Methodology
How our AI visibility checker works
Our AI visibility checker measures both your AI readiness and your real-world AI visibility. It analyzes whether AI assistants can easily read and understand your content, then tests how often ChatGPT, Gemini, and Perplexity actually cite your website when answering buyer questions.
- ★ 5.0 on Clutch (10 reviews)
- 50+ projects delivered
- Since 2009
- Troy, Michigan
- Last updated November 2026
- Home
- AI Visibility Checker
- Methodology
Why we built the checker
Marketing teams can measure
- Website traffic
- Rankings
- Backlinks
- Conversions
But most organizations cannot easily measure
Whether ChatGPT, Gemini, or Perplexity are recommending them instead of competitors. The AI Visibility Checker was built to help companies understand and improve that visibility.
The checker at a glance
Two measurements, one report: your scores next to your competitors’, and a ranked list of fixes.
Your AEO score
- 1. Scan your page URL
- 2. Run the readiness checks: crawl access, structured data, content answerability, trust signals, technical health
- 3. AEO score, 0 to 100
Your AI visibility
- 1. Suggest 15 buyer questions (you can edit them)
- 2. Ask ChatGPT, Gemini and Perplexity, 3 times each
- 3. Record which sites each answer cites
- 4. AI visibility, % (you and 3 competitors)
The two numbers in your report
The first number is your AEO score (AI answer-readiness), from 0 to 100. The second is your AI visibility, a citation rate.
| Item | AEO score | AI visibility |
|---|---|---|
| What it measures | How easily AI assistants can find, read, understand and quote the page | How often ChatGPT, Gemini and Perplexity cite the page or site for buyer questions |
| How it’s measured | Checks of the page itself: crawler access, structured data, content, trust signals, Google PageSpeed | Real answers from the engines’ official APIs, several runs per question |
| Scale | 0 to 100 | A percentage, always shown with the number of answers behind it ("9%, 11 of 126") |
| What it tells you | What you can fix | Whether your fixes are turning into citations |
A page can score well on readiness and still be cited rarely, for example when competitors have stronger pages for the same questions. That’s why the report shows both, side by side with your competitors.
How the AEO score is calculated
The AEO score combines five areas into a score from 0 to 100. Each area is made of individual checks, and each check is worth a share of the area’s points.
| Item | Weight | What it covers |
|---|---|---|
| Content answerability | 35% | Whether the page answers buyer questions directly: question-style headings, a direct answer at the start of each section, steps and comparisons in lists or tables, specific facts |
| Structured data | 25% | Whether JSON-LD schema (Organization, FAQ, Article, Product, Breadcrumb and others) is present, valid, has its required properties and matches what the page shows |
| Crawl access | 20% | Whether search engines and AI crawlers may read and index the page, whether the main content is in the HTML without JavaScript, whether answers may quote it, the canonical link, a direct 200 response, the sitemap and llms.txt |
| Technical health | 10% | Google PageSpeed Insights: performance, accessibility, best practices, SEO and agentic browsing |
| Trust signals | 10% | Whether it’s clear who is behind the page, whether a named, qualified person stands behind the content, whether claims are backed by sources, and whether the page shows it is up to date |
Content answerability carries the most weight because pages that answer buyer questions clearly and directly tend to be cited more often by AI assistants. These are starting weights: we check them against real citation results and adjust them when the data shows a better fit. Every report states the scoring version it used.
How content and trust checks are judged
Checks like "headings ask the questions people search for" need judgment, not a yes or no. An AI model (Anthropic’s Claude) reads the page and scores each of these checks on a 5-point scale against a fixed written rubric, so every page is judged the same way. The report shows the score and what we found, for example "The main sections use generic titles. (Score 3 of 5.)" Treat scores in these areas as well-grounded guidance, not exact measurements.How technical health uses PageSpeed Insights
We run Google PageSpeed Insights on the page (mobile) and use the scores it shows, from 0 to 100. Each one is worth a set number of the AEO score’s 100 points.
| Item | Worth in the AEO score |
|---|---|
| Performance | 3.3 points |
| SEO | 2.2 points |
| Agentic browsing | 2.2 points |
| Accessibility | 1.1 points |
| Best practices | 1.1 points |
Core Web Vitals from real visitors are shown when Chrome has enough data for the page, but aren’t scored. Agentic browsing checks whether AI agents can browse and use the page; Google still marks it as under development, so it may change.
