FAQ

Going global: an AI search visibility self-check for Chinese companies

When a company expanding overseas is invisible in AI search, the cause is rarely one thing: it is five stacked problems — cannot get in, cannot be read, cannot be recognised, does not match the question, and nobody mentions them. This 17-item self-check, benchmarked against Negency's measurement of 30 global-facing corporate websites, locates which layer you are stuck on and what to fix first.

Figure 01 · The going-global AI visibility self-check: five groups, 17 items

  • Group 1 · Cannot get in (crawling and reachability) — 4 items
  • ✓Homepage and key pages fetched as an AI crawler
  • ✓robots.txt does not block major AI crawlers
  • ✓Sitemap exists, is current and submitted
  • ✓Target markets can access the site, no geo-blocking
  • Pass standard: all four “yes” — otherwise the other groups do not matter yet. Sample benchmark: reachability mean 91.1, 21 sites at full marks.
  • Group 2 · Cannot be read (machine readability) — 4 items
  • ✓Exactly one on-topic H1 per page
  • ✓Organization JSON-LD present
  • ✓Core concept pages define within the first three sentences
  • ✓Each product and service has its own page
  • Pass standard: all four “yes”; pure engineering, clearable in two to three weeks. Sample benchmark: mean 69.2 (median 68.5), only 2 sites at full marks.
  • Group 3 · Cannot be recognised (entity consistency) — 3 items
  • ✓Company facts identical across site and directories
  • ✓Old names and domains cleaned up
  • ✓Footer and about page carry complete company facts
  • Pass standard: all three “yes”; schedule consistency work monthly. Sample benchmark: 17 of 30 with no JSON-LD, 13 with a missing or non-unique H1.
  • Group 4 · Does not match (content–question fit) — 3 items
  • ✓Most real customer questions answered head-on by a page
  • ✓Genuinely translated and localised pages per target language
  • ✓Cases and numbers with sources and dates
  • Pass standard: all three “yes”; schedule content work monthly. Sample benchmark: qualitative check, no separate score.
  • Group 5 · Nobody mentions us (citation ecosystem) — 3 items
  • ✓Accurate profiles in target-market directories
  • ✓Third-party public mentions in the past year
  • ✓Continuing original research or data
  • Pass standard: all three “yes”; run as a quarterly programme. Sample benchmark: qualitative check, no separate score.
Check in order: fail Groups 1–2 and fix the engineering first; Group 5 is the slowest — and sets the ceiling. Five groups, 17 items (4+4+3+3+3), with each group's pass standard. Benchmarks: Negency Flagship Report 2026 (Issue 1) (30 global-facing websites of Chinese exporting companies, sampled on a single day, 30 October 2026; median total 80/100, mean 80.1).

How to use this checklist

Work through all 17 items: answer "yes / no / not sure"; count "not sure" as "no". Five groups map to the five problems behind invisibility in AI search; every item states its check method and pass standard. Benchmarks: Negency's published measurement of 30 global-facing websites of Chinese exporters (Negency Flagship Report 2026, Issue 1, sampled on a single day, 30 October 2026; measured with the Negency AI visibility audit engine, around 30 signals across five categories; technical website visibility only, not brand mention rates in AI answers; median total 80/100, mean 80.1). Locate the blockage first, then sequence the fixes — in the wrong order, money and time go to the slowest layer.

Group 1: Cannot get in (reachability, 4 items)

  1. Key pages fetched as an AI crawler. Check: fetch as a crawler, not a browser; inspect status and body. Pass: key pages return normally, body not empty.
  2. robots.txt does not block major AI crawlers. Check: verify GPTBot, ClaudeBot, PerplexityBot unblocked; confirm CDN/firewall allow their IP ranges. Pass: target crawlers unblocked, able to fetch.
  3. sitemap.xml current and submitted. Check: spot-check links resolve and include latest pages; confirm submission to Google Search Console and Bing Webmaster Tools. Pass: few dead links; submitted in both consoles.
  4. Target markets load acceptably; no geo-blocking. Check: fetch key pages from target-market nodes. Pass: opens normally, no blocking.

The entry ticket: the sample averaged 91.1/100 on reachability, 21 at full marks. Fail here, and the other four groups do not matter yet.

Group 2: Cannot be read (readability, 4 items)

  1. Exactly one on-topic H1 per page. Check: inspect core page source; count and read H1s. Pass: one H1 per page, matching the topic.
  2. Organization structured data (JSON-LD) present. Check: inspect homepage source or test structured data. Pass: Organization JSON-LD exists; name, website, contact machine-readable.
  3. Core concept pages define within the first three sentences. Check: read the first three sentences. Pass: complete definition there, not a slogan.
  4. Each product and service has its own page. Check: confirm dedicated URLs, not one "solutions" page. Pass: hero products and services each have one.

The weakest layer, averaging 69.2/100 (median 68.5): 17 of 30 had no JSON-LD, 13 a missing or non-unique H1, only 2 full marks. Pure engineering — clearable in two to three weeks.

Group 3: Cannot be recognised (entity, 3 items)

  1. Company facts identical everywhere. Check: compare name (both languages), founding year, HQ and operating cities across website, LinkedIn, Crunchbase and Clutch. Pass: identical everywhere, no contradictions.
  2. Former names and domains cleaned up. Check: search old name and domain; confirm redirects or corrections. Pass: old name and domain no longer live facts.
  3. Footer and about page carry complete company facts. Check: confirm legal name, address, public contact present. Pass: all three present, not just a vision statement.

Group 4: Does not match (question fit, 3 items)

  1. Real customer questions answered head-on. Check: list questions customers ask AI (category, comparison, price, timeline); search the site for each. Pass: most answered directly by a page.
  2. Target languages genuinely translated and localised. Check: spot-check target-language pages for item-by-item correspondence and natural wording. Pass: genuine translation and localisation, not raw machine output.
  3. Cases and numbers sourced and dated. Check: spot-check case and data passages. Pass: key figures carry source and date, not adjectives alone.

Group 5: Nobody mentions us (ecosystem, 3 items)

  1. Accurate profiles in target-market directories. Check: search the company name in the main directories, association lists and review sites. Pass: profiles exist, facts accurate.
  2. Third-party public mentions in the past year (owned channels excluded). Check: search third-party mentions from the past year. Pass: at least one on record.
  3. Original research or data published regularly. Check: confirm continuing publication over the past year. Pass: continuing, not one piece.

The slowest group, and the ceiling-setter: recommendations in AI answers come mainly from third-party sources. Built over quarters; it cannot be bought or gamed. Our wedge guide Choosing an AI consulting and GEO partner follows this logic.

After the check: what to do next

Any "no" in Groups 1–2: fix the basic engineering first — the best-value two to three weeks. Any "no" in Groups 3–4: schedule content and consistency monthly. Any "no" in Group 5: run it as a quarterly programme with directory profiles and original research. For a scored machine check, run the free AI visibility audit and work through its report; scoring is documented in the flagship 2026 AI Search Visibility Report. For the four-step path behind this checklist, see the going-global path.

Industry versions: consumer electronics, medical devices, industrial equipment.

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.