aeo

---
name: aeo
description: Use when planning content for AI-first audiences, auditing E-E-A-T signals, tracking LLM citations, or building a citation strategy for ChatGPT, Perplexity, Claude, Gemini, or Mistral.
category: Marketing
version: 1.0.0
tools: []
---

# Answer Engine Optimization (AEO)

Help content get cited by ChatGPT, Perplexity, Claude, Gemini, and Mistral as an authoritative source. AEO optimizes for **citation** in LLM-generated answers — distinct from SEO, which optimizes for search rankings.

## AEO vs. SEO
| | SEO | AEO |
|---|---|---|
| Optimizes for | Click-through rankings | Being cited as the authoritative source |
| Audience | Humans browsing search results | LLMs answering questions |
| Success metric | Position 1-10, organic traffic | Citation count across LLMs |
| Key signals | Backlinks, keywords, page speed | E-E-A-T, structured data, factual density |
| Update cadence | Weeks-to-months | Days-to-weeks (LLM training cycles) |

Both can coexist on the same content, but the techniques differ: SEO rewards keyword density + backlinks; AEO rewards primary-source signals + structured facts.

## When to Use
- Planning content for an AI-first audience
- Auditing existing content for E-E-A-T gaps before an AI Overview rollout
- Tracking which pages get cited by which LLM
- Researching what queries LLMs cite sources for vs. answer from training
- Benchmarking against competitors' citation rates
- Building a long-term AEO strategy aligned with SEO

Not a fit for: pure click-through SEO with no citation intent; brand-voice content with no factual claims (nothing to cite); topics where LLMs already have strong training signal (e.g. elementary math); or breaking news (LLM training lag means citations arrive months later).

## Pre-flight: Bot Access
Before anything else, check whether AI crawlers are blocked in robots.txt — GPTBot, PerplexityBot, ClaudeBot, Google-Extended. A blocked crawler zeroes out citations from that platform and is always the first fix.

## Content Audit: E-E-A-T Scoring
Score content 0-100 across four dimensions:
- **Experience**: first-person evidence, dated examples, case studies, "we ran X in [year]" claims
- **Expertise**: author bio, credentials, citations to peer-reviewed sources, technical depth
- **Authoritativeness**: external backlinks from authority domains, schema.org markup, structured data
- **Trustworthiness**: HTTPS, contact info, transparent corrections policy, factual density (verifiable claims per 1,000 words)

Composite score = average of the four (0-100), with a per-dimension breakdown and specific fix recommendations for each gap.

Industry raises the bar — apply the stricter threshold and critical-signal checks for the relevant industry:

| Industry | Min Composite | Critical Signals |
|---|---|---|
| Healthcare | 85 | Medical reviewer byline, peer-reviewed citations, FDA disclosure |
| Finance | 85 | Author CFA/CPA credentials, "not investment advice" disclaimer, dated examples |
| Legal | 85 | Jurisdiction disclosed, attorney bio, "not legal advice" disclaimer |
| SaaS | 70 | Product manager byline, case study with metrics, ROI calculator |
| E-commerce | 65 | Aggregated product reviews, return policy, schema.org Product |
| B2B | 70 | Industry analyst quotes, customer logos, ROI data |
| Media | 70 | Editorial policy, fact-check link, original reporting |
| Education | 75 | Instructor bio, learning outcomes, accreditation if applicable |

## Content Optimization
Recommend AEO improvements at one of three intensities: **conservative** (touch <10% of the words), **balanced** (<30%), or **aggressive** (rewrite for maximum AEO). Improvements typically include: H2/H3 structure rewritten for LLM parsing, added citation density (numbered references to sources), JSON-LD schema for FAQ/HowTo/Article markup, and a fact-first lede that moves verifiable claims into the first 200 words.

## Citation Tracking
Since there's no persistent storage here, maintain a citation ledger directly in the conversation as a table: URL, LLM, query, date. When the user reports a citation they found (e.g., pasted from ChatGPT or Perplexity output), add it to the running table and compute per-page citation count, citation velocity, and LLM coverage on request. Offer to format it for export (e.g., as CSV-ready text) at the end of a session.

## Report Format
Present findings as a markdown audit: title, URL, date, industry, composite score with letter grade, the four-dimension breakdown table, a "Top 3 Fixes" list ranked by point impact, and the full recommendation list. Offer a JSON-shaped version if the user wants it for pipeline integration.

## Anti-Patterns to Reject
- **Keyword stuffing for AI** — LLMs extract topic from semantics; keyword density doesn't boost citation odds
- **Pure AI-generated content with no human review** — generic output gets de-prioritized by RAG retrieval looking for distinctive signal
- **Citation farms / link wheels** — modern RAG penalizes low-authority linked networks
- **Schema spam** — false or unverifiable schema.org claims get filtered; only mark up real, verifiable claims
- **Optimizing for one LLM at the expense of others** — citation distributions correlate highly across major LLMs since they share training-data sources; optimize for the shared E-E-A-T signals, not per-LLM hacks
- **Ignoring SEO entirely** — AEO citations often originate from sources that already rank well organically; treat AEO and SEO as complements

If the task is really about click-through SEO, programmatic SEO at scale, broader content strategy, copy voice/tone, or structured-data implementation specifically, @mention the teammate who owns that specialty.