SEO/AEO/GEO Skill
Advanced search, answer engine & generative engine playbook
Preview
# Master Skill Guide: Advanced SEO, AEO, and GEO Implementation for Modern Engines
An authoritative, technical, and actionable execution blueprint for optimizing websites across Traditional Search Engines (SEO), Answer Engines (AEO), and Generative AI Search Engines (GEO).
---
## 1. Core Definitions & Paradigm Shifts
### Search Engine Optimization (SEO)
The practice of optimizing web architecture, content relevance, and domain authority to maximize visibility in traditional index-based search engine result pages (SERPs) like Google and Bing.
* **Primary Vectors:** Crawlability, indexation speed, core web vitals, backlink profiles, internal link topologies, and E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness).
### Answer Engine Optimization (AEO)
The optimization of information structure to ensure immediate extraction by search snippet features, voice assistants, and multi-turn chat agents.
* **Primary Vectors:** Atomic information fragments, question-to-answer semantic mapping, structured data markup (JSON-LD), and targeted feature-snippet hunting.
### Generative Engine Optimization (GEO)
The engineering of digital assets to maximize inclusion, citation probability, and positive sentiment within the context windows and generative outputs of Large Language Models (LLMs).
* **Primary Vectors:** Entity-relationship anchoring, raw data parity (HTML tables vs. visual layouts), `llms.txt` architecture, citation context alignment, and high-density terminology injection.
```
┌───────────────────────────────────────────────────────────────────────────┐
│ THE MODERN SEARCH TRINITY │
├──────────────────────────┬─────────────────────────┬──────────────────────┤
│ SEO │ AEO │ GEO │
│ (Search Optimization) │ (Answer Engine Opt.) │ (Generative Engine) │
├──────────────────────────┼─────────────────────────┼──────────────────────┤
│ • Focus: Index Rankings │ • Focus: Quick Answers │ • Focus: Citations │
│ • Target: Google / Bing │ • Target: AI Overviews │ • Target: LLM Context│
│ • Core: Links & Vitals │ • Core: Schema & Q&A │ • Core: llms.txt │
└──────────────────────────┴─────────────────────────┴──────────────────────┘
```
---
## 2. Technical Infrastructure & Performance Architecture
### Client-Side vs. Server-Side Rendering (SSR)
Modern Single Page Applications (SPAs) built using standard React, Vue, or frameworks like client-rendered Lovable and Bolt often result in empty root DOM nodes (`<div id="root"></div>`) during initial crawl phases. While Googlebot executes JavaScript asynchronously, the "second wave" of indexing introduces latency and execution risks. Generative engine crawlers often fail to execute complex JS bundles entirely.
* **Execution Blueprint:**
* Deploy Server-Side Rendering (SSR) or Static Site Generation (SSG) utilizing frameworks such as Next.js, Remix, or Astro to ensure immediate HTML payload availability.
* For existing SPAs, implement hydration strategies or pre-rendering engines (e.g., Puppeteer/Prerender.io) to server-deliver semantic HTML to user-agents matching `Googlebot`, `Bingbot`, `ClaudeBot`, `GPTBot`, `PerplexityBot`, and `OAI-SearchBot`.
### Performance & Core Web Vitals Optimization
Page load speed directly impacts crawl budgets and user experience metrics, serving as an explicit ranking and citation factor.
* **Asset Compression & Delivery:**
* Convert legacy raster graphics (PNG, JPEG) to WebP or AVIF formats. Re-render UI icons and logotypes exclusively as optimized inline SVGs.
* *Example:* Reduce a 179KB uncompressed PNG asset to a <7KB compressed SVG path array.
* **Bundle Optimization:**
* Enforce Code Splitting and Dynamic Imports for non-critical modules below the fold.
* Eliminate render-blocking resources by deferring non-essential script execution (`<script defer>`).
* Target explicit metrics: **Largest Contentful Paint (LCP) ≤ 1.2s**, **First Input Delay (FID) ≤ 100ms** (or **Interaction to Next Paint (INP) ≤ 200ms**), and **Cumulative Layout Shift (CLS) ≤ 0.1**.
### Indexation Acceleration & Multi-Engine Syndication
* **Sitemap Optimization:** Build a strict, dynamic `sitemap.xml` containing only canonical HTTP 200 URLs. Segment sitemaps by type (e.g., `sitemap-pages.xml`, `sitemap-tools.xml`, `sitemap-blog.xml`).
* **Robots.txt Directive Architecture:**
```http
User-agent: *
Allow: /
Disallow: /api/
Disallow: /admin/
# Explicitly permit AI Crawlers to ensure comprehensive extraction
User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
Sitemap: https://yourdomain.com/sitemap.xml
```
* **IndexNow Automation:** Integrate an automated webhook or API handler inside your CMS / continuous deployment pipeline to broadcast content mutations (mutated, created, or deleted URIs) to the IndexNow endpoint (`https://api.indexnow.org`). This triggers immediate multi-engine recrawls (Bing, Yandex, Seznam) within minutes.
