← how ARIA works
Help · AEO readiness

How I score your AEO readiness

More and more of your customers don't search — they ask. AEO (Answer Engine Optimization) readiness is my measure of how easily AI assistants can read, trust, and cite your website when they answer.

Why this matters

When someone asks ChatGPT, Claude, or Perplexity "who does this near me" or "what's the best way to…", the assistant doesn't return ten blue links. It reads a handful of pages it can parse and trust, composes one answer, and cites its sources. Either your site is one of those sources, or someone else's is.

Classic SEO optimized for a position on a results page. Answer engines work differently: they extract passages, prefer clearly structured and dated content, and lean on signals of authorship and evidence. That is what I measure — not where you rank, but whether your pages are built to be quoted.

How the score works
01
I read your site

Once a domain is verified I crawl it on a schedule, rendering each page the way a browser — and a modern AI crawler — sees it, JavaScript included.

02
I score every page, 0–100

Each page is checked against the five dimensions below, weighted into a single score. Every check comes with a plain reason, so you always know why.

03
Your domain score is the average

The plain mean of your page scores — no tricks. Lifting your weakest pages lifts the score.

80–100 Cite-ready — assistants can parse, date, and attribute your content.
50–79 Readable, but easy to skip — some signals are missing.
0–49 Mostly invisible to answer engines.
What makes up the score

Five dimensions, weighted by how much they move an assistant's decision to cite a page. Each one scores from 0 to 1; I combine them into the page's 0–100 readiness with a weighted mean:

readiness = ( Σdimensions result(0–1) × weight ) × 100

= ( Structured data×0.20 + Answer structure×0.25 + Freshness×0.15 + Extractability×0.15 + Authority / E-E-A-T×0.25 ) × 100

The bars show each dimension's share of the total.

  • Structured data · 0.20
  • Answer structure · 0.25
  • Freshness · 0.15
  • Extractability · 0.15
  • Authority / E-E-A-T · 0.25
0 50 100 50 · wait 80 · go 0.20 Structured data 0.25 Answer structure 0.15 Freshness 0.15 Extractability 0.25 Authority / E-E-A-T
Bar height is each dimension's weight on the same 0–100 scale as the score. The dashed lines mark the go and wait cut-offs — a perfect showing on every dimension stacks to 100.
Why these weights, and why a mean

Structured data and answer structure carry the most weight because they most determine whether an engine can lift a clean, quotable answer from your page at all. Freshness, extractability, and authority refine that judgment rather than make or break it, so each weighs less. I combine the five as a weighted mean, not a minimum — one weak dimension lowers the score in proportion to its weight instead of zeroing the whole page. That keeps the number honest and actionable: it tells you how much headroom is left and which lever moves it most.

The five dimensions — and how to improve each
Structured data heaviest signal

JSON-LD structured data is the label on the tin: it tells machines, unambiguously, what a page is — an article, an FAQ, a product, an organization. Pages that declare a relevant type are far easier to cite; valid JSON-LD with only a generic type earns partial credit.

To improve: most CMSs and SEO plugins can emit JSON-LD for you. Mark articles up as Article or BlogPosting, question pages as FAQPage, products as Product, and your homepage as Organization or WebSite.

Answer structure

Assistants lift passages that directly answer a question. I look for one clear h1, and for H2/H3 section headings phrased the way people actually ask — "What does it cost?", "How do I get started?".

To improve: turn section headings into the questions your customers ask, and answer each one in the first sentence or two below it. The more of your headings that read as questions, the better.

Freshness

Models prefer sources they can date — an undatable page is a risky citation. Best is a published or modified time in your page metadata; visible date elements on the page earn partial credit; no date signals at all earn none.

To improve: emit article published/modified meta tags (most CMSs can), and keep your cornerstone pages genuinely up to date.

Extractability

Before anything can be cited, it has to be read. I measure how much of a page is actual text versus markup, and whether the content lives in a clear, semantic main content region rather than anonymous markup.

To improve: wrap your content in a clear main content region, make sure the words arrive as server-rendered text rather than being drawn by script, and trim boilerplate markup.

Authority / E-E-A-T

Assistants weigh experience, expertise, authoritativeness, and trust. I check three concrete proxies: a named author in your metadata, at least one outbound link citing a source, and concrete numbers — prices, dates, statistics — in the text itself.

To improve: add author metadata to your articles, link out to the sources you rely on, and be specific — real figures beat adjectives.

Where to see yours

Your score and its full breakdown live on each verified domain's page, including your weakest pages and the exact reasons they lag. I re-read your site every 6 hours — or right away when you press refresh now — and re-score everything on every pass, so improvements show up quickly.