FAQ

What does an AI visibility audit actually check?

An audit checks roughly 30 signals along one chain — can AI find you, read you and trust you — in five groups: crawler reachability (are bots allowed in, is the sitemap complete), machine-readable surface (llms.txt, Markdown versions, fetch success), entity facts (is company information structured and consistent everywhere), content structure (question-shaped organisation, heading hierarchy, short answers first) and citation readiness (do cases and numbers have public, checkable sources). Negency's free audit scores each group and points at specific pages and problems, not just a single total.

Figure 01 · The audit's five check groups

  • Group 1 · Crawler reachability
  • ✓robots.txt allows the major AI crawlers
  • ✓Sitemap exists with no dead links
  • ✓Pages return healthy status codes
  • Group 2 · Machine-readable surface
  • ✓llms.txt fact file present
  • ✓Markdown versions of core pages
  • ✓Body text extractable from raw HTML
  • Group 3 · Entity facts
  • ✓JSON-LD markup (name, business, audience, contact)
  • ✓Wording consistent between website and third-party platforms
  • Group 4 · Content structure
  • ✓Organised around real customer questions
  • ✓Each page leads with a quotable short answer
  • ✓Clean heading hierarchy
  • Group 5 · Citation readiness
  • ✓Cases and numbers carry public sources
  • ✓Sources attributed
Check items come from the article's five signal groups; the full audit engine runs about 30 signals in five groups (see the flagship report's sampling note). The audit measures technical visibility only, not brand mention rates in AI answers.

The five groups, itemised

The audit checks five groups of signals, every item traced to a specific page and problem — never just a total. Group by group: what is checked, how it is judged, what failure looks like.

Crawler reachability

What is checked: does robots.txt allow the major AI crawlers, does a complete sitemap exist without dead links, do pages return healthy codes. How it is judged: AI crawlers are simulated against the homepage and core pages, not just configuration. Typical failure: robots allows the crawler while the CDN or WAF blocks its IP ranges; or the 200 is really a challenge page.

Machine-readable surface

What is checked: is there an llms.txt fact file, do core pages have Markdown versions, is body text extractable from raw HTML. How it is judged: raw HTML is checked for the body text itself; a missing llms.txt is a warning, not a failure — an optional aid, not a requirement. Typical failure: body text script-rendered, source holding only container tags; or llms.txt contradicting the website's facts.

Entity facts

What is checked: are name, business, audience and contact details marked up in JSON-LD, and consistent between the website and third-party platforms. How it is judged: markup is reconciled against the visible page text and cross-checked against major directories. Typical failure: no JSON-LD at all; or a founding year in the markup differing from the About page.

Content structure

What is checked: is content organised around real customer questions, does each page lead with a quotable short answer, is the heading hierarchy clean. How it is judged: core pages are sampled: a definition within the first three sentences, each H1 unique and on-topic, paragraphs leading with the answer. Typical failure: one H1 shared across the site, or an H1 stating a slogan instead of a topic; a concept page opening with the motto and no definition in three sentences.

Citation readiness

What is checked: do cases and numbers carry public, attributed sources. How it is judged: key figures and case claims are checked for a clickable public source and a date. Typical failure: "industry-leading" and "significant growth" with no figures or sources; or figures attributed only to unverifiable "internal data".

A real reference point

We sampled 30 Chinese companies' global websites with the same method: mean total 80.1/100, median 80 (range 55–98), 7 above 90 and 6 below 70 — but the machine-readable surface was weakest (average 69.2, median 68.5): 17 of 30 lacked JSON-LD, 15 had sitemap problems, 13 had malformed H1 headings, 10 scored below 70 on both machine readability and entity structure. (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.)

Scope matters: the audit measures technical visibility — whether AI can reach and read a website — not brand mention rates in AI answers. A high technical score only proves you hold the entry ticket.

The audit runs on this site: no registration for the score; the full report arrives by email.

How an audit runs, step by step

  • Intake: submit the domain and confirm scope — languages, target markets, the engines that matter;
  • Automated crawl: AI crawlers are simulated against the homepage and core pages, recording allow/deny status, response codes and body-text extraction;
  • Five-dimension scoring: every deduction maps to a specific page and problem, never just a total;
  • Readout: start with the weakest dimension and entry-ticket issues before content and citation issues — in the wrong order, budget goes to the wrong fixes;
  • Fix and re-test: remediate by priority, then re-run the same instrument. Score movement and mentions are recorded separately: the score is a process metric, mentions are the outcome metric.

After the audit: the improvement path, in priority order

  • Priority 1 — the entry ticket: allow AI crawlers, repair the sitemap, ensure core pages return clean responses with extractable body text — without entry, everything else is zero;
  • Priority 2 — being understood: add JSON-LD entity markup, fix H1 and heading hierarchy, give core pages a quotable short answer;
  • Priority 3 — being trusted: align entity wording across website and third-party platforms, attach public sources to cases and numbers;
  • Ongoing — being cited: publish question-shaped content, build third-party mentions, re-sample a fixed question set regularly.

For a formal, evidence-backed diagnostic after the self-check, the diagnostic sample excerpt shows the deliverable format and granularity.

See where your brand stands in AI search

Start with the free AI visibility audit: your scores for crawler reachability, machine readability and entity structure in minutes. For the full competitor comparison and fix list, see the AI search visibility diagnostic package.

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.