AEO Content Checklist

TL;DR

Run a page through this 30-item, five-pillar checklist before publishing or as part of a quarterly audit. Score one point per box. 25-30 is publish-ready, 20-24 needs polish, below 20 needs structural work. The five pillars map directly to the Citation Readiness Score used across the rest of the /aeo/ hub.

How to use this checklist

Work through the five pillars in order: Answerability, Authority, Freshness, Structure, and Entity Clarity. Each pillar has six checkboxes. Score one point per box that passes; total out of 30. Run it once before publishing, then on a 90-day audit cadence for any page in the citation-priority set.

The pillar grouping is deliberate. AI engines select citation candidates on a small, repeatable set of signals: can the page answer the question, can it be trusted, is it current, can it be parsed, and does it talk about entities consistently. Each pillar covers one of those signals, and the six items in each pillar are ordered roughly from highest to lowest leverage so that fixing top items first yields the largest extraction improvement.

Phase 1: Answerability (6 items)

• [ ] Direct answer in the first sentence — The very first sentence below the H1 (or below an AI-summary blockquote) answers the canonical question. Verify by skim-reading only the first sentence and asking "does this answer the question?"
• [ ] Answer block sits in the 40-60 word range — Definitions and core answers are tight enough to be quoted whole. Verify with a word-count tool on the answer paragraph; 40-60 is the Frase-recommended range for AI search FAQs.
• [ ] Answer is self-contained — The answer makes sense without surrounding context, anaphora ("this," "that," "as above"), or references to images. Verify by copy-pasting the answer block into a blank document and reading it cold.
• [ ] Canonical question matches H1 intent — The frontmatter canonical_question and the H1 target the same intent and entity. Verify by reading H1 and canonical_question side by side.
• [ ] No filler preamble — The page does not begin with "In this article…", "We will explore…", or other AI-extraction-killing scaffolding. Verify with a Find for "In this article" and "we will explore."
• [ ] AI summary blockquote present — A single > AI summary: … blockquote sits between the H1 and the TL;DR, with two to three factual sentences. Verify the page has exactly one such blockquote, not two or three competing summaries.

Phase 2: Authority (6 items)

• [ ] Author identified — A real, named author is shown above the fold and in author schema. Anonymous "team" bylines are weaker; a named author with credentials is stronger. Verify with View Source and the Rich Results Test.
• [ ] Author has linked credentials or bio — Author has either a Person schema with hasCredential or a linked author page with bio and qualifications. Verify by clicking the byline and inspecting the resulting page.
• [ ] At least one primary or industry-standard citation per long-form page — Specific claims are anchored to identifiable named sources (official docs, schema.org, peer-reviewed papers, established industry sources). Verify by Finding ](http and auditing the link list.
• [ ] No unsourced numeric multipliers — No "3x more", "50% faster", or "10x improvement" without a footnote or inline citation. Verify with a regex Find for \d+x and \d+% and confirm each match has a source.
• [ ] Domain or organization signals present — Page is on a topically focused domain or in a clearly authored section, with Organization schema set on the site. Verify with View Source and the Rich Results Test.
• [ ] No hallucination-prone claims — Claims are either widely-known facts, sourced, or framed as opinion / pattern observation. Verify by re-reading every numeric, comparative, or "X is the leader" claim and asking "where is the source?"

Phase 3: Freshness (6 items)

• [ ] datePublished present in schema — The page exposes a published date both visibly near the byline and inside Article / TechArticle JSON-LD. Verify with View Source for datePublished.
• [ ] dateModified present and accurate — The schema dateModified matches the visible "Last updated" text. Stale or missing dateModified is a strong negative AEO signal. Verify with the Rich Results Test.
• [ ] Visible "Last updated" date near the byline — A reader can see the freshness without inspecting source. Verify by skim-reading the header area of the published page.
• [ ] review_cycle_days set in frontmatter — Typical value is 90 days for evergreen and 30 days for fast-moving topics. Verify by opening the MDX frontmatter.
• [ ] last_reviewed_at updated on every audit — Even when the body does not change, a successful review bumps last_reviewed_at. Verify against your audit log or git history.
• [ ] Time-sensitive claims flagged — Any claim tied to a specific year, version, or release ("as of 2026", "Chrome 124+") is explicitly dated in the body. Verify with a Find for the current year and major prior years.

