• AI-driven search traffic is growing rapidly and represents a significant share of organic discovery
• LLMs cite only a handful of domains per response vs. 10 blue links in traditional search
• Traditional SEO rank is a weak predictor of AI citation — many cited pages rank outside the top 20 in Google
• Content freshness, structure, and extractability matter more than backlinks for LLM visibility
• Google AI Overviews appear on a large and growing share of searches, and most AI-assisted searches result in fewer outbound clicks

1. Extractable, not just readable: Content must be structured in self-contained chunks that LLMs can pull verbatim
2. Answer-first: Lead with the direct answer, then provide supporting context
3. Entity-rich: Reference entities and relationships, not just keywords
4. Multi-engine: Optimize for Bing (ChatGPT), Google (AI Overviews), and direct AI crawlers (Perplexity) simultaneously
5. Machine-readable: Schema, llms.txt, and clean HTML structure help LLMs understand page semantics

[One-sentence description of what this site covers]

• Category Hub A: Description of this category
• Category Hub B: Description of this category

• [Page Type]: [What these pages contain and why they're useful]

• [Where the data comes from, how often updated]

• /llms-full.txt: Complete content index

Also create `/llms-full.txt` — a comprehensive version with more detail that LLMs can fetch when they need deeper context.

For pSEO specifically, the llms.txt should:
- List all category hub pages
- Describe what each page type contains
- Note the data source and update frequency (freshness signal)
- Link to the sitemap for full page discovery

<section>
  <h2>[Question or topic as heading]</h2>
  <!-- Answer capsule: 134-167 words, self-contained, directly answers the heading -->
  <p>[Direct answer in the first 1-2 sentences. Then supporting detail.
     The entire paragraph should make sense if extracted without any
     surrounding context.]</p>
</section>

• Each section under an H2/H3 should be a complete, self-contained answer (134-167 words optimal)
• Lead with the conclusion/answer, then provide reasoning
• Never assume the reader has seen other sections — each chunk must stand alone
• Use clear heading-to-content mapping (the heading is the question, the content is the answer)
• Pages with 120-180 words between headings get 70% more ChatGPT citations

• Long unstructured paragraphs with no headings
• Content that requires reading previous sections to understand
• Vague headings that don't indicate what the section answers
• Burying the answer after paragraphs of context

• Surface numeric data from the content model as explicit statistics ("4.8 average rating from 500+ reviews")
• Include specific numbers, percentages, dates, and measurements
• Cite data sources explicitly ("According to [source], ...")
• If the business has proprietary data, surface it — LLMs prefer content with information not found elsewhere
• Add relevant expert quotations or attributions where possible

• Reference the primary entity by its full canonical name, not just abbreviations
• Include entity relationships ("made by [company]", "located in [city]", "similar to [related entity]")
• Use schema markup to explicitly define entities (Organization, Product, Place, Person)
• Link to authoritative entity sources (Wikipedia, official sites) to help LLMs disambiguate
• Use consistent entity naming across all pages (don't alternate between "NYC" and "New York City")

• Verify Bing indexation via Bing Webmaster Tools (not just Google Search Console)
• Submit sitemap to both Google Search Console and Bing Webmaster Tools
• Verify AI crawlers are not blocked in robots.txt
• Use SSR or SSG — AI crawlers cannot execute JavaScript (client-rendered pages are invisible to them)

• Neutral and factual — not promotional or salesy
• Comprehensive but concise — covers the topic fully without padding
• Comparative — comparison/list formats make up ~33% of all AI citations
• Authoritative — backed by sources, data, and expertise

• Use neutral, informational tone even on commercial pages
• Include comparison sections where relevant (vs. alternatives, compared to similar options)
• Add "at a glance" summary sections that LLMs can extract as complete answers
• Avoid superlatives without data backing ("best", "top", "#1" without evidence)

• Use ISR with short revalidation intervals to keep pages fresh
• Display a visible "Last updated: [date]" on every page
• Include

  dateModified

  in schema markup with accurate dates
• If data changes (prices, ratings, availability), reflect changes quickly
• Consider automated data refresh pipelines that trigger page regeneration

• /llms.txt

  exists at site root with page type descriptions and category links
• robots.txt allows GPTBot, ChatGPT-User, PerplexityBot, ClaudeBot, Bingbot
• Site is indexed by both Google and Bing (verify in both webmaster tools)
• All pages use SSR or SSG (no client-side-only rendering)
• Each content section is a self-contained 134-167 word answer capsule
• Headings clearly indicate what the section answers
• Statistics and specific numbers are present on every page
• Entity names are consistent and schema-defined across all pages
• Content tone is neutral and factual, not promotional
• Comparison/list sections are included where relevant
• "Last updated" date is visible on every page and in schema
• FAQ sections use Q&A format that matches natural language queries

• Builds on: pseo-templates (page structure), pseo-schema (JSON-LD), pseo-metadata (crawlability), pseo-linking (robots.txt, sitemap)
• Extends: pseo-performance (SSR/SSG requirement, freshness via ISR)
• Validated by: pseo-quality-guard (content quality is the foundation — thin pages won't be cited by LLMs either)