Free AI visibility audit for global brands.

What an AI visibility audit actually checks

Generative search has two stages, not one. Citation selection decides whether a page enters the engine's source pool. Citation absorption decides whether the page actually shapes the answer language versus sitting unused. A page can be selected often but absorbed weakly. The user-visible mention is a third stage downstream of both.

A real audit measures all three layers, separately. Most "AI visibility checker" tools collapse them into a single score. We don't, because the fix for each is different — and a single score hides which layer is the actual blocker for your brand.

The audit also reports both onsite readiness (what the engines see when they fetch your domain) and offsite visibility (how engines surface you across third-party sources). For brand visibility in Mainland China, offsite signal is often the dominant factor — third-party citations are observed roughly 6.5× more often than self-citations in published GEO research.

The six layers we score

The audit grades your site against the citation pyramid. Each layer is a prerequisite for the next:

1. Search-engine indexed. Bing for ChatGPT search; Google for AI Overview / Gemini. If you're not in the index that powers the AI surface, none of the rest matters.
2. AI-crawler reachable. Granular robots.txt across five bot buckets — search/retrieval, user-triggered, training, opt-out tokens, undeclared. A common failure: blocking training-bots also accidentally blocks OAI-SearchBot via overbroad rules.
3. AI-parseable. Clean HTML, semantic URLs, llms.txt, Markdown alternates. Claude Code and Cursor send Accept: text/markdown natively — pages without a Markdown alternate return HTML noise.
4. Selection-worthy. Relevance, specificity, length, authority signals. Pages with real numbers, dated comparisons and named entities are cited 50%+ more than vague pages.
5. Absorption-worthy. Quotable phrasing, evidence density, structured comparisons. The length sweet spot is 1,000–3,000 words; pages under 500 words rarely absorb.
6. Third-party validated. Cited from Reddit, Zhihu, Baidu Baike, Hacker News, vertical media. This is the highest-leverage signal — and the slowest to build.

What makes this audit different — China specificity

Most free AI-visibility checkers test ChatGPT, Google AI Overview, Perplexity and Gemini. They do not test the engines a Mainland-Chinese consumer actually uses. Across a 540-call panel (30 prompts × 3 LLMs × 3 reps × 2 turns) we ran in May 2026, the three Chinese-trained engines we tested cited Mainland-CN sources at materially different rates than Western engines:

Top-15 source overlap (Jaccard) between the three engines was 0.20–0.30. They operate on substantially different source substrates. This means an audit built for ChatGPT or Gemini systematically under-reports what Mainland Chinese consumers see, because the source ecosystem the Chinese engines draw from barely overlaps with the Western set.

On a re-run of the identical 30-prompt panel one week later, source mention rates correlated at Pearson r 0.97–0.99 across all three LLMs (ICC 0.97–0.99). Top-5 source membership was perfectly stable (κ = 1.00). The pattern replicates, with a caveat: Doubao top-15 κ was 0.46 — long-tail source ranking is less stable for Doubao specifically. We treat top-5 with high confidence and the long tail with the appropriate hedge.

For more on the engine differences, see our research briefing Why your brand looks different on every Chinese AI.

What you get back — the report walkthrough

The free audit returns a multi-section report rendered live in your browser. The structure is the same as our paid client audits, just narrower in engine and source-graph coverage:

A — Executive snapshot

Aggregate visibility score (0–100) plus the single most-impactful blocker. Calibrated against the citation pyramid, not against a vendor-internal heuristic.

B — Onsite readiness

Per-bucket robots.txt audit (training / retrieval / user-triggered / opt-out / undeclared), llms.txt presence and parse-quality, llms-full.txt presence (if relevant), Markdown alternate detection on key pages, semantic URL structure, structured-data sanity (with the Bing/Copilot-only caveat), and word-count distribution flagging.

D — Per-LLM × category × niche surfacing

Whether each engine surfaces your brand at the broad-category (L1) level and at the positioning-niche (L2) level. Includes an absorption-depth column showing whether your brand explanation is reused by the engine or only mentioned in passing.

E — Audience hot topics

The actual mainland-CN consumer questions in your category — recommendation, how-to, comparison, authenticity, education prompts — drawn from the live prompt panel.

F — Top fixes

Three to seven prioritised actions with effort estimate (1-hour / multi-week / multi-quarter) and expected layer-of-pyramid impact. We label each fix as measured evidence, prior-knowledge hypothesis, or planned intervention test — never as a guarantee.

What this audit does not do

We are deliberately conservative about what the audit promises, because the AI-visibility tooling space is full of overclaims. Specifically, the audit does not:

• Guarantee a ranking position, in any number of days. Anyone promising "rank in DeepSeek in 7 days" is selling rather than measuring.
• Claim that JSON-LD schema "makes AI find you." JSON-LD is a Bing/Copilot signal in our sample, not a universal one.
• Recommend "add an FAQ section to boost AI score." Pure FAQ-format pages underperform — they add noise without new information, lowering signal-to-noise per page.
• Recommend "increase content length" as a generic fix. Length only helps if each chunk adds new information. Padding hurts.
• Measure sales, conversion, or attributable revenue. AI mention is a brand-visibility signal. Referral traffic from AI search is currently under 1% of total referral traffic globally, though intent quality is materially higher when it does occur.
• Inspect what is in any engine's training data. We can only observe what the engine self-attributes when answering.

Who this audit is built for

Three buyer profiles use the free audit most often. Each is consistent with the China-facing GEO thesis — global brands operating in or expanding to Mainland China — but with different content priorities:

Western luxury, fashion and beauty brands

Brands evaluating China expansion, particularly in categories where Zhihu, Xiaohongshu, SMZDM and Bilibili surface heavily. On a 1,620-response handbag panel we ran, DeepSeek surfaced Zhihu in 97% of responses and Xiaohongshu in 96% — but the source-mix collapses at the ultra-luxury price tier, where The Purse Forum, Vogue Business and auction-house archives carry more weight. Cross-vertical generalisation should be done carefully.

Travel, hospitality and destination marketing

Teams tracking inbound Mainland-CN demand and how DeepSeek, Qwen and Doubao recommend destinations, hotels and itineraries. Travel queries are heavily multi-turn — first turn is "where to go," follow-ups dig into accommodation and authenticity — which is why our audit panel is multi-turn rather than single-shot.

B2B technology and SaaS

Especially developer-facing products. DeepSeek's developer-corpus weight is more pronounced than Qwen or Doubao for technical questions — code generation, API behaviour, model deployment — so brand recall on developer-leaning categories surfaces differently. We separate consumer-facing vs developer-facing prompt pools in the panel.