Good morning.
For twenty years, search worked one way. Rank on Google. Earn the click. Capture the traffic.
That game still matters. Google still sends billions of visits per day. But a second game is running alongside it now, and most businesses haven't noticed.
A growing share of queries never reach a search results page. They go to ChatGPT. Claude. Perplexity. Gemini. The user asks a question and gets one synthesized answer, sometimes with citations, sometimes without. No ten blue links. No browsing. One answer, and done.
If your content gets cited, you get traffic and credibility from a source people trust. If it doesn't get cited, you're invisible in that channel entirely. There's no second-place finish.
This issue covers both games. The first section breaks new ground with a framework for AI search optimization that most businesses haven't touched yet. The second section makes traditional SEO systematic with skills you can deploy repeatedly. Both disciplines reinforce each other. Winning one makes winning the other easier.
— Sam
IN TODAY’S ISSUE 🤖
The GEO framework: how AI models choose what to cite
Traditional SEO skills: audit, technical, programmatic, competitive
How the two disciplines reinforce each other
Your project: the dual-channel search audit
Download the skills
Let’s get into it.
1. The New Search Game
First, important:
If you haven't set up your skill environment and you’re not using Claude yet, start with the previous issue. This issue assumes you have skills installed and understand the basics.
If you’re ready, let’s move on:
Every SEO guide covers Google. This section covers the game nobody has a playbook for yet.
When someone asks Perplexity "what's the best project management tool for freelancers," it searches the web, reads multiple pages, then synthesizes one answer with citations. The user reads Perplexity's answer. Maybe clicks a citation. Maybe doesn't. Either way, the sources Perplexity chose to cite got the value. Everyone else got nothing.
The same pattern applies across ChatGPT with search, Claude with web search, Google's AI Overviews, and every other AI interface that retrieves and synthesizes web content.
This is Generative Engine Optimization. GEO. And the mechanics differ from traditional SEO in ways that matter.
How AI Models Select Sources
Three layers determine whether your content appears in AI-generated answers.
Layer 1: Training data. Large language models learn from the web. Content that appears across many authoritative sources gets absorbed into model weights during training. When a model answers a question "from memory" without searching, it's drawing on this layer. You can't optimize for a specific query here, but consistent, high-quality publishing in your domain builds presence in future training data over time.
Layer 2: Retrieval. Perplexity, ChatGPT with browsing, and Claude with web search all retrieve real-time content. They run a search query, get results, then read the actual pages. This layer is directly optimizable. The model searches, retrieves, reads, and selects. What you write and how you structure it determines whether your content gets selected.
Layer 3: Citation selection. After retrieving content, the model decides which sources to cite in its answer. This is where the competition happens. The model has read 5-10 sources. It cites 2-3. What separates the cited from the skipped?
Observable patterns from testing across multiple AI search tools:
Gets Cited | Gets Skipped |
Direct, specific answers to the query | Vague overviews that hedge every claim |
Original data, statistics, research | Repackaged secondary content |
Clear structure with extractable claims | Walls of text without clear takeaways |
Named entities and specific details | Generic advice without examples |
Authoritative domain with expertise signals | Thin content on generic domains |
Recent, dated content on time-sensitive topics | Undated or clearly stale content |
You can test this yourself. Ask ChatGPT or Perplexity a question your content should answer. Look at what gets cited. Look at what those sources have in common. The patterns are consistent.
The CITE Framework
This is the optimization framework for AI search citability:
C — Citable Structure
AI models extract individual claims from content. If your content is structured so that specific statements stand alone as quotable facts, it's more likely to be cited.
What this looks like in practice:
Clear topic sentences that make a specific claim per paragraph
Headers that telegraph what follows (not clever, not vague)
Tables and lists with concrete data points
Statements that answer a question directly without requiring surrounding context
Paragraphs that each make one clear point
I — Information Density
AI models prefer sources that contain original, specific information. Density means every paragraph adds something new.
What counts as dense:
Original data, percentages, specific numbers
Named examples with real companies, people, or products
First-party research, surveys, or analysis
Step-by-step instructions with specific actions (not "consider doing X")
Definitions and clear explanations that resolve ambiguity
What counts as thin:
Rephrasing common knowledge without adding insight
"Experts say" without naming the experts
Advice that applies to everything and therefore applies to nothing
Summaries of other summaries
T — Topical Authority
AI models (especially retrieval systems) assess whether a source has depth in a topic area. One article doesn't establish authority. A body of work does.
How to build it:
Multiple pieces covering the same domain from different angles
Internal linking between related content
Author credentials visible on the page and verifiable
Consistent publishing in the same subject area over time
Content that other sources reference and link to
E — Entity Clarity
AI models work with entities: people, companies, products, concepts. Content that clearly identifies and consistently names entities is easier for models to parse and cite.
What this means:
Use proper nouns consistently (don't swap "the company" for the actual name)
Define terms when first used
Use schema markup to reinforce entity relationships (covered in the seo-technical skill)
Answer "what is X" and "how does X work" directly when introducing concepts
What GEO Is Not
Some honest framing, because this discipline is new.
GEO is not a guaranteed way to get cited. Models are probabilistic. The same query can produce different citations on different days. There's no equivalent to "ranking #1" because there's no fixed result page.
