AEO score methodology: every factor, every weight
The AEO score rates from 0 to 100 how ready a page is to be selected and quoted by answer engines. Here is exactly how each of the 22 factors is measured.
The formula
AEO score = points earned ÷ points available × 100, across 22 factors weighted 35% answer readiness, 20% structured data, 25% content quality and 20% trust and access. Bands: 85+ excellent, 70 to 84 good, 50 to 69 needs work, below 50 poor.
Answer readiness
| Factor | Weight | How it is measured |
|---|---|---|
| Direct answer in the opening paragraph | 8 | Opening paragraph is 20 to 90 words and does not start with "This", "We", "Welcome" or "Here". |
| Question-style headings | 7 | H2 to H6 headings that end with "?" or start with how/what/why/when/where/which/who/can/does/is/should. Full credit at 3. |
| FAQ section | 6 | A heading containing FAQ/questions, or FAQPage schema. |
| Snippet-length answer blocks | 6 | Paragraphs of 35 to 75 words that do not open with a pronoun. Full credit at 4. |
| Bulleted or numbered lists | 4 | Bulleted or numbered lists in the main content. Full credit at 2. |
| Comparison or data table | 2 | At least one HTML table in the main content. |
| Summary or key takeaways | 2 | A heading such as Summary, Key takeaways, TL;DR or Quick answer. |
Structured data
| Factor | Weight | How it is measured |
|---|---|---|
| Valid JSON-LD structured data | 6 | Every JSON-LD block parses and has a schema.org @type (invalid blocks earn 30%). |
| FAQPage, HowTo or QAPage schema | 6 | FAQPage, HowTo or QAPage present. |
| Organization or Person entity | 4 | Organization, Person, LocalBusiness or Corporation present. |
| Page-type schema | 4 | Article, BlogPosting, Product, Service, SoftwareApplication, Recipe, Course, Event or LocalBusiness present. |
Content quality
| Factor | Weight | How it is measured |
|---|---|---|
| Easy to read | 6 | English: Flesch Reading Ease 60+ = full, 50+ = 80%, 30+ = 45%. Other languages: average sentence length. |
| Factual density | 6 | Numbers, prices, dates and measurements per 100 words. Full credit at 3. |
| Content depth | 5 | Main-content words: under 300 = 20%, 600 = 50%, 1,000 = 75%, 1,000+ = 100%. |
| Clear section structure | 4 | At least 3 H2 sections (60%) plus a heading at least every 350 words (40%). |
| Covers the subtopics that matter | 4 | Share of the key subtopics (identified by AI from all compared pages) that your page covers. Excluded when AI is unavailable. |
Trust & access
| Factor | Weight | How it is measured |
|---|---|---|
| AI search bots allowed | 6 | robots.txt rules for this exact URL, per RFC 9309, for OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Bingbot. |
| Indexable and snippet-eligible | 3 | No noindex, nosnippet or max-snippet:0 in the robots meta tag or X-Robots-Tag header. |
| Author or byline | 3 | Byline, rel=author, author meta or author in schema. |
| Recently updated | 3 | Newest date from schema, meta tags and <time> elements: 180 days = full, 1 year = 70%, 2 years = 40%, older = 15%, none = 0. |
| Cites external sources | 2 | Distinct external domains linked from the main content (social links excluded). Full credit at 2. |
| HTTPS and canonical URL | 3 | HTTPS (50%) plus a canonical tag (50%). |
Competitors, citations and content gap
Competitors are the URLs you enter, or the top pages from a live web search for your question (your own domain and video or social platforms are skipped). Every page is scored with the same 22 factors.
The citation test searches your question on the live web, then an AI writes an answer from only those results and cites them as [n]. We record which results were cited, whether your page was among the results, and your position among cited sources. AI answers vary by wording and day, so treat each test as a sample.
The content gap asks an AI to list the subtopics a complete answer needs and mark which page covers each, using only the text of the pages. It also drafts a 40 to 60 word answer from facts on your page.
"They have it, you don't" lists factors where you score under 50% and at least half of the competitors score 50% or more. Your advantages is the reverse.
Fix priority: high when you score under 50% and the fix is worth 4+ points or most competitors already do it; medium for other gaps under 50%; low for partial credit. Points show how much your score rises when that factor is fully fixed.
What the score does not claim
- It does not show private rankings inside ChatGPT, Perplexity or Gemini; no tool can.
- It is not a Google ranking factor.
- It analyses the URL you submit; check each important page separately.
Methodology FAQ
How is the AEO score calculated?
Each of the 22 factors earns 0 to 100% of its weight. The AEO score is the sum of earned points divided by the points available, so factors that cannot be measured are excluded rather than counted as zero.
How are competitors chosen?
You can enter up to three. If you do not, the tool uses the top pages returned by a live web search for your target question, skipping your own site and video or social platforms.
How does the citation test work?
Your question is searched on the live web. An AI writes an answer using only those results and cites them by number, the same retrieve-then-cite process answer engines use. We record which results were cited.
Is the AI used to score my page?
Only for content coverage and the suggested answer. All other 21 factors are measured directly from your page HTML and robots.txt.