AEO, GEO and SEO readiness checker. Scored against a published method.
Answer 18 questions about your company and your site. You get a weighted score out of 100 across five pillars, your pillar by pillar breakdown, the three fixes worth doing first, and every weight and source we used to reach the number.
The five pillars
Weights out of 100. Access is heaviest because it gates every other pillar. Each bar below is a weight, not a score.
Your readiness score
0 of 100. Scoring
What to do first
Scored against the method published below, version 1.0, last reviewed 14 October 2026. Your answers and this score have been sent to our team, so there is nothing else to submit. Read the methodology
Pillar by pillar
Each bar is your score against the points available for that pillar. The weights are published below and the browser scores with those exact numbers.
The three fixes worth doing first
An access gap always takes the first slot, because nothing downstream can be retrieved or quoted from a page an engine cannot fetch. The other two slots go to whatever is leaving the most points on the table, and ties break in pillar order. That rule is the method, not a preference.
Want a second pair of eyes on it?
Twenty minutes with a senior operator. Bring the score. We will run your category questions against the engines before the call and tell you who is being named instead of you.
Methodology
Audit the score before you trust it. Weights, bands, limits and sources.
Version 1.0, last reviewed 14 October 2026. The scoring model lives in one file, and the same file renders these tables and runs the arithmetic in your browser, so what is published here is what scored you.
The five pillars and their weights
Grounded in what the platforms themselves publish about how their AI features retrieve and cite, listed in the sources below.
| Pillar | Weight | What it measures | Why it is weighted that way |
|---|---|---|---|
| Crawlability and access | 25 | Whether the engines can fetch your pages at all: robots.txt rules for AI search crawlers, your CDN and firewall, text present in the raw HTML, and pages indexed in both Google and Bing. | It is weighted highest because it gates everything else. Google will only use a page in its AI features if that page is indexed and eligible to appear with a snippet. OpenAI says a site that disallows OAI-SearchBot will not appear in ChatGPT search answers. Perplexity asks owners to allow PerplexityBot to appear in its results. A perfect page nobody can fetch scores zero in every answer. |
| Answerability | 25 | Whether a passage on your page can be lifted out and used as the answer: a direct answer in the first two sentences under a question heading, the page types buyers ask for (comparisons, cost explanations, implementation detail, original data), and visible freshness. | Equal to access because retrieval only puts you in the candidate pool. Google names unique, non-commodity content with first-hand expertise as likely the most influential factor over the long run, and it asks for a good reading experience rather than fragmented text. Comparison and question queries are also where AI answers show up most, so these are the pages that get read. |
| Corroboration | 20 | Whether independent sources agree about you: review platforms, the best-of lists in your category, partner and analyst directories, press, LinkedIn, video and podcasts, and a Wikipedia article or Wikidata item if you qualify. | Your own site is one witness. Orbit Media's 2026 study of B2B buyer questions found engines leaning heavily on third parties, and leaning on different ones: Clutch was the most cited third-party domain for Claude, while LinkedIn and YouTube led for Perplexity and listicle roundups led for Gemini. Review platforms and curated lists carry the most weight inside this pillar, and Wikipedia the least, because most mid-market B2B companies do not meet its notability bar. |
| Proof and entity clarity | 15 | Whether your claims can be checked and your facts are stated the same way everywhere: named client outcomes with numbers, named authors with bio pages, Organization markup with sameAs, and one wording for who you are across your site and every profile. | Weighted below corroboration on purpose. Google states plainly that structured data is not required for generative AI search and that there is no special markup to add, so a tool that treated schema as a gate would be overselling it. Markup and named proof still earn their place: they make your facts machine-readable, they keep rich results working, and inconsistent facts are the most common reason an engine describes a company wrongly. |
| Measurement and ownership | 15 | Whether you can see what is happening and somebody owns it: a fixed prompt set run on a schedule, the generative AI report in Search Console, AI Performance in Bing Webmaster Tools, a GA4 channel for AI assistants, crawler counts from server logs, a free-text question on how buyers found you, and a named internal owner. | Lowest weight because measurement does not change what an engine sees this week. It is still scored, because AI answers move run to run, platform reporting now exists on both Google and Microsoft sides, and work with no owner and no baseline stops after one quarter and cannot be defended to a board. |
| Total | 100 | Eleven scored checks. The seven questions about your company, markets, competitors and expected outcome score nothing: no market or competitor is inherently better for AI visibility, and weighting one would be the kind of invented number this page argues against. | |
Every question and what it scores
Answer values in brackets. Checkbox questions add up, up to the cap shown.
