CClarivyBilingual AI Search Visibility
MOCK SAMPLE — Quick diagnostic, not a complete audit. This is a sanitized public sample, not a paid customer report. The subject brand is the fictional Example Cross-Border Brand (URL example.com), and the 5 user queries + 35 datapoints are generated by the audit-runner mock orchestrator (RUN_MODE=mock). Numbers, citation map, and priority actions are illustrative; a paid Snapshot ($149) of your brand will produce a different report. See the sample home for the full disclaimer.
Clarivy · AI Search Visibility Snapshot · Public Mock Sample

Example Cross-Border Brand

Order ord_sample_snapshot_20260615 · Sampled 2026-06-15T10:00:00+08:00 · 5 user queries × 7 endpoints = 35 datapoints · $149

0 / 35

endpoint × prompt datapoints that mention example.com · 0 brand mentions across 35 datapoints

Deterministic GEO score pair

These mock scores are computed from observed LLM responses and structured citation fields. They are GEO-scoped report fields, not SEO scores, backlink scores, traffic estimates, CORE-EEAT scores, CITE scores, or ranking forecasts.

ScoreSample resultWhat drove it
AI Citation Readiness 0/100 · absent 0 / 35 endpoint-query datapoints mentioned example.com; 0 / 7 endpoints mentioned the brand.
Source Trust Baseline 15/100 · absent The subject domain example.com is known, but no mock response cited it as a source; evidence completeness is 35 / 35.

Per-endpoint mention count

EndpointMethodMentions / PromptsNotes
Doubao Volcengine Ark API · retention controls configured per account 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.
Kimi Moonshot API · retention controls configured per account 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.
DeepSeek DeepSeek API · no-retention header 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.
Qwen DashScope OpenAI-compatible API · no-log header 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.
ChatGPT OpenAI API · retention controls configured per account 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.
Claude Anthropic API · retention controls configured per account 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.
Gemini / Google AI Overviews Vertex AI generateContent · retention controls configured per account 0 / 5 All 5 prompts returned non-empty responses; none mentioned example.com.

Confidence: high (35 of 35 contract endpoints returned non-empty, non-error responses) — see methodology.html §11.2.

The 5 user queries (Snapshot product)

  1. best GEO audit service for 跨境品牌 in 2026 (zh + en blend, brand-agnostic)
  2. is example.com legit (brand-specific, purchase intent)
  3. example.com vs Profound (brand-specific, competitive)
  4. best AI search visibility audit for DTC brand (category comparison)
  5. how to improve example.com citations in AI answers (improvement intent)

These 5 queries were chosen to mimic a buyer's Snapshot briefing form. In a paid run, the buyer specifies their own 5 queries at the time of order. See methodology.html §5 (Snapshot vs Standard/Enterprise).

Top 3 priority actions (lift-ordered, placeholders in this sample)

  1. Publish a structured About page + llms.txt on your canonical domain. Why: all 35 datapoints returned 0 mentions, so the foundation (retrievable entity description + llms.txt) is the highest-leverage first step. Effort: 1 day. Owner: marketing. Expected lift: +0–2 mentions in 30 days on stochastic endpoints (deepseek) · +0–0 on deterministic endpoints (doubao, kimi, qwen). Evidence: every (endpoint, prompt) datapoint in sample-raw-data-index.md.
  2. Submit your brand URL to the crawler / source endpoints relevant to the model surfaces you care about (for example, Volcengine Juliang, Baidu Zhanzhang, Sogou, Bing). Why: some API endpoints are deterministic at temperature 0 but only after their source index has seen the URL. Effort: 1 day. Owner: marketing. Expected lift: +0–1 mention in 30 days on retrieval-dependent endpoints (within stochastic noise on Doubao / Kimi / Qwen). Evidence: the market-specific datapoints in sample-raw-data-index.md.
  3. Add structured data (schema.org Organization + sameAs to social profiles) to the canonical domain. Why: the LLMs that returned generic competitor mentions (no citation, no entity) can pick up an Organization entity if it is on the page in a machine-readable form. Effort: 1 day. Owner: dev. Expected lift: +0–1 mention in 60 days on stochastic endpoints (deepseek) — long tail, do not over-rotate from this single lever. Evidence: none of the 35 datapoints had a structured-data parse, which is the gap.

Disclaimer. The 3 actions above are template placeholders that illustrate the shape of a real audit's actions. In a paid run, the actions are derived from the per-(endpoint, prompt) findings of the actual responses, ranked by lift-per-effort, and each links to a real finding_id in sample-ai-employee-brief.md.

Top cited domains

No structured citations captured at this run — none of the 35 datapoints cited any domain for example.com. In a paid run, this table would list the top 3-5 cited domains across the actual LLM responses, with endpoint attribution.