GEO is not separate from good content strategy. Everything in the CITE framework also makes content better for human readers. Citable structure is readable structure. Information density is useful content. This isn't a hack. It's disciplined content creation with structural awareness.
GEO is not static. AI search is evolving fast. The CITE framework is based on observable patterns in current AI tools. These patterns may shift as models and retrieval systems improve. What won't change: models will always prefer clear, specific, authoritative content over vague, thin, generic content.
2. Traditional SEO — Made Systematic
GEO is where search is going. SEO is where search still is. Google still drives most discovery traffic for most businesses. These four skills make the established discipline repeatable and thorough.
The geo-optimization Skill
What it does: Audits content for AI search citability. Scores your content against the CITE framework and identifies specific changes that improve your chances of being cited.
When it triggers: Requests to optimize for AI search, improve citability, audit content for AI visibility.
Key frameworks:
CITE framework scoring (Citable structure, Information density, Topical authority, Entity clarity)
Content citability scorecard
AI search query classification (informational, comparative, recommendation)
Source selection gap analysis
Red flags it looks for:
Content that summarizes without adding original insight
No specific data, numbers, or named examples anywhere
Vague headers like "Introduction" or "Overview" or "Conclusion"
No author attribution or expertise signals
Undated content on time-sensitive topics
No single paragraph that answers a specific question directly
Example prompt: "Audit this blog post for AI search citability. Score it against the CITE framework and tell me what specific changes would make it more likely to be cited by ChatGPT, Perplexity, or Claude."
Here’s an example of what the SKILL.md file would look like:
---
name: geo-optimization
description: Audit and optimize content for AI search citability. Use when improving chances of being cited by ChatGPT, Perplexity, Claude, or other AI models that retrieve and synthesize web content.
---
# GEO Optimization
Generative Engine Optimization. This skill audits content for AI search citability using the CITE framework and provides specific recommendations to improve the likelihood that AI models will cite your content as a source.
## When to Use This Skill
- Auditing existing content for AI search citability
- Creating new content designed for AI search visibility
- Analyzing why competitors get cited and you don't
- Improving content structure for machine readability
- Evaluating content after AI search testing
- Building a content strategy that serves both Google and AI interfaces
## How AI Models Select Sources
### The Three Layers
**Layer 1: Training Data**
Models learn from the web during training. Content that appears across multiple authoritative sources gets absorbed into model weights. This is a long-term play. Consistent, high-quality publishing builds presence in future training data. Not directly optimizable per-query, but directionally important.
**Layer 2: Retrieval**
Perplexity, ChatGPT with browsing, Claude with web search, and Google AI Overviews all retrieve real-time content. They search, read pages, and select which sources to use. This layer is directly optimizable. What you write and how you structure it determines whether your content gets retrieved and read.
**Layer 3: Citation Selection**
After reading 5-10 sources, the model cites 2-3 in its answer. This is where competition happens. The model has options. It selects based on relevance, specificity, authority, and extractability.
### Citation Selection Patterns
Based on observable behavior across multiple AI search tools:
| Gets Cited | Gets Skipped |
|-----------|-------------|
| Direct, specific answers to the query | Vague overviews that hedge every claim |
| Original data, statistics, research | Repackaged secondary content |
| Clear structure with extractable claims | Walls of text without clear takeaways |
| Named entities and specific details | Generic advice without examples |
| Authoritative domain with expertise signals | Thin content on generic domains |
| Recent, dated content on time-sensitive topics | Undated or clearly stale content |
| Content that commits to a position | Content that lists every option without evaluating |
## The CITE Framework
### C — Citable Structure
Content organized so individual claims can be extracted by a model without requiring full-page context.
**Checklist:**
| Element | Citable | Not Citable |
|---------|---------|-------------|
| Paragraph structure | Each paragraph makes one clear, extractable claim | Paragraphs blend multiple ideas with no clear takeaway |
| Headers | Specific, descriptive ("How to Calculate CAC" ) | Vague or clever ("Getting Started," "The Big Picture") |
| Data presentation | Tables and lists with concrete data points | Data buried inside prose paragraphs |
| Definitions | Clear, direct definitions when introducing terms | Terms used without explanation |
| Answers | Direct answer in first sentence, then elaboration | Answer buried after three paragraphs of context |
**The Inverted Pyramid Test:**
For any section, can you read just the first sentence of each paragraph and get the key claims? If yes, the content is citable. If no, restructure.
**Practical actions:**
- Lead each section with its strongest, most specific claim
- Use headers that contain the keyword or question being answered
- Put data in tables rather than embedding numbers in prose
- Write topic sentences that stand alone as quotable facts
- Break long paragraphs into single-claim paragraphs
### I — Information Density
Every paragraph adds something the reader (or model) couldn't get from a generic source.