| Question | Pillar | Points | How the answers score |
|---|---|---|---|
| Your name, work email, company, website and role | Context | 0 | Context only, not scored |
| What would a buyer type to find what you sell? | Context | 0 | Context only, not scored |
| Which markets do you sell into? | Context | 0 | United States (undefined); Canada (undefined); United Kingdom and Europe (undefined); India (undefined); Singapore and Southeast Asia (undefined); United Arab Emirates (undefined); Saudi Arabia (undefined); Qatar, Kuwait or Bahrain (undefined); Australia and New Zealand (undefined); Other (undefined) |
| Who is the ideal customer, in one line? | Context | 0 | Context only, not scored |
| Which three competitors do you actually lose deals to? | Context | 0 | Context only, not scored |
| What outcome do you expect from AEO, GEO and SEO work? | Context | 0 | Context only, not scored |
| By when do you need it? | Context | 0 | This quarter (0); Within 6 months (0); Within 12 months (0); No date yet, still exploring (0) |
| In robots.txt, and at your CDN or firewall, are the AI search crawlers allowed through? | Crawlability and access | 10 | All of them are allowed, and I have checked the CDN too (10); Some are allowed, some are blocked (5); Blocked, or I do not know (0) |
| Is the main text of your key pages present in the raw HTML? | Crawlability and access | 8 | Yes, the full text is in View Page Source (8); Partly, some sections are injected (4); No, or I do not know (0) |
| Are your key pages indexed in both Google and Bing? | Crawlability and access | 7 | Yes, in both (7); In one of the two (4); No, or I do not know (0) |
| Do your key pages answer the question directly in the first two sentences, under a heading phrased as the question? | Answerability | 9 | Yes, on every page that matters (9); On some pages (5); No, pages open with context or positioning (0) |
| Which of these do you publish today? | Answerability | 8 | Comparison or alternatives pages that name competitors (2); An explanation of how you are priced, even without a rate card (2); Public implementation or integration detail, or docs (2); Original data or research of your own (2); Question and answer pages built from real buyer questions (2). Capped at 8 |
| Do your guides, comparisons and cost pages show a visible update date, with a review cadence behind it? | Answerability | 8 | Yes, visible dates and a review cadence by page type (8); Dates on some pages, no cadence (4); Neither (0) |
| Can a stranger check your claims: named client outcomes with numbers, and named authors with bio pages? | Proof and entity clarity | 8 | Both: outcomes with numbers, and named authors with bios (8); One of the two (4); Neither (0) |
| Do you have valid Organization markup with sameAs links, and the same wording about who you are everywhere? | Proof and entity clarity | 7 | Both: valid markup, and one consistent description everywhere (7); One of the two (4); Neither, or I do not know (0) |
| Which of these are true about you off your own site? | Corroboration | 20 | A complete review platform profile with recent reviews, on Clutch, G2 or your category equivalent (5); Included, and described accurately, in the best-of lists for your category (5); Listed in the partner or analyst directories that apply to you (3); Independent press or trade coverage in the last 12 months (3); A complete company LinkedIn page that matches your site description (2); Video or podcast appearances with real transcripts (1); A Wikipedia article or a sourced Wikidata item (1) |
| How do you measure AI visibility today? | Measurement and ownership | 10 | A fixed set of buyer prompts, run against the engines on a schedule (4); The generative AI performance report in Google Search Console (2); AI Performance in Bing Webmaster Tools (2); A GA4 channel group for AI assistant referrals (1); Crawler hits and status codes from server logs (0.5); A free-text "how did you hear about us" question you actually read (0.5) |
| Does anyone own this work inside the company? | Measurement and ownership | 5 | Yes, a named owner with time set aside for it (5); Shared across the team, nobody accountable (2.5); Nobody owns it (0) |
The bands, and what each one means
The three lower readings are the ones published in chapter 2 of our own AEO and GEO playbook, unchanged, so you can check the checker against the method it came from. The top band splits the highest readings, because the advice at 95 is not the advice at 72.
