What does an AEO score actually measure?

The score measures two things: extractability (schema, atomic answer blocks, heading hierarchy, retrieval-ready chunking) and citability (original claims, proprietary data, demonstrable expertise). A 95+ score means the page clears a publishable hygiene-and-substance floor. It is a floor, not a forecast — answer engines are not deterministic, and no score guarantees citation.

How is the AEO Standard different from an SEO audit?

SEO audits optimize for a ranked results page; the AEO Standard optimizes for extraction and citation by AI answer engines. The overlap is real — semantic HTML, schema, and meta hygiene serve both — but the standard adds what SEO tooling ignores: substance scoring, atomic answer blocks, RAG chunk quality, and query fan-out coverage across a content cluster.

Can a page score well and still never get cited?

Yes. Structure makes content extractable; substance makes it citable. A page of perfectly formed answer blocks that says nothing a model could not generate itself earns no citations — which is why Substance & Originality carries 25 of the 100 points and cannot be scored by regex alone. The standard says this plainly rather than selling certainty.

How do I check a page against the standard?

Two ways today: the free web scanner at isimplifyme.com/tools/aeo-scanner scores any URL in the browser, and the aeo-scan CLI preview on npm checks the core mechanical signals from the terminal with npx aeo-scan. The full CLI — sitemap crawls, CI exit codes, optional LLM-scored substance checks — is in development against this standard.

Why publish the rubric openly?

Scores only mean something when the rubric behind them is stable, inspectable, and versioned. Publishing under CC BY 4.0 lets anyone adopt or adapt the methodology with attribution, and versioning it like software means a score always references a specific revision. The weights are experiential — refined across a seventeen-site production network — and the standard says so.

The AEO Standard — the 100-Point AEO Rubric by iSimplifyMe

The 100-point Answer Engine Optimization rubric — published, versioned, open for adoption.

Abstract

The AEO Standard is a 100-point scoring system for Answer Engine Optimization — the discipline of structuring content so AI answer engines like ChatGPT, Perplexity, Gemini, and Claude can extract, cite, and recommend it. iSimplifyMe developed the rubric internally, ran every page in a seventeen-site production network through it, and now publishes it as a versioned, openly licensed standard. This page is the standard.

Why publish a scoring standard?

Answer engines are replacing the search results page as the place where buying decisions form, yet most "AEO tools" measure structure alone — schema present, headings tidy — and call it optimization. That misses the half that matters: whether the content says anything a model would bother to cite.

A published standard fixes a second problem: consistency. Scores only mean something when the rubric behind them is stable, inspectable, and versioned. Ours is now all three.

The principle

AI engines cite content that is genuinely worth citing. Structure makes content extractable; substance makes it citable. A page can be perfectly formatted and still earn zero citations because it says nothing a model could not generate itself — which is why this standard scores both, and why a structural score alone is a floor, not a forecast.

Gating rules — automatic rejection

Before any points are scored, two gates apply. Failing either one is an automatic REJECT regardless of the point total:

Hidden machine-only content. Text placed in the page for crawlers but hidden from human readers — screen-reader-only blocks of body copy, clip: rect(0,0,0,0) or 1×1-pixel containers, off-screen or white-on-white text. Hidden text is a spam pattern, it does not work (engines render the page and discount it), and on health, legal, and financial sites it is a manual-action risk.
Fabricated authority. Invented credentials, invented statistics, or citations to sources that do not exist.

The gates exist because both failures are common, both are tempting, and both destroy the trust the rest of the rubric is trying to build. Fix the gating violation first; then score.

The 100 points

The rubric scores seven sections. Sections 1 and parts of 4 require judgment — a human reviewer or an LLM pass. The rest is mechanical and can be scored by software. Checks are marked accordingly: [M] mechanical, [J] judgment.

Section 1 — Substance & Originality (25 points)

Check	Points
Makes a claim, take, or synthesis not already in the generic top-10 results [J]	5
First-hand experience, proprietary data, original analysis, or named-expert insight [J]	5
Specific named entities, numbers, and examples — not generic advice [J]	5
Would NOT be reproduced by a one-line generic LLM prompt [J]	5
Demonstrable expertise: credentialed expert named in-content, organizational authority markers, or proprietary data [J]	5

Section 1 carries the most weight deliberately. It is the section that commodity content fails, and it is the reason a page full of well-formed answer blocks can still earn nothing.

Section 2 — Atomic Answer Blocks (15 points)

Check	Points
Blocks are visible on the page — never hidden or screen-reader-only [M]	3
Answer blocks are 40–60 words and self-contained [M]	3
The primary question is answered in the first 40 words of its section [M]	3
3–5 atomic blocks minimum, each targeting a distinct question [M]	3
Definitive language — "X is…" not "X might be…" [J]	3

An atomic answer block is a direct, factual, 40–60-word answer to a specific question, preceded by a question-phrased heading, that makes complete sense when extracted in isolation. No "as mentioned above," no dependence on surrounding copy. The block near the top of this page is one.

