page speaks about a specific company, what it does and where its official channels are. LLMs' entity knowledge has to attach to something — structured facts are the anchor; without markup, an engine can only guess which namesake a name refers to.

Understandable content units: Service, FAQPage and Article markup declare in machine terms what a page is and which section answers which question, instead of forcing the engine to infer it from prose.

Verifiable facts: dates, authors, sources and issuing organisations in markup are extractable at crawl time — verifiable facts are what answers prefer to cite.

On format, Google's official documentation lists JSON-LD as the recommended format: it sits in its own script tag, separate from the visible text, and is the easiest to implement and maintain. Google also requires markup to match the visible page content — marking up claims nobody can see is worse than no markup.

What it cannot do

Structured data is not a citation switch. Without third-party mentions it will not lift mention rates; it raises two different things: the probability that facts get read correctly, and the probability of being quoted correctly once you are in scope. Google's official docs are explicit that no AI-specific structured data is needed for AI Overviews and AI Mode — structured data is not a requirement for AI search. What it does is make the citations that do happen more accurate.

One change from 2026 is worth knowing: Google stopped showing FAQ rich results in Search in May 2026. FAQPage remains a valid schema.org type with real value for AI engines and other parsers, but do not build it as a Google Search display play; the reason to add FAQPage is machine readability, not a search feature.

The cost of the work is low and the effect is testable — in our sample of 30 Chinese exporter websites (Negency Flagship Report 2026, Issue 1, sampled 30 October 2026), missing JSON-LD was the single most common major gap (17 of 30), and the machine-readable dimension averaged only 69.2 (median 68.5). That profile — cheap to fix, most people have not, fixes a real error class — is exactly what should be done first. (Sampling note: the sample is the global homepages of 30 Chinese exporter websites, measured in a single-day sweep on 30 October 2026 with the Negency audit engine — about 30 signals in five groups. The report measures technical visibility only — whether AI can reach and read the site — not brand mention rates in AI answers.)

Order of work

Organisation entity first, then FAQPage for FAQ content, then Service for offerings, then Article for news and reports. Cross-check names, dates and URLs against the visible page text: markup that contradicts the page makes things worse, not better. Accept the work in two steps: validate the syntax with a checking tool such as Google's Rich Results Test, then hand-check that the names, dates and links in the markup match what a human can see on the page, item by item.

The four types to do first, in order

  • Organization: pin down who you are — legal name, website, logo, public contact and founding facts — so engines stop confusing you with namesakes;
  • WebSite: declare the site's name and how its language versions relate, so multilingual pages roll up to one entity;
  • FAQPage / Article: structure the title, author, publication and update dates of Q&A and article pages, so engines can extract and attribute them cleanly;
  • Service / Product: state each service and product's name, description and audience as structured entries, so "who offers what" can be read directly.

The figure above this article shows a JSON-LD skeleton for Organization + FAQPage, using Negency's own public facts as the example; every field comes from the schema.org specification, so you can write against it directly.

When structured data "does nothing", the usual cause

When structured data "does nothing", the cause is usually not format but mismatch: the structured facts disagree with what humans can see on the page, only the homepage has it while inner pages do not, or a plugin's default output conflicts with the company's real facts. A typical case: the markup says the company was founded in 2014 while the About page says 2016 — faced with two versions, the engine trusts neither. Engines trust cross-consistent information; structured data only states consistent information more clearly — it is not a pass for inconsistent information. Reconcile before you mark up, and re-check page by page afterwards; only then is the markup actually in place.

Not sure which markup your site is missing? A free AI visibility audit checks JSON-LD item by item — look at the result before you start fixing.

Related questions

How is GEO different from SEO?

SEO optimises a page's rank in a list of links, aiming for clicks. GEO (Generative Engine Optimization) optimises how discoverable, understandable, trustworthy and citable a company is inside AI-generated answers — the goal is to be in the answer itself. They share technical foundations (crawlability, structured data, content quality), but GEO additionally requires governed entity facts, a machine-readable surface (llms.txt, Markdown mirrors) and a third-party mention ecosystem. A site that ranks well is not automatically visible to AI: engines cite the companies whose facts they can fetch, parse and verify. The concrete differences sit in four dimensions — goal, object, content structure and evaluation — compared item by item in the body below.

How can Chinese companies going global build AI search visibility?

For Chinese companies going global, the visibility gap is rarely content volume — it is the machine-readable surface. Our audit of 30 Chinese outbound companies found machine-readability the weakest category (69.2/100 average): 17 had no JSON-LD on the homepage, 15 had a missing or thin sitemap, 13 had a missing or duplicated H1. The path has four steps: audit bot reachability and machine-readability first; govern entity facts so Chinese and English statements agree; publish question-shaped content built on real overseas buyer questions; then earn third-party mentions (media, communities, partners) and verify with periodic resampling against a fixed question set.

What should a corporate website change for the AI search era?

Five changes. (1) Bot reachability: robots.txt explicitly allows AI crawlers and live fetches are not stopped by firewalls or challenge pages. (2) A machine-readable surface: an llms.txt fact file (optional aid), a complete sitemap, Markdown versions of core pages (optional). (3) Entity structure: JSON-LD naming entity types such as Organization, clean title/H1 discipline, hreflang pairing on bilingual sites. (4) Question-shaped content: pages organised around buyer questions, short answers first, FAQ schema attached. (5) Measurability: analytics that separate human and bot traffic and attribute AI-referred inquiries. Negency's own site was rebuilt to exactly this standard, and the public audit tool scores against it.

What is llms.txt, and should every company website have one?

llms.txt is a plain-text brief at the site root written for AI systems: who the company is, what it does, where the core pages live, and the canonical facts and definitions. It does not by itself raise citation rates — in Ahrefs' May 2026 study of 137,210 domains, 97% of published llms.txt files received zero requests that month, so its direct effect is near zero — but it is the cheapest fact-governance move available: one file anchors a single version of the truth that crawlers can rely on. Verdict: worth adding, useless alone; it works as part of entity-fact governance, structured data and question-shaped content. Negency's llms.txt is a full fact bible: company definition, offering boundaries, sourced cases and citation rules.

What should an enterprise prepare before deploying AI agents?

Four preparations, and we advise against starting without any of them. (1) Scenario selection: pilot on a workflow that is frequent, comparatively rule-clear and verifiable — not the most glamorous one. (2) Knowledge and data foundations: govern the knowledge base first (deduplication, versioning, permissions); a dirty knowledge base only scales errors. (3) Governance design: permission boundaries, human approval points and audit logs decided before launch, not after. (4) A measurement baseline: record the manual process's time and error rates before the pilot, or value can never be proven afterwards. Negency's method is a small controlled pilot, verified before scaling, with governance as a design input rather than a patch.