llm-optimization
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseLLM Optimization Skill
LLM优化技能
Purpose
目标
Make websites appear in AI assistant recommendations and citations. Different from traditional SEO - optimized for how LLMs parse and recommend content.
让网站出现在AI助手的推荐和引用中。与传统SEO不同——针对LLM解析和推荐内容的方式进行优化。
Core Rules
核心规则
- Structured > Prose — LLMs extract facts from clear structure
- Schema.org is Critical — Speakable, FAQPage, HowTo schemas
- Answer the Question — First paragraph must directly answer intent
- Cite Sources — Links to authoritative sources build trust
- Entity Clarity — Clear business name, location, service definitions
- Freshness Signals — Last updated dates, recent content
- No Walls — Content must be crawlable, no JS-only rendering
- Never Override Truth — LLM optimization NEVER overrides factual accuracy or legal compliance
- 结构化内容＞散文式内容 —— LLM会从清晰的结构中提取事实
- Schema.org至关重要 —— 需包含Speakable、FAQPage、HowTo等Schema类型
- 直接回答问题 —— 第一段必须直接响应用户的搜索意图
- 引用权威来源 —— 链接到权威来源可建立信任
- 实体清晰明确 —— 清晰的企业名称、地点、服务定义
- 新鲜度信号 —— 最后更新日期、近期内容
- 无访问壁垒 —— 内容必须可被抓取，不能仅依赖JavaScript渲染
- 绝不违背事实 —— LLM优化绝不能违背事实准确性或法律法规要求
LLM Crawlers to Support
支持的LLM爬虫
| LLM | Crawlers | Notes |
|---|---|---|
| OpenAI/ChatGPT | GPTBot, OAI-SearchBot, ChatGPT-User | GPTBot = training, others = real-time |
| Google Gemini | Google-Extended | robots.txt control token, not a distinct UA |
| Perplexity | PerplexityBot, Perplexity-User | Bot = indexing, User = real-time fetch |
| Claude | ClaudeBot, Claude-User, Claude-SearchBot | Official Anthropic crawlers |
| Microsoft Copilot | Bingbot | Uses Bing's crawler |
| LLM | 爬虫 | 说明 |
|---|---|---|
| OpenAI/ChatGPT | GPTBot, OAI-SearchBot, ChatGPT-User | GPTBot = 训练用途，其他为实时调用 |
| Google Gemini | Google-Extended | 通过robots.txt控制令牌，并非独立UA |
| Perplexity | PerplexityBot, Perplexity-User | Bot = 索引用途，User = 实时抓取 |
| Claude | ClaudeBot, Claude-User, Claude-SearchBot | Anthropic官方爬虫 |
| Microsoft Copilot | Bingbot | 使用Bing的爬虫 |
robots.txt Configuration
robots.txt配置
txt
undefinedtxt
undefinedOpenAI crawlers
OpenAI crawlers
User-agent: GPTBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: GPTBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
Google AI (control token)
Google AI (control token)
User-agent: Google-Extended
Allow: /
User-agent: Google-Extended
Allow: /
Perplexity crawlers
Perplexity crawlers
User-agent: PerplexityBot
Allow: /
User-agent: Perplexity-User
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Perplexity-User
Allow: /
Anthropic/Claude crawlers
Anthropic/Claude crawlers
User-agent: ClaudeBot
Allow: /
User-agent: Claude-User
Allow: /
User-agent: Claude-SearchBot
Allow: /
undefinedUser-agent: ClaudeBot
Allow: /
User-agent: Claude-User
Allow: /
User-agent: Claude-SearchBot
Allow: /
undefinedContent Structure for LLM Extraction
适配LLM提取的内容结构
html
<!-- 1. Direct Answer (first 150 chars) -->
<p class="lead">
[Business Name] provides [service] in [location].
[Key differentiator]. [Call to action].
</p>
<!-- 2. Quick Facts Box -->
<aside class="quick-facts" itemscope itemtype="https://schema.org/LocalBusiness">
<h2>Quick Facts</h2>
<dl>
<dt>Service Area</dt><dd itemprop="areaServed">[Areas]</dd>
<dt>Price Range</dt><dd itemprop="priceRange">[Range]</dd>
</dl>
</aside>
<!-- 3. FAQ Section (critical for LLM) -->
<section itemscope itemtype="https://schema.org/FAQPage">
<!-- Each Q&A as schema -->
</section>html
<!-- 1. Direct Answer (first 150 chars) -->
<p class="lead">
[Business Name] provides [service] in [location].
[Key differentiator]. [Call to action].
</p>
<!-- 2. Quick Facts Box -->
<aside class="quick-facts" itemscope itemtype="https://schema.org/LocalBusiness">
<h2>Quick Facts</h2>
<dl>
<dt>Service Area</dt><dd itemprop="areaServed">[Areas]</dd>
<dt>Price Range</dt><dd itemprop="priceRange">[Range]</dd>
</dl>
</aside>
<!-- 3. FAQ Section (critical for LLM) -->
<section itemscope itemtype="https://schema.org/FAQPage">
<!-- Each Q&A as schema -->
</section>Forbidden
禁止事项
- ❌ Content behind JavaScript-only rendering
- ❌ Blocking LLM crawlers in robots.txt
- ❌ Missing Speakable schema
- ❌ Vague, marketing-speak first paragraphs
- ❌ No FAQ section on service pages
- ❌ Missing lastModified dates
- ❌ No structured data
- ❌ 内容仅依赖JavaScript渲染
- ❌ 在robots.txt中阻止LLM爬虫
- ❌ 缺失Speakable schema
- ❌ 第一段使用模糊的营销话术
- ❌ 服务页面无FAQ板块
- ❌ 缺失lastModified日期
- ❌ 无结构化数据
Definition of Done
完成标准
- robots.txt allows all LLM crawlers
- Speakable schema on all key pages
- FAQPage schema on service pages
- First paragraph directly answers search intent
- Quick Facts box with structured data
- lastModified meta tag present
- Content renders without JavaScript
- Entity names consistent across site
- robots.txt允许所有LLM爬虫访问
- 所有关键页面添加Speakable schema
- 服务页面添加FAQPage schema
- 第一段直接响应用户搜索意图
- 添加包含结构化数据的快速信息框
- 存在lastModified元标签
- 内容无需JavaScript即可渲染
- 全站实体名称保持一致
References
参考资料
- ChatGPT Optimization
- Gemini Optimization
- Perplexity Optimization
- Schema Speakable
- ChatGPT优化
- Gemini优化
- Perplexity优化
- Schema Speakable