Phase 4: Structure (6 items)

• [ ] Logical heading hierarchy — H1 → H2 → H3 with no skipped levels and no orphaned H4s. Verify with a heading-outline browser extension or a manual scan.
• [ ] Tables for comparisons, lists for steps and features — Comparisons are tables, sequential instructions are ordered lists, and unordered features are bulleted lists, not prose. Verify by re-reading body sections and asking "could this paragraph be a table?"
• [ ] Scannable paragraphs — Primary content paragraphs cap at roughly three sentences. Long blocks are broken into sub-sections or lists. Verify by skim-reading the rendered page.
• [ ] FAQ section at the bottom — A real FAQ with 4-8 question/answer pairs, each answer 2-4 sentences. Verify by counting Q&A pairs and answer length.
• [ ] FAQPage schema present where the FAQ exists — Even though Google's August 2023 change restricted the rich result to gov/health, the schema continues to act as an AI-extraction signal. Verify with the schema.org Validator.
• [ ] BreadcrumbList schema reflecting site hierarchy — Page exposes its place in the hub → pillar → article structure. Verify with View Source for BreadcrumbList.

Phase 5: Entity Clarity (6 items)

• [ ] One canonical name per entity across body, frontmatter, and schema — "AEO" is "AEO" everywhere, not also "Answer Engine Optimisation" mid-paragraph. Verify with a global Find for known entity names and their variants.
• [ ] Aliases declared in frontmatter — Any entity with more than one common name has its variants in aliases:. Verify by reading the frontmatter aliases block.
• [ ] Acronyms expanded on first use — First use is "answer engine optimization (AEO)"; subsequent uses are just "AEO." Verify by Finding the acronym and confirming the first hit is the expansion.
• [ ] Internal links use canonical anchor text — Links to related concept pages use the canonical name as the anchor text, not "click here" or "this article." Verify by reading the body link text aloud.
• [ ] External entities link to canonical pages on first mention — First mention of schema.org, Google Search Central, OpenAI, Perplexity, etc., links to the canonical home or doc. Verify by scanning the link list.
• [ ] canonical_concept_id present in frontmatter — Page declares its stable, kebab-case canonical ID. Verify by opening the frontmatter.

Scoring rubric mapped to Citation Readiness Score

The 30-item score maps directly to the Citation Readiness Score (CRS) used across this hub. Each pillar contributes one CRS axis with a fixed weight:

To compute CRS from the checklist: score each pillar out of 6, divide by 6 to get a per-pillar ratio (0-1), multiply by the weight, and sum. A page scoring 6/6 Answerability, 5/6 Authority, 4/6 Freshness, 6/6 Structure, and 5/6 Entity Clarity computes to (1.00 × 25) + (0.83 × 20) + (0.67 × 15) + (1.00 × 25) + (0.83 × 15) = 89, which lands in the "publish, expect citation" band. CRS ≥ 85 corresponds to a checklist score of 25/30 or higher and is the operational threshold for citation-priority pages.

Quick-fix priorities by severity band

When a page lands below the publish threshold, the order in which you fix items matters more than the count. Use the following priority order, mapped to the Severity bands derived from CRS.

Critical (score below 15). Page is not publish-ready and not auditable in its current form.

1. Add a direct first-sentence answer (Answerability item 1).
2. Add an AI summary blockquote (Answerability item 6).
3. Add datePublished and dateModified (Freshness items 1 and 2).
4. Add Article or TechArticle schema as the structural foundation.
5. Re-outline the body to a logical heading hierarchy (Structure item 1).

High (score 15-19). Page has bones but the AI-extraction layer is missing.

1. Tighten the answer block to 40-60 words (Answerability item 2).
2. Anchor unsourced claims; remove unsourced multipliers (Authority items 3 and 4).
3. Add the FAQ section and FAQPage schema (Structure items 4 and 5).
4. Add canonical_concept_id and aliases to frontmatter (Entity Clarity items 6 and 2).