| Score | Reading | What to do first |
|---|---|---|
| 0 to 39 Invisible | AI assistants either cannot read you or do not know you. Nothing further up the list is worth funding until the engines can fetch your pages and find you in an index. | Fix access and indexing this week, then write the one-paragraph definition of what your company is and who it is for, and use that exact wording everywhere. |
| 40 to 69 Known but not chosen | You exist and you are readable, but competitors get recommended ahead of you. This is the most common place for a B2B site with good SEO habits and no answer-engine work. | Build a fixed prompt set so you can see where you stand, then write the comparison and cost pages buyers ask for, and get accurately represented on the third-party sources those answers already cite. |
| 70 to 89 In the conversation | You are readable, answerable and corroborated, so you are named in some answers and cited in a few. What is left is holding position on the questions that produce pipeline. | Formalize measurement across Google, Bing and the assistants, then defend your high-intent questions with original data, named authors and real update dates. |
| 90 to 100 Defending the answer | The foundations are in place. The risk now is drift: a firewall rule, a stale page or a competitor publishing better proof than you. | Keep the prompt set running on a fixed cadence, watch crawler status codes in your server logs, and put your pages on a review schedule by page type. |
What this tool cannot see
Published next to the score on purpose. A readiness score that hides its blind spots is a sales device.
- This tool does not fetch your website. Every answer is self-reported, so the score is exactly as accurate as your answers. Answering "I do not know" scores zero, which is the honest reading: if nobody can confirm it, you cannot rely on it.
- It cannot see your analytics. It knows nothing about your sessions from AI assistants, your impressions or citations in Google and Bing, your rankings or your pipeline. Those take access to Search Console, Bing Webmaster Tools, GA4 and your CRM.
- It does not run prompts against the engines. It cannot tell you whether ChatGPT, Perplexity, Gemini or Copilot name you today, who they name instead, or how they describe you. That needs a prompt set run against each engine and scored.
- It does not read your robots.txt, your CDN rules, your HTML or your markup. A real audit does all four, and the CDN is where the surprises usually are.
- AI answers move from run to run. Even a measured snapshot is a snapshot, not a trend, which is why the method asks for a fixed prompt set on a cadence rather than one good screenshot.
- This is a readiness score, not a forecast. It describes how ready your site is to be retrieved, selected and corroborated. It does not predict traffic, rankings, citations or revenue, and no number on this page should be read as a projection.
Sources
Platform documentation first, independent studies second, our own client work labeled as ours. Every date is the date we last checked the page.
- Google Search Central, "Optimizing for generative AI in Google Search", last updated 10 July 2026. Indexing and snippet eligibility, unique non-commodity content, no chunking requirement, no special markup.
- Google Search Central, "AI features and your website", checked 14 October 2026. How Google AI features use the index, and the controls site owners have.
- Google Search Central, "Organization structured data", checked 14 October 2026. The Organization fields and sameAs, used in the entity clarity line.
- OpenAI, "Overview of OpenAI crawlers", checked 14 October 2026. OAI-SearchBot must be allowed to appear in ChatGPT search. GPTBot is a separate, training-only decision.
- Perplexity, "Perplexity crawlers", checked 14 October 2026. Allow PerplexityBot to be surfaced and linked. Perplexity-User is user-triggered and is handled at the firewall.
- Anthropic, "Does Anthropic crawl data from the web, and how can site owners block the crawler?", checked 14 October 2026. ClaudeBot and the search-side crawlers, and how to allow or block them.
- Microsoft Bing Webmaster Blog, "Introducing AI Performance in Bing Webmaster Tools Public Preview", 10 February 2026. Citations in AI answers across Copilot and Bing summaries, cited pages and grounding queries.