**Density scoring:**
| Density Level | What It Looks Like |
|---------------|-------------------|
| High | Original data, specific percentages, named case studies, first-party research, unique analysis |
| Medium | Curated examples, specific recommendations with reasoning, expert attribution with names |
| Low | General advice that applies to everything, "experts say" without naming them, restated common knowledge |
| Zero | Filler paragraphs, throat-clearing intros, conclusion summaries that repeat earlier content |
**Information types that increase density:**
- Specific numbers ("43% of users abandon" vs. "many users abandon")
- Named examples ("Stripe's onboarding flow" vs. "a popular SaaS company")
- Original analysis (your interpretation of data, not restating someone else's)
- Concrete steps ("Set X to Y in Z" vs. "configure your settings appropriately")
- Comparisons with specifics ("Tool A costs $49/mo and includes X; Tool B costs $79/mo but adds Y")
- Timelines and benchmarks ("expect results in 2-4 weeks with 1,000+ monthly visitors")
**The Replacement Test:**
For any paragraph, ask: could this paragraph appear in any article on this topic? If yes, it's generic. Replace with something only you could write based on your data, experience, or analysis.
### T — Topical Authority
Retrieval systems assess whether a source has depth and credibility in the topic area.
**Authority signals:**
| Signal | How to Build It |
|--------|----------------|
| Content depth | Multiple articles covering the same topic from different angles |
| Internal linking | Pieces that reference and link to each other |
| Author expertise | Visible credentials, bio, verifiable experience |
| Publishing consistency | Regular content in the same domain over months/years |
| External validation | Other authoritative sources citing or linking to your content |
| Content freshness | Updated dates, current information, maintained accuracy |
**Topical authority audit questions:**
1. How many pieces do you have on this topic? (1 piece = weak, 5+ = strong)
2. Do pieces link to each other? (Isolated pieces = weak, interconnected = strong)
3. Is author expertise visible on the page? (Anonymous = weak, credentials shown = strong)
4. How recently was content updated? (2+ years old = weak, <6 months = strong)
5. Do other sites reference your content? (No inbound references = weak)
**Building authority over time:**
- Create a pillar page on your core topic
- Build supporting content that covers subtopics in depth
- Link supporting content to and from the pillar page
- Update content when information changes
- Add original data or research when available
### E — Entity Clarity
AI models parse content using entities: named people, companies, products, concepts. Clear entity usage makes content easier for models to understand, categorize, and cite.
**Entity clarity checklist:**
- Use proper nouns consistently (don't alternate between "the company," "the firm," "the organization")
- Define terms when first introduced
- Use the same name for the same entity throughout (pick "Google Analytics 4" or "GA4" and stick with one)
- Include structured data (schema markup) that reinforces entity relationships
- Answer "what is X" directly when introducing a concept
- Name people, companies, and products rather than using vague references
**Schema markup for entity clarity:**
- Article schema with author information
- Organization schema for your business
- FAQ schema for question-and-answer content
- HowTo schema for procedural content
(See `seo-technical` skill for implementation details.)
## AI Search Query Classification
Different query types require different optimization approaches:
| Query Type | Example | What Gets Cited |
|-----------|---------|-----------------|
| Informational | "What is conversion rate optimization?" | Clear definitions, comprehensive explanations, authoritative sources |
| Comparative | "Notion vs. Asana for project management" | Specific feature comparisons, honest tradeoffs, hands-on experience |
| Recommendation | "Best email marketing tool for small business" | Opinionated recommendations with reasoning, specific use cases, pricing details |
| How-to | "How to set up Google Analytics 4" | Step-by-step instructions, specific settings, screenshots/details |
| Current events | "Latest changes to Instagram algorithm" | Dated content, recent publication, specific details about what changed |
**Optimize based on the queries you want to appear in.** Informational queries reward depth and definitions. Comparative queries reward specifics and honest tradeoffs. Recommendation queries reward opinionated, well-reasoned positions.
## Red Flags
Issues that reduce AI search citability:
- [ ] Content summarizes others without adding original insight
- [ ] No specific data, numbers, or named examples in the entire piece
- [ ] Headers are vague ("Introduction," "Overview," "Conclusion," "Key Takeaways")
- [ ] No author attribution or visible expertise signals
- [ ] Content is undated on a time-sensitive topic
- [ ] No single paragraph directly answers a specific question
- [ ] Content hedges every claim ("it depends," "results may vary" as the entire position)
- [ ] Entities referenced vaguely ("a leading company" instead of naming it)
- [ ] Content is thin on a topic competitors cover in depth
- [ ] No internal links to related content on the same topic
- [ ] Page has no schema markup
- [ ] Content reads identically to 10 other articles on the same topic
## Output Format
When auditing content for AI search citability:
```
## GEO Audit: [Content Title/URL]
### CITE Framework Score
| Dimension | Score (1-5) | Key Issues |
|-----------|------------|------------|
| Citable Structure | ___ | [Primary issue] |
| Information Density | ___ | [Primary issue] |
| Topical Authority | ___ | [Primary issue] |
| Entity Clarity | ___ | [Primary issue] |
| **Overall** | **___** | |
### Target Query Analysis
| Query This Should Answer | Would It Get Cited? | Why/Why Not |
|--------------------------|-------------------|-------------|
| [Query 1] | Yes/No/Maybe | [Reason] |
| [Query 2] | Yes/No/Maybe | [Reason] |
### Citable Structure Assessment
- Extractable claims per section: [Assessment]
- Header specificity: [Assessment]
- Inverted pyramid structure: [Present/Missing]
- Key issue: [Most important structural problem]
### Information Density Assessment
- Original data/research: [Present/Missing]
- Named examples: [Count and quality]
- Specific vs. generic ratio: [Assessment]
- Key issue: [Most important density problem]