Medium (score 20-24). Page is close to publish-ready; finish the polish.

1. Standardise entity names and acronym handling (Entity Clarity items 1 and 3).
2. Add BreadcrumbList schema (Structure item 6).
3. Set review_cycle_days and refresh last_reviewed_at (Freshness items 4 and 5).
4. Convert any prose comparisons to tables (Structure item 2).

Low (score 25-27). Publish; queue the remaining items for the next audit cycle and avoid blocking release on cosmetic gaps.

Copy-paste markdown template

Drop the following block into a Notion or Linear ticket per page audit. Replace the slug placeholder, then check off items as you go. The block deliberately mirrors the structure of this checklist so audit notes drop straight back into the page metadata when the fixes are made.

# AEO Audit — <slug>

Date: YYYY-MM-DD

Reviewer:

Score target: 25 / 30

Phase 1: Answerability (___ / 6)

• [ ] Direct answer in first sentence
• [ ] Answer block 40-60 words
• [ ] Answer self-contained
• [ ] canonical_question matches H1 intent
• [ ] No filler preamble
• [ ] AI summary blockquote present

Phase 2: Authority (___ / 6)

• [ ] Author identified
• [ ] Author credentials linked
• [ ] >= 1 primary citation per long-form page
• [ ] No unsourced numeric multipliers
• [ ] Org / domain signals present
• [ ] No hallucination-prone claims

Phase 3: Freshness (___ / 6)

• [ ] datePublished in schema
• [ ] dateModified in schema (matches visible date)
• [ ] Visible "Last updated" date
• [ ] review_cycle_days set
• [ ] last_reviewed_at current
• [ ] Time-sensitive claims dated

Phase 4: Structure (___ / 6)

• [ ] Logical heading hierarchy
• [ ] Tables for comparisons, lists for steps
• [ ] Scannable paragraphs
• [ ] FAQ section (4-8 Q&A)
• [ ] FAQPage schema
• [ ] BreadcrumbList schema

Phase 5: Entity Clarity (___ / 6)

• [ ] One canonical name per entity
• [ ] Aliases declared in frontmatter
• [ ] Acronyms expanded on first use
• [ ] Internal links use canonical anchor text
• [ ] External entities linked on first mention
• [ ] canonical_concept_id present

Total: ___ / 30

Band:

Top 3 fixes:

1.

2.

3.

FAQ

Q: Do I need every checklist item to publish?

No. Aim for 25 / 30 or higher before publishing for citation-priority pages, and 20 / 30 for lower-priority utility pages. Below 20 / 30, AI engines are unlikely to cite the page reliably and the highest-leverage move is structural rewriting rather than item-by-item polish.

Q: Why are there exactly five pillars and six items per pillar?

The five pillars correspond to the five signals AI engines use when selecting citation candidates: extractability, trust, recency, parseability, and disambiguation. Six items per pillar is the empirical count where each pillar covers its signal without redundancy and the total (30) is small enough to run in a single review pass within a 30-minute audit slot.

Q: Is FAQPage schema still required after Google's 2023 change?

For non-gov/health sites it is optional for rich results but still recommended for AEO. The August 2023 Google update removed the visual rich result for general sites, but FAQPage schema continues to act as an AI-extraction signal across ChatGPT, Perplexity, and Google AI Overviews (Frase: FAQ schemas for AI search; Search Engine Land: schema and AI search).

Q: How often should I re-run this checklist?

At least every 90 days for evergreen pages, every 30 days for fast-moving topics (model releases, platform updates), and after every major refactor. The frontmatter review_cycle_days and last_reviewed_at fields exist to track the cadence per page rather than running a single global schedule.

Q: What is the highest-leverage single fix on a failing page?

Moving the direct answer into the first sentence of the answer block (Answerability item 1). Most pages already contain the answer somewhere; promoting it to position one is usually a one-line edit that meaningfully improves AI extraction across all engines and unlocks several downstream items in the Answerability pillar.