Section 3 — Structured Data & Schema (10 points)

Check	Points
FAQPage schema with 3+ question–answer pairs [M]	2
Article or BlogPosting schema with all required fields [M]	2
BreadcrumbList schema [M]	2
HowTo, Person, or Organization schema where applicable [M]	2
Schema validates with zero errors [M]	2

BreadcrumbList is the most commonly missing schema across every site we have audited — including, at one point, our own homepage. Note that JSON-LD delivered inside a @graph container is valid and common; a scanner that fails to descend into @graph will report schema as missing when it is not.

Section 4 — RAG / Retrieval Readiness (20 points)

Check	Points
Key facts front-loaded — the answer appears in the first 1–2 sentences of each section [J]	3
Short paragraphs, 2–4 sentences — clean chunk boundaries [M]	3
Content chunks on real structural boundaries — headings and paragraphs, no fake blocks [M]	3
Entity consistency — the same name for the same thing throughout [J]	3
Data-backed claims with specific numbers [J]	3
Source attribution for statistics and claims [M]	3
Definitive statements, no ambiguous pronouns [J]	2

Retrieval pipelines chunk content on structural boundaries and embed the chunks. Walls of text chunk badly, hedged language embeds weakly, and a chunk that reads "it monitors them" retrieves nothing.

Section 5 — Semantic HTML & Heading Hierarchy (10 points)

Check	Points
Single H1 matching the target query [M]	2
H2s map to subtopics; H3/H4 nested without skipping levels [M]	2
Headings phrased as questions where natural [M]	2
Table of contents with anchor links [M]	2
Clear, readable structure — no reliance on exotic markup [J]	2

Section 6 — Internal Linking & Fan-Out Coverage (10 points)

Check	Points
Links to the pillar or hub page with keyword anchor text [M]	2
3+ internal links per 1,500 words, with varied anchor text [M]	2
Links to related cluster content [M]	2
The page or its cluster covers the query fan-out — the adjacent sub-queries a model generates from the head query [J]	2
No orphan pages — new content is linked FROM existing pages [M]	2

Fan-out is the least-understood check in the rubric. When a model receives a question, it expands it into adjacent sub-queries before retrieving; a page that answers the head query in isolation, with no cluster covering the expansion, loses citations to sites that cover the neighborhood.

Section 7 — SEO Meta & Technical (10 points)

Check	Points
SEO title 50–60 characters [M]	2
Meta description 150–160 characters [M]	2
Clean URL slug — short, evergreen, no dates [M]	2
Image alt text on all images [M]	1
OG and Twitter card meta tags [M]	1
Canonical URL [M]	1
Page load under 3 seconds [M]	1

Score thresholds

The sections total exactly 100 points: 25 + 15 + 10 + 20 + 10 + 10 + 10. Verdicts map to ranges:

Score	Verdict
95–100	PUBLISH — the hygiene and substance floor is cleared
85–94	REVISE — minor fixes needed
70–84	REWRITE — structural or substance gaps
Below 70	REJECT
Any gating violation	REJECT, regardless of score

A 95+ score means the content is publishable. It does not guarantee citation or ranking — answer engines are not deterministic, and any tool or consultant promising otherwise is selling certainty that does not exist. The score is a floor, not a forecast.

How to check a page against the standard

There are two ways to run the rubric today. The free AEO scanner scores any URL in the browser. For the terminal, an early preview of the CLI is on npm — npx aeo-scan <url> checks the core mechanical signals, with the full 100-point CLI (sitemap crawls, CI exit codes, optional LLM-scored substance checks) in development.

The judgment checks — Section 1 above all — cannot be fully mechanized without a model in the loop, and that is by design. A standard that only measured what regex can see would reward exactly the commodity content the standard exists to filter out. For a scored audit with the judgment sections included, the full AEO audit runs the complete rubric, and the Nexus platform applies it continuously across a content pipeline.

Versioning and license

This is version 1.0 of the standard, published July 2026. Revisions are versioned like software: point releases for clarifications, minor versions for threshold changes, major versions for structural changes to the rubric. The changelog lives with the standard.

The methodology is licensed CC BY 4.0 — use it, adapt it, score against it, with attribution to iSimplifyMe. The weights encode several years of operating a seventeen-site production network against this rubric; they are experiential, not the output of a controlled study, and we say so plainly because a standard that overstates its own evidence would fail its own Section 1.

Status

The standard is published and versioned; this page is its canonical home.
The aeo-scan CLI preview is live on npm; the full CLI tracks adoption of the standard.
Internally, one unified scanner package now scores every site in the network against this rubric — the same checks, the same thresholds, in CI.
The white-paper PDF edition is available for download, and the versioned methodology lives in the public reference repository.

The AEO Standard

What is the AEO Standard?