### Topical Authority Assessment
- Related content on domain: [Count]
- Internal linking: [Present/Missing]
- Author credentials visible: [Yes/No]
- Content freshness: [Date and assessment]
- Key issue: [Most important authority problem]
### Entity Clarity Assessment
- Consistent entity naming: [Yes/No]
- Terms defined: [Yes/No]
- Schema markup: [Present/Missing]
- Key issue: [Most important clarity problem]
### Prioritized Recommendations
| Priority | Recommendation | CITE Dimension | Expected Impact |
|----------|---------------|----------------|-----------------|
| 1 | [Specific action] | [C/I/T/E] | [High/Med/Low] |
| 2 | [Specific action] | [C/I/T/E] | [High/Med/Low] |
| 3 | [Specific action] | [C/I/T/E] | [High/Med/Low] |
| 4 | [Specific action] | [C/I/T/E] | [High/Med/Low] |
| 5 | [Specific action] | [C/I/T/E] | [High/Med/Low] |
### Quick Wins
[Changes that take <1 hour and meaningfully improve citability]
### Structural Improvements
[Changes that require more work but have lasting impact]
### Rewrite Suggestions
[If specific sections need rewriting, provide guidance on what to change]
```
## Testing Your Optimization
After making changes, test effectiveness:
1. **Run AI search queries** that your content should answer
2. **Document** which sources get cited (including yours or not)
3. **Compare** cited sources against your updated content
4. **Re-run** the same queries 1-2 weeks later to track changes
5. **Iterate** based on what you observe
Testing template:
```
QUERY: _______________
MODEL: [ChatGPT / Perplexity / Claude]
DATE: _______________
CITED: [Yes/No]
SOURCES CITED: [List URLs]
WHAT CITED SOURCES HAVE THAT MINE DOESN'T: _______________
ACTION: _______________
```
## Chaining to Other Skills
GEO optimization connects to broader search strategy:
- **Technical issues blocking retrieval** → Chain to `seo-technical` for crawlability and schema
- **Content not ranking on Google either** → Chain to `seo-audit` for on-page optimization
- **Competitors dominating both channels** → Chain to `competitor-seo` for gap analysis
- **Need to build topical authority at scale** → Chain to `programmatic-seo` for content volume
When chaining, pass along: CITE scores, specific gaps identified, AI search test results.The seo-audit Skill
What it does: On-page SEO analysis for any URL or content piece. Covers title tags, meta descriptions, header structure, keyword usage, internal linking, content quality signals, and search intent alignment.
When it triggers: Requests to audit SEO, check on-page optimization, review content for search performance.
Key frameworks:
On-page element checklist (title, meta, headers, URL, images)
Keyword intent alignment (informational, navigational, transactional, commercial)
Content quality signals (depth, originality, E-E-A-T markers)
Internal linking assessment
Search intent match analysis
Red flags:
Title tag missing target keyword or over 60 characters
Meta description missing or over 155 characters
No H1 or multiple H1s
Keyword stuffing or unnatural density
Thin content (<300 words for a competitive topic)
No internal links to or from the page
Content doesn't match the search intent behind the target keyword
Example prompt: "Audit this page for on-page SEO. Here's the content and my target keyword: [keyword]. What's working, what's broken, and what should I fix first?"
Here’s an example of what the SKILL.md file would look like:
---
name: seo-audit
description: On-page SEO analysis for any URL or content piece. Use for auditing search performance, diagnosing ranking issues, checking keyword targeting, or optimizing content before publishing.
---
# SEO Audit
Comprehensive on-page SEO analysis. This skill evaluates title tags, meta descriptions, header structure, keyword usage, content quality, internal linking, and search intent alignment to identify why pages underperform and what to fix first.
## When to Use This Skill
- Auditing existing pages for search performance
- Optimizing content before publishing
- Diagnosing why a page isn't ranking for target keywords
- Checking keyword targeting and intent alignment
- Reviewing content quality signals
- Comparing your page against ranking competitors
## On-Page Element Checklist
### Title Tag
| Criteria | Pass | Fail |
|----------|------|------|
| Contains target keyword | Keyword present, preferably near the front | Keyword missing or buried at the end |
| Length | 50-60 characters | Under 30 or over 60 (truncated in SERPs) |
| Unique | Different from all other page titles on site | Duplicates another page's title |
| Compelling | Gives reason to click over competitors | Generic or purely descriptive |
| Brand | Brand name included (usually at end) | Missing when brand has recognition |
### Meta Description
| Criteria | Pass | Fail |
|----------|------|------|
| Contains target keyword | Keyword present naturally | Keyword missing (won't be bolded in SERPs) |
| Length | 120-155 characters | Under 70 or over 155 (truncated) |
| Includes CTA or value prop | Gives searcher reason to click | Purely descriptive without motivation |
| Unique | Different from all other meta descriptions | Duplicates another page or is auto-generated |
| Matches intent | Aligns with what the searcher wants | Promises something the page doesn't deliver |
### Header Structure
| Element | Best Practice |
|---------|---------------|
| H1 | Exactly one per page, contains target keyword, matches page topic |
| H2s | Logical subsections, include related keywords naturally |
| H3s | Sub-subsections where needed, not used for styling |
| Hierarchy | H1 → H2 → H3 (never skip levels) |
| Keyword placement | Primary keyword in H1, secondary/related keywords in H2s |
### URL Structure
| Criteria | Good | Bad |
|----------|------|-----|
| Length | Short, descriptive | Long, parameter-heavy |
| Keywords | Contains target keyword | No keyword relevance |
| Format | Lowercase, hyphens | Underscores, mixed case, special characters |
| Hierarchy | Reflects site structure | Flat with no logical path |
| Example | `/blog/conversion-rate-optimization` | `/p?id=4827&cat=3` |
### Image Optimization
| Element | Requirement |
|---------|-------------|
| Alt text | Descriptive, includes keyword where natural, under 125 characters |
| File name | Descriptive with hyphens (`conversion-funnel-diagram.png`) |
| File size | Compressed for web (use WebP where supported) |
| Dimensions | Sized appropriately (not relying on CSS to resize large images) |
| Lazy loading | Implemented for below-fold images |
## Search Intent Analysis
Every keyword has intent. Content must match it.
