AIGEO 不是玄学，也不是替代 SEO。它先检查企业事实能否被抓取、理解、匹配和核验，再用有范围的跨平台测试持续复测。五层回答“卡在哪里”，七步回答“接下来怎么做”。AIGEO is not magic or a replacement for SEO. Five readiness layers locate the bottleneck; seven execution steps show what to fix next and how to measure it again.
从技术地基到复测闭环，任何一层断掉，后面的内容和信源投入都可能被浪费。From technical access to repeatable measurement, a broken layer can waste everything built above it.
关键页面可访问、可索引、以网页文字呈现，robots、CDN/WAF 与 sitemap 不制造障碍。Accessible, indexable text pages with crawler, CDN/WAF and sitemap basics in place.
企业身份、中英文名称、产品树、目标市场和定制边界清楚且一致。Clear, consistent entity identity, product tree, markets and customization limits.
内容回答真实采购约束：MOQ、打样、交期、材质/剂型、工艺、合规和质量节点。Content answers real buying constraints: MOQ, samples, lead time, format, process, compliance and quality gates.
认证、测试、产能和案例有主体、范围、日期，并与目录、展会和外部页面一致。Certificates, tests, capacity and cases have an owner, scope and date, consistent with external sources.
固定企业、品类、市场、问题集、平台、轮次与日期，保留原始答案并跨周期比较。Fix scope, queries, platforms, rounds and dates; retain original answers and compare over time.
25 题免费自测用于准备度筛查；真实 AI 答案、竞品和信源需要专业基线测试。The free check screens readiness; real answers, rivals and sources require a scoped expert baseline.
做 3 分钟自测 →Take the 3-minute check →不是每家企业七步全部重做。先量基线，再按最低层和真实平台缺口选择动作。Not every company needs every step rebuilt. Measure first, then act on the lowest layer and real platform gaps.
按企业、品类、市场与买家角色建立问题集，在约定平台和轮次下记录出现率、域名、竞品、信源与答案语境。Scope company, category, market and buyer roles; record mentions, domains, rivals, sources and answer context across agreed platforms and rounds.
把 5 类买家 × 采购旅程的提问，加上 AI 自动衍生的子查询，列成完整清单。每个问题都是一个可被命中的内容入口。List every buyer question across the journey, plus the sub-queries AI generates. Each one is a content entry point.
把核心事实与采购答案做成可读取网页；结构化数据必须和可见文字一致，llms.txt 只作为辅助清单，不是收录开关。Publish readable answers; keep structured data consistent with visible text. llms.txt is an aid, not an indexing switch.
官网、B2B 目录、展会/协会、媒体和社交页面统一企业名、身份、品类、地址与官网，不把“关键词写满”当作策略。Align names, identity, category, address and domain across directories, events, media and social profiles.
根据实际信源榜选择目录、行业文章、协会或媒体；不自称榜单第一，不付费伪造第三方背书。Choose directories, industry articles, associations or media from actual source gaps; never fabricate third-party proof.
在适合的行业社区与内容渠道回答真实问题，不刷帖、不冒充客户，不把硬广包装成口碑。Answer real questions in relevant communities; no spam, impersonation or disguised advertising.
按约定周期复用同平台、问法、轮次与地区 / 语言，比较变化后再决定补网页、证据或信源。Repeat the agreed platforms, queries, runs and region / language; use the comparison to choose site, evidence or source fixes.
目标不是保证推荐，而是让每次变化都有原始记录、原因判断和下一步动作。The goal is not a guaranteed recommendation, but evidence for every change, diagnosis and next action.
下面是公开抓取/索引入口与样本中常见的差异，不是固定算法结论。实际项目仍以测试日期、模式、地区、问题集和原始答案为准。These are public crawler/indexing paths and recurring sample patterns, not fixed algorithm rules. Each project remains scoped by date, mode, region, queries and original answers.
如希望公开页面进入 ChatGPT 搜索与引用，应核对是否误拦 OAI-SearchBot；它与训练抓取控制不是同一件事。Check that OAI-SearchBot is not blocked when search visibility is desired; it is separate from training controls.
仍以搜索基础为地基：页面需可索引、可显示摘要，重要事实以文字呈现，结构化数据与页面可见内容一致。Search fundamentals remain the base: indexable pages, eligible snippets, text-based facts and matching structured data.
单独记录 Gemini 的回答、来源和地区/语言；不要把 Google AI Overviews 的结果直接等同为 Gemini。Record Gemini answers, sources and region/language separately; do not treat AI Overviews as the same result.
核对 Bing 索引与站长工具数据；Bing 的 AI Performance 可提供引用页面和触发查询，但引用不等于排名。Check Bing indexing and Webmaster data; AI Performance can show citations and queries, but citations are not rankings.
核对 PerplexityBot 抓取与答案实际引用来源；目录、榜单或媒体是否重要，应由该项目的信源榜决定。Check PerplexityBot access and actual answer sources; let the project's source list determine which external sites matter.
记录是否联网、所用模式与原始答案。平台入口会变化，不把某个搜索合作方写成永久、唯一来源。Record browsing mode and original answers. Platform integrations change, so no search partner is treated as permanent or exclusive.
正文提及、官网引用与综合出现必须分列，再与竞品、信源和答案语境一起解释。Answer-body mentions, owned-domain citations and combined presence are reported separately, then interpreted with rivals, sources and answer context.
只检查访客可见的回答正文是否出现规范企业名，不把推理、来源标题或摘要混入。Checks only visible answer text for the canonical entity name; reasoning and source metadata do not count.
检查引用链接是否指向正确官网或其子域，并分别记录旧域名与错误域名。Checks whether citations resolve to the correct owned domain or subdomain, with old and wrong domains recorded separately.
正文提及或官网引用至少发生一个。它是辅助指标，绝不冒充正文提及率。Answer mention OR owned-domain citation. It is a secondary metric and is never labelled mention rate.
按问题、平台和轮次统计反复出现的同行，不用一张截图定义竞争格局。Tracks recurring competitors by query, platform and round rather than defining competition from one screenshot.
保留原始链接，区分官网、目录、媒体、协会和社区，并判断来源是否真正支撑答案。Retains original URLs, classifies source types and checks whether a citation materially supports the answer.
区分路过提及、进入短名单与明确推荐，并核验实体、域名、能力和证据是否准确。Separates passing mentions, shortlists and recommendations while checking entity, domain, capability and evidence accuracy.
同一批企业问法，在不同平台可能相差几十个百分点。看真实日期与样本范围下的「平台分诊」。The same company and query set can differ by tens of points across platforms. See a dated, scoped platform diagnosis.
看「平台分诊」洞察 →Read the platform-split insight →自己先做 25 题免费自测；需要真实答案、竞品和信源时，再申请专业测试。Start with the free 25-question check; request an expert audit for real answers, rivals and sources.