---
## 3. High-Leverage Content Engineering
### Content Pillar & Topical Cluster Framework
Isolate broad commercial topics into high-authority structural units to capture deep topical authority.
```
┌─────────────────────────────┐
│ PILLAR PAGE │
│ (Ultimate Freelance Pricing)│
└──────────────┬──────────────┘
│
┌───────────────────────┼───────────────────────┐
▼ ▼ ▼
┌───────────────┐ ┌───────────────┐ ┌───────────────┐
│ CLUSTER PAGE │ │ CLUSTER PAGE │ │ CLUSTER PAGE │
│ (Hourly Rate) │ │(Project Rates)│ │ (Scope Creep) │
└───────────────┘ └───────────────┘ └───────────────┘
(All cluster elements maintain bidirectional, in-body hyperlinks)
```
* **Pillar Page:** Build an exhaustive, comprehensive resource (2,500–4,000 words) addressing an overarching domain entity.
* **Cluster Nodes:** Produce 5–12 targeted supporting documents (800–1,500 words) deep-diving into specific long-tail variants.
* **Linking Rule:** Every cluster node must pass equity back to the pillar page using exact-match or semantic anchor text, and the pillar page must contextually reference every cluster node.
### The Inverted Pyramid & Atomic AEO Blocks
To feed Answer Engine features and LLM context extraction algorithms, redesign your content layout to serve answers inline immediately.
* **Atomic Quick-Answer Block:** Position a structured 40–60 word conceptual summary directly beneath the primary `<h1>` element. This block bypasses stylistic narrative and answers the core query immediately.
* **Section Design (The Question H2 Layout):** Phrase subheadings explicitly as transactional or informational questions.
* *Anti-Pattern:* `## Claude Code Skill Framework Locations`
* *Optimized Pattern:* `## Where Does Claude Code Store Skill Files?`
* **Structural Mechanics:** Follow an inverted pyramid sequence inside each subsection:
1. *Sentence 1-2:* Direct, definitive answer to the heading's question.
2. *Sentence 3-5:* Granular supporting metrics, technical steps, or operational definitions.
3. *Sentence 6+:* Extended contextual exceptions, references, and peripheral theory.
### Data Parity for LLM Comprehension
Large Language Models interpret text and structured markup far more efficiently than embedded media, complex SVG canvases, or JavaScript chart instances.
* **Implementation Strategy:** Accompany every infographic, chart, or data visualization with a native HTML `<table>` element containing identical raw data points. Apply standard semantic styling (`<thead>`, `<tbody>`, `<th>`, `<td>`). This provides clean, scrapeable matrices for generative engines trying to cite comparative metrics.
---
## 4. Advanced Structured Data Architecture (JSON-LD)
Deploy semantic entities using JSON-LD script blocks injected directly into the HTML document header. Ensure data points mirror visible on-page assets perfectly to prevent structural trust penalties.
### 1. Corporate Identity & Entity Anchor (`/about` or Homepage)
Establishes the organization, links it to natural persons, and anchors the domain to definitive knowledge graphs.
```json
{
"@context": "https://schema.org",
"@type": "Organization",
"@id": "https://yourdomain.com/#organization",
"name": "Agensi AI",
"url": "https://yourdomain.com",
"logo": "https://yourdomain.com/assets/logo.png",
"sameAs": [
"https://twitter.com/agensiai",
"https://github.com/agensiai"
],
"founder": {
"@type": "Person",
"name": "Jane Doe",
"jobTitle": "Founder & Principal Architect",
"sameAs": "https://linkedin.com/in/janedoe"
}
}
```
### 2. Interactive Asset Definition (`WebApplication` for Tools)
Optimizes visibility for interactive assets, free utility scripts, or calculators, compelling high-intent backlink aggregation.
```json
{
"@context": "https://schema.org",
"@type": "WebApplication",
"@id": "https://yourdomain.com/tools/freelance-budget-calculator/#webapp",
"name": "Advanced Freelance Budget Calculator",
"url": "https://yourdomain.com/tools/freelance-budget-calculator",
"applicationCategory": "BusinessApplication",
"operatingSystem": "All",
"browserRequirements": "Requires JavaScript. Requires HTML5.",
"offers": {
"@type": "Offer",
"price": "0",
"priceCurrency": "USD"
}
}
```
### 3. Editorial Authority Injection (`BlogPosting` / `Article`)
Validates author entities and publication timelines for index and generative aggregators.
```json
{
"@context": "https://schema.org",
"@type": "BlogPosting",
"headline": "How to Configure Custom SKILL.md Files for Claude Code",
"datePublished": "2026-06-15T08:00:00+00:00",
"dateModified": "2026-07-18T14:30:00+00:00",
"author": {
"@type": "Person",
"name": "Jane Doe",
"url": "https://yourdomain.com/about"
},
"publisher": {
"@type": "Organization",
"name": "Agensi AI",
"logo": {
"@type": "ImageObject",
"url": "https://yourdomain.com/assets/logo.png"
}
},
"description": "A deep technical guide showing exactly where to store and how to structuralize custom skill markdown files inside the Claude Code runtime environment."
