/009-substack-aeo-geo-optimizer

name: substack-aeo-geo-optimizer
description: >
  Optimizes any Substack content for AEO (Answer Engine Optimization) and GEO
  (Generative Engine Optimization) — making it citable by ChatGPT, Perplexity,
  Claude, Gemini, and Google AI Overviews. Use this skill whenever the user
  provides a Substack URL or pastes Substack post content and wants to increase
  AI visibility, get cited by AI search engines, optimize for generative search,
  or asks things like "optimize this for AI search", "make this citable by
  ChatGPT or Perplexity", "GEO optimize this post", "AEO this article",
  "how do I get AI to cite my newsletter", "optimize for AI overviews",
  "make this show up in AI answers". Always trigger this skill when the goal
  is to make a Substack post more visible inside AI-generated answers, not just
  Google rankings. This is separate from SEO metadata optimization.

Substack AEO/GEO Optimizer

Transforms a Substack post (URL or pasted text) into content that AI engines (ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews) are structurally likely to cite in their answers.

Built on the CITED framework from Passionfruit's AEO methodology and Princeton KDD 2024 GEO research showing 30-40% citation lift from structured optimization.

---

Core insight: why most Substack content never gets cited by AI

AI engines don't cite content at random. They cite content that: 1. Answers a specific question **directly and immediately** (in the first 40-60 words) 2. Contains **verifiable facts** with named sources (not vague claims) 3. Uses **structured, extractable formats** (FAQ, lists, definition blocks) 4. Signals **authority** through author credentials and entity clarity 5. Covers the topic **comprehensively enough** to be the best available answer

Most Substack posts are written for human readers who tolerate narrative buildup. AI engines are pattern-matching for extraction — they need the answer upfront, the evidence inline, and the structure explicit.

---

Step 1 — Read and classify the content

If a URL is provided, fetch and read the full post content. If text is pasted, read it directly.

Classify the post into one of these query types (this determines optimization priority):

| Query type | Signal | Primary AI citation format | |---|---|---| | **Definitional** | "What is X", "X explained" | Definition block (40-60 words), then depth | | **How-to / Process** | "How to X", numbered steps | Numbered list with clear step verbs | | **Comparison** | "X vs Y", "best X for Y" | Comparison table, named criteria | | **Opinion / Contrarian** | "Why X is wrong", "The truth about X" | Named expert quotes, cited data | | **List / Roundup** | "Best X", "Top Y tools" | Structured list with consistent criteria | | **Case study / Story** | "How I did X", "My results with Y" | Result-first opening, metrics inline |

---

Step 2 — Run the CITED audit

Score the existing post on each of the 5 CITED dimensions. Mark each as ✅ Present / ⚠️ Weak / ❌ Missing.

C — Context clarity

Does the post make clear **who it's for** and **what specific question it answers**?

• ✅ Present: clear audience, specific question answered in title/intro
• ⚠️ Weak: topic is clear but audience or query angle is vague
• ❌ Missing: post reads as general exploration with no specific answerable question

I — Intent match

Does the content structure match the **intent** of someone asking this question to an AI?

• ✅ Present: informational → direct answer; comparison → table; how-to → steps
• ⚠️ Weak: intent is right but format doesn't match (e.g., steps buried in paragraphs)
• ❌ Missing: content doesn't answer the query an AI user would actually ask

T — Truth signals

Does the post contain **verifiable, citable evidence**? Count: named statistics with source, expert quotes with attribution, specific data points.

• ✅ Present: at least 1 cited stat or data point every 150-200 words
• ⚠️ Weak: some stats present but no sources named, or fewer than 1 per 300 words
• ❌ Missing: all claims are opinion-based, no named sources

E — Extraction format

Is the content **structurally easy for an AI to lift and cite**?

• ✅ Present: direct-answer intro in 40-60 words, FAQ section, or structured list
• ⚠️ Weak: good content but buried in narrative without extractable structure
• ❌ Missing: wall of text, no headers, no extractable blocks

D — Differentiation

Does the post contain **at least one thing no other source says** in this way? Original framework, proprietary data, personal result, counter-intuitive claim.

• ✅ Present: clear unique angle, named framework, or original data
• ⚠️ Weak: good writing but no unique insight AI couldn't find elsewhere
• ❌ Missing: standard take on a standard topic

---

Step 3 — Produce the AEO/GEO optimization report

Output a structured report with these sections:

Section 1: CITED Audit Score

C — Context clarity:    [✅/⚠️/❌] [1-line explanation]
I — Intent match:       [✅/⚠️/❌] [1-line explanation]
T — Truth signals:      [✅/⚠️/❌] [1-line explanation]
E — Extraction format:  [✅/⚠️/❌] [1-line explanation]
D — Differentiation:    [✅/⚠️/❌] [1-line explanation]

AI Citation Readiness: [LOW / MEDIUM / HIGH]

Section 2: Top 3 AI queries this post could answer

List the 3 most likely questions a user would ask ChatGPT or Perplexity that this post *should* show up in — but probably doesn't yet.

Format:

1. "[Exact query a user would type]"
   → Currently blocked by: [what's missing]

2. "[Exact query a user would type]"
   → Currently blocked by: [what's missing]

3. "[Exact query a user would type]"
   → Currently blocked by: [what's missing]

Section 3: Rewritten opening (the AEO intro block)

The single highest-impact fix. Rewrite the post's opening paragraph as an **AEO-optimized direct-answer intro** of 40-60 words.

Rules for this block:

• Answers the primary query directly in sentence 1
• Includes the most searchable term naturally
• Contains at