### Intent Types
| Intent | What Searcher Wants | Content Format |
|--------|-------------------|----------------|
| Informational | Learn or understand something | Guides, tutorials, explanations, definitions |
| Navigational | Find a specific page or site | Brand pages, login pages, specific product pages |
| Commercial | Research before buying | Comparisons, reviews, "best of" lists, vs. pages |
| Transactional | Complete an action or purchase | Product pages, pricing pages, signup pages |
### Intent Alignment Check
1. Search your target keyword in Google
2. Look at the top 5 results
3. What content type dominates? (Guides? Lists? Product pages? Videos?)
4. What format are they? (Long-form? Short? With images? With tools?)
5. Does your content match this pattern?
If the top results are all comparison articles and you've written a single-product review, you have an intent mismatch. The content format matters as much as the content quality.
## Content Quality Signals
### E-E-A-T Assessment
| Signal | What to Check |
|--------|---------------|
| Experience | Does the content show first-hand experience with the topic? |
| Expertise | Is the author qualified? Are credentials visible? |
| Authoritativeness | Is the site recognized in this topic area? |
| Trustworthiness | Is information accurate, sourced, and current? |
**Practical E-E-A-T improvements:**
- Add author bio with relevant credentials
- Include first-person experience ("We tested this..." "In our experience...")
- Cite sources for claims and statistics
- Show date published and date last updated
- Include original images, screenshots, or data
### Content Depth Analysis
| Indicator | Sufficient | Insufficient |
|-----------|-----------|-------------|
| Word count vs. competitors | Comparable or deeper | Significantly thinner |
| Subtopics covered | All major subtopics addressed | Key subtopics missing |
| Questions answered | Covers "People Also Ask" queries | Leaves common questions unanswered |
| Original value | Adds perspective, data, or insight | Restates what's already ranking |
| Freshness | Information is current | Contains outdated information |
### Internal Linking
| Check | What to Look For |
|-------|-----------------|
| Links to this page | Other relevant pages on your site link to this one |
| Links from this page | This page links to related content on your site |
| Anchor text | Descriptive, keyword-relevant (not "click here") |
| Link depth | Page is reachable within 3 clicks from homepage |
| Broken links | No links pointing to 404s or redirects |
## Keyword Analysis
### Primary Keyword Assessment
| Factor | Evaluation |
|--------|-----------|
| Placement | Present in title, H1, first 100 words, URL, meta description |
| Density | Natural usage (typically 0.5-2%), not forced |
| Variations | Synonyms and related terms used naturally |
| LSI keywords | Related concepts and terminology included |
### Keyword Cannibalization Check
Multiple pages targeting the same keyword compete against each other.