}
```
---
## 5. Generative Engine Optimization (GEO) & Machine-Readable Layer
### The llms.txt Specification
The `llms.txt` file is a newly established standard served at the root directory (`/llms.txt`) designed to provide structured context directly to modern LLM wrappers, developer agents, and AI aggregators. It acts as an intentional sitemap tailored for machine comprehension.
#### Execution Blueprint: Root `/llms.txt`
```markdown
# Agensi AI
> A premium open-source marketplace and repository for modular SKILL.md configurations designed to optimize Claude Code, Cursor, and Codex CLI runtimes.
## Core Products & Tools
- [Freelance Budget Calculator](/tools/freelance-budget-calculator): An interactive utility providing multi-currency overhead projections and baseline hourly rate metrics for engineering consultants.
- [AI Agent Skills Marketplace](/skills): The primary index containing production-tested skill schemas categorized by execution environment.
## Technical Guides & Documentation
- [Claude Code Skill Installation Guide](/blog/installing-claude-skills): Detailed technical instructions detailing shell paths and environment parameters required to inject local skill matrices.
- [Structuring Valid SKILL.md Schemas](/blog/structuring-skills-guide): An architecture review defining semantic headings, model constraints, and system prompt formatting rules.
```
#### Secondary Layer: Content Specific `/llms-full.txt`
For expansive document ecosystems, deploy an `/llms-full.txt` asset that appends raw text blocks, systemic code snippets, or markdown definitions of the core documentation pages directly inside a single file. This allows AI utilities to consume your site's entire informational matrix in a single, high-density context fill.
### Citation Engineering
To maximize the likelihood that an LLM outputs your brand name or URL reference when answering questions:
* **High-Density Technical Nomenclature:** Incorporate precise industry terminology, version numbers, framework metrics, and distinct jargon. LLMs rely heavily on strong semantic embeddings to bridge query concepts to source documents.
* **Contextual Validation:** Wrap key conclusions with high-authority phrase structures. *Example: "Based on multi-variant field benchmarks executed across 96 standalone React instances, optimizing the SSR payload structure directly increased citation likelihood by 219%..."*
---
## 6. Data-Driven Growth Loop & Continuous Integration
Transition SEO away from arbitrary content generation and structure it as a programmatic, closed-loop engineering cycle driven by exact metrics.
```
┌────────────────────────────────────────────────────────┐
│ THE WEEKLY METRIC GROWTH LOOP │
└───────────────────────────┬────────────────────────────┘
│
▼
┌───────────────────────────────────┐
│ Export GSC Data & Analysis │
│ (Isolate impressions vs. clicks) │
└─────────────────┬─────────────────┘
│
▼
┌───────────────────────────────────┐
│ Isolate Content Defects │
│ (Fix cannibalization & low CTR) │
└─────────────────┬─────────────────┘
│
▼
┌───────────────────────────────────┐
│ Content & Structure Injection │
│ (Deploy question H2s + Schema) │
└─────────────────┬─────────────────┘
│
▼
┌───────────────────────────────────┐
│ Trigger Multi-Engine Ping │
│ (Deploy IndexNow + Manual GSC) │
└───────────────────────────────────┘
```
### The Weekly Analytical Engine Loop
1. **Metric Extraction:** Every 7 days, export user interaction datasets from Google Search Console (GSC) and Bing Webmaster Tools, capturing `Queries`, `Pages`, `Impressions`, `Clicks`, and `Average Positions`.
2. **Gap Isolation:** Feed the data to an analytical instance or processing pipeline to compute specific programmatic anomalies:
* *High Impressions / Zero Clicks:* Identifies queries where AI Overviews are intercepting clicks, signaling the need to refactor content into an explicit **Atomic Quick-Answer Block** to win the AI citation position.
* *Keyword Cannibalization:* Isolates instances where multiple distinct URIs compete for identical search tokens. Remediate by combining duplicate bodies and implementing strict 301 server redirects or defining unambiguous `<link rel="canonical" href="...">` declarations.
* *Low Click-Through Rate (CTR) at High Positions:* Signals truncated titles or generic description tags. Keep final `<title>` elements strictly **under 60 characters** to ensure clean cross-engine rendering.
3. **Content Refactoring & Synthesis:** Authors or automated pipelines update weak nodes based on the gap analysis, ensuring all structural criteria are met (Question H2 layouts, data tables, structural JSON-LD).
4. **Syndication Broadcast:** Issue immediate index execution commands via IndexNow for Bing/Yandex, and execute API-driven or manual fetch requests via the GSC URL Inspection interface to guarantee accelerated processing.