How we test AI visibility
-
The questions
We suggest 15 questions a buyer is likely to ask, based on what the page offers, your industry and your location, and you can edit them. The free preview tests the first 3. Questions that name a brand are reported separately: AI visibility uses only the generic questions.
-
The engines
Each question goes to ChatGPT, Gemini and Perplexity through their official APIs, with web search turned on, so answers draw on the live web. The report lists the exact model version used for each engine.
-
Repeated runs
AI assistants write each answer fresh, so the sources they cite change from run to run. We ask each question 3 times on each engine: 15 questions give 135 answers, and every result is a rate over those answers.
-
Counting citations
For every answer we record which websites it cites, then report your site citation rate, your page citation rate (ChatGPT and Perplexity, because Gemini names only the site), your share of voice across all citations, and rates by question and by engine.
How competitors are compared
We suggest up to 3 competitors (local, national or global) and you can change them. For each competitor we:
Run the same readiness checks
On the competitor’s page for the same offering (the report shows which page we used), so you see their AEO score and area scores next to yours. Technical health is measured for your page only.
Count their citations
In the same answers as yours, so the rates are directly comparable.
Show where they’re ahead
The checks they pass that your page doesn’t, with the points you’d gain by fixing each.
How issues and fixes are ranked
Every check that doesn’t pass becomes an issue with:
The action plan puts the issues in order: what to do next, then what to do when you can. Checks the page already passes are listed under "Already right", and anything we couldn’t check is listed with the reason.
What we found
On your page, and why it matters for AI assistants.
How to fix it
Often with ready-to-use code (such as schema markup) or draft content (such as question headings written from your page).
Impact and effort
The AEO score points you’d gain, and whether it’s a quick fix (minutes) or some work (hours).
Who
Who usually does it: web developer, content writer or marketing.
How fresh the results are
Results are reused for 24 hours
A page checked recently, by anyone, isn’t read again, and the results say how old they are ("Results from 3 hours ago"). Rescan checks everything fresh.
Rescan after 6 to 8 weeks
After you make fixes, AI citations usually take about 6 to 8 weeks to change, so rescan then to measure the difference.
What the results can and can’t tell you
- Not an official ranking. No AI company publishes how it chooses sources, so no tool can guarantee citations. A high AEO score means the page removes the common reasons AI assistants skip it.
- API answers, not the apps. We measure through the engines’ official APIs. Answers in the ChatGPT, Gemini or Perplexity apps can differ, for example because of a person’s location, chat history or settings.
- Rates, not certainties. Answers vary from run to run, so a rate over many answers is more reliable than any single check. The number of answers is always shown so you can judge it.
- Some judgment is AI-based. Content and trust checks use an AI model with a fixed rubric. They are consistent, but they’re guidance, not exact measurements.
- A snapshot, not tracking. Each report measures one page at one point in time. To see change over time, rescan the same page later.
Frequently asked questions about the methodology
Which AI assistants does the checker test?
ChatGPT, Gemini and Perplexity, through their official APIs with web search turned on. Each report lists the exact model version used for each engine.
Why does my AI visibility change between scans?
AI assistants generate each answer fresh, and the web pages they draw on change too. We reduce the noise by asking every question 3 times on each engine and reporting a rate, but some movement between scans is normal. Look for consistent changes across several questions.
Why does a branded question not count toward AI visibility?
A question that names your company almost always cites your own site, which would inflate the score. AI visibility uses only generic questions, the ones buyers ask before they know who you are; branded questions are shown separately.
Why doesn’t the checker use Google rankings?
Google rankings and AI citations are related, but they are not the same. AI assistants often cite sources that answer specific questions clearly, contain structured data, and provide trustworthy information. A page can rank well in search and still receive few AI citations, or vice versa.
Why is my AEO score high but my AI visibility low?
Readiness removes the reasons AI assistants skip a page, but they still choose between pages. If competitors have pages that answer the same questions more directly, or are better known for the topic, they may be cited instead. The competitor comparison and the per-question results show where.
What data do you store?
We store the page URLs you scan, the competitors and questions you choose, your report results, and your email address if you request a full report. Your email is the only personal information we collect; everything else is publicly available information about websites. See our privacy policy.
Is the checker free?
The free preview, with your AEO score and a 3-question test, needs no signup, and your first 3 full reports are free when you enter your email. After that, each additional report costs $19.99, or $79.99 for a pack of 5. See pricing.
See how visible your page is to AI
Get your AEO score and AI visibility, compared with 3 competitors, in a few minutes.