**How to check:**
1. Search `site:yourdomain.com "target keyword"` in Google
2. If multiple pages appear, you have cannibalization
3. Solution: Consolidate content, differentiate targeting, or use canonical tags
## Red Flags
Issues that commonly hurt search performance:
- [ ] Title tag missing target keyword or over 60 characters
- [ ] Meta description missing, duplicate, or over 155 characters
- [ ] No H1 tag, or multiple H1 tags on one page
- [ ] H1 doesn't contain or relate to target keyword
- [ ] Content doesn't match search intent (format mismatch)
- [ ] Thin content compared to ranking competitors
- [ ] Keyword stuffed (unnatural, forced repetition)
- [ ] No internal links to or from the page
- [ ] Images missing alt text
- [ ] Content is outdated with no "last updated" date
- [ ] No author attribution or expertise signals
- [ ] Key subtopics covered by competitors are missing
- [ ] Page is more than 3 clicks from homepage
- [ ] URL is non-descriptive or contains parameters
- [ ] No external citations or sources for claims
## Output Format
When auditing a page:
```
## SEO Audit: [Page Title / URL]
### Page Overview
- Target keyword: [Primary keyword]
- Secondary keywords: [List]
- Search intent: [Informational / Commercial / Transactional / Navigational]
- Current ranking: [Position if known]
- Current traffic: [Monthly organic visits if known]
### On-Page Elements
| Element | Current | Status | Recommendation |
|---------|---------|--------|----------------|
| Title tag | "[Current title]" | Pass/Fail | [Fix if needed] |
| Meta description | "[Current meta]" | Pass/Fail | [Fix if needed] |
| H1 | "[Current H1]" | Pass/Fail | [Fix if needed] |
| URL | [Current URL] | Pass/Fail | [Fix if needed] |
| Images | [Alt text status] | Pass/Fail | [Fix if needed] |
### Header Structure
[H1 → H2 → H3 hierarchy assessment]
### Search Intent Alignment
- Target keyword intent: [Type]
- Top SERP format: [What's ranking]
- Content format match: [Yes/No/Partial]
- Recommendation: [If mismatch, what to change]
### Content Quality Assessment
- Word count: [Count] vs. competitor average [Count]
- E-E-A-T signals: [Present/Missing]
- Original value: [Assessment]
- Subtopics covered: [Covered/Missing]
- Freshness: [Current/Outdated]
### Keyword Analysis
- Primary keyword placement: [Where it appears]
- Density: [Percentage and assessment]
- Related keywords: [Present/Missing]
- Cannibalization risk: [Yes/No]
### Internal Linking
- Links to this page: [Count and quality]
- Links from this page: [Count and quality]
- Broken links: [Count]
### Red Flags
1. [Issue with specific evidence]
2. [Issue with specific evidence]
3. [Issue with specific evidence]
### Prioritized Recommendations
| Priority | Issue | Recommendation | Expected Impact |
|----------|-------|----------------|-----------------|
| 1 | [Issue] | [Specific fix] | [High/Med/Low] |
| 2 | [Issue] | [Specific fix] | [High/Med/Low] |
| 3 | [Issue] | [Specific fix] | [High/Med/Low] |
### Title Tag Alternatives
[If title needs rewriting, provide 2-3 options]
### Meta Description Alternatives
[If meta needs rewriting, provide 2-3 options]
### Content Recommendations
[Specific sections to add, expand, or restructure]
### Next Steps
1. [Immediate fix]
2. [Secondary fix]
3. [Longer-term improvement]
```
## Chaining to Other Skills
SEO audits often reveal issues requiring deeper analysis:
- **Technical problems (speed, schema, indexing)** → Chain to `seo-technical`
- **Competitor content outranking you** → Chain to `competitor-seo` for gap analysis
- **Content not being cited by AI models** → Chain to `geo-optimization` for CITE framework audit
- **Need to scale content for multiple keywords** → Chain to `programmatic-seo`
When chaining, pass along: target keywords, current rankings, content quality assessment, specific gaps identified.The seo-technical Skill
What it does: Technical SEO analysis covering site architecture, crawlability, indexation, page speed, schema markup, and structured data. The infrastructure that determines whether good content gets discovered.
When it triggers: Questions about technical SEO, site speed, schema markup, structured data, crawl issues, indexation, Core Web Vitals.
Key frameworks:
Technical health checklist (robots.txt, XML sitemap, canonical tags, redirects)
Core Web Vitals assessment (LCP, INP, CLS)
Schema markup selection and implementation by page type
Crawl budget optimization
Mobile-first indexing requirements
Site architecture and URL hierarchy
Schema markup guidance (folded into this skill rather than a standalone):
Schema type selection by page type (Article, Product, FAQ, HowTo, LocalBusiness, Organization)
JSON-LD implementation templates
Rich snippet eligibility assessment
Testing with Google's Rich Results Test
Common schema mistakes that invalidate markup
Red flags:
No XML sitemap or outdated sitemap
Pages blocked by robots.txt unintentionally
Missing or incorrect canonical tags
No schema markup on key page types
Core Web Vitals failing (LCP >2.5s, CLS >0.1)
No HTTPS
Broken internal links or redirect chains (3+ hops)
Duplicate content without canonicalization
Mobile rendering issues or missing mobile viewport
Example prompt:
Review the technical SEO health of my site. Here's my setup: [platform, hosting, known issues]. What schema markup should I add to my [page type] pages, and what technical issues should I fix first?The programmatic-seo Skill
What it does: Strategy and execution for building SEO-optimized pages at scale. Template design, data sources, content patterns, and quality controls for creating dozens or hundreds of pages from structured data.
When it triggers: Requests about scaling content production, building template-based pages, programmatic SEO, creating comparison or location pages at volume.
Why it matters for solo operators: One person with AI and a good template can build 200 optimized pages that would have required a content team and months of manual work. Comparison pages, location pages, "best X for Y" pages, tool directories, glossary pages. High leverage.
Key frameworks:
Page type selection (comparison, location, tool, glossary, "best X for Y," alternatives)
Template design (fixed structure + variable data + unique value per page)
Data source identification (public data, APIs, proprietary research)
Quality control checklist (each page must add unique value beyond variable swaps)
Internal linking architecture for programmatic page sets
Staged launch and indexation strategy
Red flags:
Thin pages with only variable data swapped and no unique analysis
No differentiation between pages beyond keyword/location insertion
Template produces content that reads as obviously AI-generated
No internal linking strategy connecting the page set
Targeting keywords with zero search volume
Publishing hundreds of pages simultaneously (crawl budget issues)
Pages that exist for Google but provide no value to humans
Example prompt:
I want to build comparison pages for [product category]. I have data on 50 products. Help me design a template that produces pages with genuine unique value, not just keyword-stuffed variations.The competitor-seo Skill
What it does: SEO competitive analysis. Identifies keyword gaps, content opportunities, backlink patterns, and structural advantages competitors have that you can target.
