Answer-Engine Optimization Brief

---
name: answer-engine-optimization-brief
description: Use when briefing content for LLM answer engines (ChatGPT, Perplexity, Claude, Google AI Overviews). Triggers on "AEO", "LLM ranking", "answer engine", "AI Overviews", or "Perplexity citation".
---

# Answer-Engine Optimization Brief

Classical SEO optimizes for "make the user click." AEO optimizes for "make the LLM quote you." Different mechanics, different content shape.

## Required inputs

1. **Target query** — the exact wording (or close paraphrase) of the question the LLM is being asked
2. **Current SERP / LLM answer** — what's already being cited, by whom
3. **Your authority signals** — what makes your source quotable (data, original research, named author, brand recognition)
4. **Your business goal** — citation alone, citation that drives click, citation that drives mention by name

## What gets cited

Observed across ChatGPT, Perplexity, Google AI Overviews:
- **Extractable chunks** — short, declarative sentences that can be quoted standalone
- **Numbers + dates** — specific stats, recent (≤18 months) preferred
- **Clear attribution** — author name, publication, date in the page metadata
- **Structured answers** — H2/H3 questions, lists, comparison tables
- **First-party data** — surveys, internal benchmarks, original research
- **Direct topical match** — page is unambiguously about the query

What doesn't get cited:
- Walls of prose that need summarizing to extract
- Pages full of caveats ("it depends," "many factors")
- Conversational fluff between facts
- Anything that requires reading 3 paragraphs to get to the claim

## Brief structure

### Page-level
- **URL slug** — match the natural-language query as closely as possible
- **Title** — the query as a complete question or declarative answer
- **Meta description** — the 1-sentence direct answer, in case it gets pulled
- **Schema.org** — Article + FAQPage + Author + DatePublished + DateModified

### Content structure
- **First sentence**: the direct answer to the query. Not an intro. Not a setup. The answer.
- **Followed by**: the 1-paragraph expansion with the qualifications
- **Then**: H2-structured sections that each answer a sub-question the original query implies
- **Tables**: where comparison is involved (LLMs love structured data they can quote)
- **Lists**: where enumeration matters (3-step processes, criteria sets)

### Extractable atoms
For the brief, identify 5–8 standalone declarative sentences that, if quoted, would be the LLM's answer. These should:
- Be ≤30 words
- Stand alone without needing context
- Contain a number or named entity if possible
- Include a citation-ready source (your data, named study, recent date)

Example: "In 2025, 64% of B2B buyers under 30 said they ask an AI tool before contacting a vendor (source: [data], 2025-04)."

### Authority signals
- Named author with credentials visible
- Last-updated date visible (LLMs prefer recent)
- First-party data or original research
- Internal links to related pages (signals topical depth)
- External links to authoritative sources (signals you've done research)

## Output

```
## Brief
- Target query: ...
- Direct 1-sentence answer (this is what we want LLMs to quote): ...

## Page structure
- Slug: ...
- Title: ...
- Meta: ...
- H2 sections: [list]

## 5 extractable atoms
[Each ≤30 words, each citation-ready]

## Authority signals to include
[Author bio, date, original data, etc.]

## Comparison table content
[If applicable — what columns, what rows]

## What to NOT do
- No "it depends"
- No 3-paragraph intro
- No conversational throat-clearing
- No vague qualifiers that block extraction
```

## Tone

- Direct. The LLM is looking for a quote, not a vibe.
- Numerical when possible. "Most companies" → "73% of companies (2025 survey of 1,200)."
- Recent. Date everything.
- Confident. Hedged language doesn't get quoted.