Define the buyer set
Category, audience, pain, competitors, alternatives, integrations, and trust requirements.
Methodology
AI visibility is measured with prompts, positions, citations, and fixable gaps.
We do not treat AI outputs as static rankings. We test buyer journeys, compare competitors, review source trails, and convert patterns into website work.
Scope
Generate prompt groups
Shortlist, comparison, switching, integration, pricing, and proof prompts.
Record answer behavior
Mention rate, position, sentiment, omitted competitors, and cited sources.
Recommend website changes
Owned pages, third-party listings, structured data, FAQ coverage, and positioning fixes.
Article retrofit
Existing articles get a separate GEO self-check.
99-point article score
We review authoritative direct quotes, statistical data completeness, citability and traceable sources, structure, fluency, semantic density, trust signals, terminology, robustness, cross-domain fit, and readability.
Suggestion labels
Unsupported but useful improvements are marked as 建议补充来源, 建议补充数据口径, or 建议补充原文引语 so the rewrite stays faithful to the original article.