When it triggers: Requests for competitor SEO analysis, keyword gaps, content opportunities, competitive search positioning.
Key frameworks:
Keyword gap analysis (they rank, you don't)
Content gap analysis (topics they cover, you don't)
SERP feature ownership (featured snippets, People Also Ask, image packs)
Content format analysis (what types of content rank in your niche)
Backlink profile comparison
Competitive content quality assessment
Red flags:
Targeting identical keywords as competitors with 10x your domain authority
Ignoring long-tail opportunities where you can actually win
Copying competitor content strategy without understanding their structural advantages
No awareness of which SERP features competitors own
Competing on volume instead of depth and specificity
Example prompt:
Here are my top 3 competitors: [URLs]. I currently rank for roughly [X] keywords. Find the keyword and content gaps where they rank and I don't, and identify the easiest opportunities to close first.If you want all the SEO Skills, join Cortex (see details below)
3. Two Games, One System
These two disciplines look separate. They reinforce each other.
Good SEO structure (clear headers, schema markup, logical organization) directly improves GEO citability. AI retrieval systems read your pages the same way Googlebot does. If your content is well-structured for crawlers, it's well-structured for AI models.
Content built for AI citability (original data, specific claims, entity clarity) also ranks better on Google. Google's own quality guidelines increasingly reward the same signals the CITE framework targets: expertise, specificity, and original value.
Programmatic SEO pages built with CITE principles serve both channels. A comparison page with real data, specific claims, and clear structure ranks on Google and gets cited by Perplexity.
Competitor SEO gaps often reveal GEO opportunities. If no competitor has genuinely citable content on a topic, there's an opening in both channels simultaneously.
The skills work together. Optimizing for one channel accelerates the other. That's the system advantage.
4. Your Project: The Dual-Channel Search Audit
Time: 1-2 business days
Output: Search audit covering both Google SEO and AI citability, with prioritized action plan
Getting Content Into Claude
Use the same methods from Issue #2:
Markdown export (MarkDownload extension, Reader Mode + copy) for accurate text capture
Manual capture for JavaScript-heavy pages or gated content
Screenshots for layout and visual hierarchy as a supplement
For SEO audits specifically, also capture:
Title tag (view page source or browser tab)
Meta description (view source or SEO browser extension)
URL structure
Any existing schema markup (view source, search for "application/ld+json")
Phase 1: Select Your Audit Target (30 minutes)
Pick one content asset or page that meets at least two of these criteria:
Already gets some organic traffic
Covers a topic you want to own
You can edit and republish it
Competitors rank for the same topic
If unsure, start with your homepage or your highest-traffic blog post.
Gather what you have:
Search Console data (impressions, clicks, average position)
Target keywords for this page
URLs of 2-3 competitors ranking for the same terms
Current page content (markdown export)
If you don't have Search Console data, use estimates. Check your analytics for organic traffic, or search your target keywords and note where you appear (if at all).
Phase 2: Google SEO Audit (2-3 hours)
Step 1: On-page audit
Run seo-audit on your target page:
Audit this page for on-page SEO.
## Page Content
[Paste markdown export here]
## Page Metadata
- Title tag: [exact title]
- Meta description: [exact description]
- URL: [full URL]
- Target keyword: [primary keyword]
## Current Performance
- Monthly organic visits: [number or estimate]
- Average position for target keyword: [if known]
- Search intent for target keyword: [informational/transactional/etc.]
Analyze on-page elements, keyword usage, content quality, and search intent alignment. Prioritize fixes by expected impact.Step 2: Technical check
Run seo-technical on your site:
Review the technical SEO setup for my site.
## Site Details
- Platform: [WordPress/Shopify/custom/etc.]
- URL: [homepage]
- Known issues: [anything you're aware of]
Check for:
1. Schema markup presence and correctness
2. XML sitemap status
3. Canonical tag implementation
4. Mobile rendering
5. Any obvious crawl/index issues
What schema types should I add to my [page type] pages? Provide JSON-LD templates.Step 3: Competitive comparison
Run competitor-seo against the top-ranking pages for your target keyword:
Compare my page against these competitors for the keyword "[keyword]":
## My Page
[Paste content or summary]
## Competitor 1: [URL]
[Paste content or summary]
## Competitor 2: [URL]
[Paste content or summary]
Identify:
1. Content gaps (what they cover that I don't)
2. Structural differences (how they organize content)
3. Keyword opportunities they're capturing
4. What I can do better than them
Document all findings before moving to Phase 3.
Phase 3: AI Search Audit (2-3 hours)
Step 1: Test your content in AI models
This is the most important step. Open ChatGPT, Perplexity, or Claude with web search and ask questions your content should answer.
Test 3-5 queries. For each one, document:
QUERY: [What you asked]
MODEL: [ChatGPT/Perplexity/Claude]
MY CONTENT CITED: [Yes/No]
SOURCES CITED INSTEAD: [List URLs]
WHAT CITED SOURCES HAVE THAT MINE DOESN'T:
- [Observation 1]
- [Observation 2]
- [Observation 3]This gives you concrete evidence of what the AI models prefer over your content.
