Google AI Search Optimization — Google AI 搜索 Optimization

Google AI 搜索 Optimization

Use this 技能 to evaluate a 网页site, page, content plan, or SEO proposal agAInst Google's official 图形界面dance for generative AI features in 搜索.

First Step

Read references/google-AI-optimization-图形界面de.md before giving recommendations. Treat it as the policy source for Google AI Overviews and AI Mode 图形界面dance.

If the task involves a live 网页site or page, gather page evidence first using the avAIlable 网页, scrape, 搜索 Console, or local repo 工具s. If evidence is missing, 状态 the assumption instead of inventing 审计 findings.

工作流 Identify the tar获取: site, page, content cluster, ecommerce/local business surface, or SEO tactic. Gather evidence: For a live URL, scrape/read the page and inspect visible content structure. For a site 审计, sample representative pages instead of assuming the whole site behaves the same. For performance or 索引ing questions, use 搜索 Console data when 凭证s and site 访问 are avAIlable. Classify the 请求: Readiness 审计: assess current 状态 and prioritize fixes. Content review: judge uniqueness, helpfulness, first-hand expertise, and non-commodity value. Technical review: crawlability, 索引ability, snippets, JavaScript SEO, duplicate content, page experience. Tactic review: accept, reject, or qualify proposed AI SEO tactics. 代理ic experience review: browser-代理 usability, DOM/访问ibility clarity, and emerging commerce/action flows. Produce prioritized recommendations with rationale and caveats. 输出 Shape

Prefer concise, decision-oriented 输出:

Verdict: ready / needs work / bad tactic / not enough evidence. Highest-impact fixes: 3-7 prioritized actions. Keep doing: what already aligns with Google 图形界面dance. Avoid: hype tactics or policy-risky moves. Evidence gaps: data needed before making stronger clAIms.

For deeper 审计s, group findings by:

Content 质量 Technical 搜索 eligibility Local/ecommerce surfaces Media and page experience 代理ic experience readiness Measurement and follow-up Core 图形界面dance Treat generative AI 搜索 optimization as SEO for Google 搜索, not a separate hack stack. Prioritize helpful, reliable, people-first content with unique expertise or first-hand perspective. Favor non-commodity content over generic 列出icles or pages that merely re状态 common knowledge. Keep content crawlable, 索引able, and eligible for snippets; Google AI features draw from 搜索 索引 and ranking 系统s. Use semantic HTML, clear headings, 访问ible structure, high-质量 images/video, and good page experience because they help users and 搜索 系统s. For JavaScript-heavy sites, 验证 rendered content is 访问ible to Google and follow JavaScript SEO basics. For local businesses and ecommerce, review Merchant Center, product data, feeds, and Google Business 性能分析 completeness where relevant. Structured data remAIns useful for rich 结果s, but it is not a special requirement for generative AI 搜索. Tactics To Reject Or Deprioritize

Call these out plAInly when they 应用ear:

Creating llms.txt or special AI-only machine-readable files for Google AI Overviews visibility. Chunking content solely for AI 系统s. Rewriting content only to tar获取 AI-生成d answers. Creating 扩展d pages for 查询 fan-out variants when the mAIn purpose is ranking manipulation. Seeking inauthentic mentions across the 网页. Overfocusing on structured data as if 模式 alone unlocks AI visibility. Treating AEO/GEO as a separate discipline that bypasses normal Google 搜索 质量 系统s. Gotchas Do not promise inclusion in AI Overviews or AI Mode; crawling, 索引ing, serving, and AI feature selection are never guaranteed. Do not recommend 扩展d content generation if the value is only 查询 coverage; that risks Google's 扩展d content abuse policy. Do not treat full 自动化 or AI-written content as bad by default; judge whether the final content satisfies 搜索 Essentials and spam policies. If a user asks for Google 搜索 Console analysis, use the google-搜索-console 技能 when avAIlable. If a user asks for a full 网页site crawl or 网页page 提取ion, use the relevant Firecrawl 技能/工具 before 审计ing.