Step 2: CITE framework audit
Run geo-optimization on your content:
Audit this content for AI search citability using the CITE framework.
## Content
[Paste markdown export]
## AI Search Test Results
[Paste your findings from Step 1]
## Target Queries
[List the queries your content should answer]
Score each CITE dimension:
- Citable structure: Can individual claims be extracted?
- Information density: Does every paragraph add original value?
- Topical authority: What signals authority in this domain?
- Entity clarity: Are entities clearly identified and consistent?
For each dimension, give specific recommendations to improve.Step 3: Analyze what's beating you
Take 2-3 sources that AI models cited instead of yours. Export them to markdown. Note:
How do they structure content differently?
What specific information do they include that you don't?
How do they handle the exact query the model answered?
What can you learn from their approach without copying it?
Phase 4: Cross-Channel Analysis (1 hour)
Now you have SEO findings and GEO findings. Look for overlaps.
Here are my SEO audit findings:
[Summary of Phase 2]
Here are my GEO audit findings:
[Summary of Phase 3]
Identify:
1. Fixes that improve BOTH Google ranking and AI citability
2. Quick wins (high impact, easy to implement)
3. Structural improvements that require more work
4. Which fixes to prioritize for maximum dual-channel impactFixes that help both channels go to the top of the priority list. These are the highest-leverage changes you can make.
Phase 5: Prioritize and Implement (2-3 hours)
Score each fix with the ICE framework:
Fix | Channel | Impact | Confidence | Ease | Score |
Add specific data to key claims | Both | High | High | High | 9 |
Restructure headers for extractability | Both | High | Medium | High | 8 |
Add FAQ schema markup | SEO | Medium | High | High | 7 |
Build 3 supporting articles for topical authority | Both | High | Medium | Low | 6 |
Score: High=3, Medium=2, Low=1. Add all three. Prioritize "Both" channel fixes.
Implement top 3-5 fixes. For each:
FIX TRACKING
Fix: _______________
Channel: SEO / GEO / Both
Implemented: [date]
BEFORE:
- SEO: Current ranking position / traffic
- GEO: Cited in AI search? [Yes/No, with test queries]
AFTER (check at intervals):
- SEO check-in: 2-4 weeks (rankings shift slowly)
- GEO check-in: 1-2 weeks (re-run AI search queries)
- Result: _______________The Output
By the end of this project, you have:
SEO audit with on-page, technical, and competitive findings
GEO audit with CITE framework scores and specific gaps
AI search test results showing what gets cited (and what doesn't)
Cross-channel analysis showing where fixes compound
Prioritized fix list with ICE scoring
3-5 implemented changes with tracking plan
Save everything in your Claude Project. The baseline AI search test results are particularly valuable. Re-run the same queries monthly to track whether your citability improves.
Do you want ALL the Skills, ready to deploy?
You got some of the skills above. If you’re a Cortex subscriber, you’ll get them ALL below.
You can either spend 30-60 minutes trying to create your own, or you can snag them all in ~30 seconds as a paying Cortex member—plus get the true value of Cortex, far beyond a bunch of skills.
What is Cortex?
Cortex is the premium monthly membership version of Bionic Business.
In addition to 3-4 full, unrestricted regular issues per month, you also get:
1x SIGNALS Strategic Briefing issue (strategic moves you must make now to secure your business revenue, market share, and profits).
1x CIRCUITS Tactical Guide issue (workflows, how-to’s, prompts, very practical implementation inside a self-driving business).
50% OFF workshops on prompt engineering, agent operating systems, and more. Only active Cortex subscribers get any kind of discount, no one else.
Free subscribers get a redacted and summarized version of regular issues (only Cortex subscribers get the full emails loaded with the good stuff).
There’s enough value for you to take action but it’s not the full stack. And some issues are in full, with no restrictions.
So, if you’d like to receive full issues and the monthly special Signals and Circuits issues, then you should sign up for Cortex, which is $50/month (and you have full archives of previous regular issues for as long as you’re a member).
One note on Signals and Circuits: You only get these special issues for the months you’re subscribed—which means if you skip a month, you won’t get them for that month. There are no back-issues for past months, so now is your chance to lock-in these upcoming special issues while you can.
Also, almost all issues on the website are only accessible with a Cortex subscription.
After you’ve upgraded to Cortex, log in to the website and come back to this issue, and the Skills will be unlocked below.
Ready? Get the Skills, full regular issues, and the upcoming Signals and Circuits issues:
Cortex is a monthly premium version of Bionic Business, helping entrepreneurs grow and scale with AI and Agents. You can cancel at any time. But there are no refunds, for any reason. All sales are final.
Join Cortex to read the rest!
Become a paying subscriber to get access to this post and other subscriber-only issues. Cortex is a monthly premium version of Bionic Business, helping entrepreneurs and business owners grow and scale with AI and Agents—and build autonomous businesses.
Upgrade to Cortex NowEvery month, you get:
- Full regular email issues
- 1x Signals Strategic Briefing
- 1x Circuits Tactical Guide