- RFC 9309, Robots Exclusion Protocol, September 2022. What a robots.txt rule does and does not oblige a fetcher to do.
- Seer Interactive, "AIO Impact on Google CTR: 2026 Update", 24 April 2026. 53 brands, 5.47 million queries, 2.43 billion impressions. AI Overviews on 95.4 percent of comparison queries and 85.9 percent of question-format queries. Brands cited inside an AI Overview earned about 120 percent more organic clicks per impression than when they were not cited.
- Orbit Media, LLM citation sources study, 2 September 2026. 72 B2B prompts, 13,184 citations. Clutch was the most cited third-party domain for Claude. LinkedIn and YouTube led for Perplexity. Used to weight the corroboration channels.
- Wikipedia, "Notability (organizations and companies)", checked 14 October 2026. Why a Wikipedia article is scored lowest in corroboration: significant independent coverage is required, and paid article creation is a common route to deletion.
- Lemniscate Growth, The B2B AEO and GEO Playbook 2026, chapter 2, checked 14 October 2026. Our own published 10-line self-audit and its three readings. The three lower bands here use those readings unchanged.
Why the questions are the questions
Retrieved, selected, corroborated. In that order.
Three things have to be true. Every pillar sits under one of them.
Retrieved. An assistant searches an index, then fetches candidate pages. Google will only use a page in its AI features if it is indexed and eligible to appear with a snippet. OpenAI publishes that a site which disallows OAI-SearchBot will not appear in ChatGPT search answers, and Perplexity asks site owners to allow PerplexityBot so pages can be surfaced and linked. This is why the access questions come first and carry the most points, and why a firewall rule nobody told you about is the most expensive thing on this page.
Selected. The model reads the candidates and quotes the passages that answer the question best. Google's own guidance says there is no requirement to break content into tiny pieces and no special markup to add, and that unique content with first-hand expertise is likely the most influential factor over the long run. So the answerability questions ask about the shape of your pages, not about tricks: a heading phrased as the question, the answer in the next 40 to 60 words, and the page types buyers actually ask for. Comparison and question queries are where AI answers appear most: Seer Interactive found AI Overviews on 95.4 percent of comparison queries and 85.9 percent of question-format queries across 5.47 million queries, published 24 April 2026.
Corroborated. Your site is one witness. Orbit Media's study of 72 B2B prompts and 13,184 citations, published 2 September 2026, found Clutch was the most cited third-party domain for Claude while LinkedIn and YouTube led for Perplexity. That is why review platforms and curated lists carry more weight inside the corroboration pillar than a Wikipedia article, which most mid-market B2B companies do not qualify for anyway.
Being cited is also worth something measurable. In the same Seer dataset, brands cited inside an AI Overview earned about 120 percent more organic clicks per impression than when they were not cited. We do not turn that into a revenue promise for your company, and neither should anyone else.
If you want the long version, the AEO and GEO playbook is the full method, and the explainers cover how AI crawlers work, how AI Overviews work and how to benchmark AI visibility.
After the score
What a real audit adds. The things a form cannot see.
Your prompt set, run
Fifty to a hundred of your buyer questions run against ChatGPT, Perplexity, Gemini and Copilot, scored on mention rate, citation rate, position and share of voice against the three competitors you named.
Access checked, not asked
Your robots.txt, your CDN rules, your rendered HTML and your crawler status codes in the server logs, read rather than self-reported.
The corroboration map
The domains the answers in your category already cite, where you are absent, and where you are present but described wrongly.
A baseline you can report
The Search Console and Bing AI reports connected, a GA4 channel for AI assistants, and the four numbers to bring to a board each month.
Proof
Where this work led.
return on every event attended
A Dubai crypto infrastructure company that could not run US ads. Search, events and ABM built its pipeline against Fireblocks-class rivals.
Read story
pipeline in 28 months
DLT Labs had no funnel. We built the GTM, website, ABM and webinars, then used a retail case study to open port and rail accounts in Dubai.
Read story
of results from inbound
AI contract lifecycle management on ServiceNow. Category content and search took qualified meetings from single digits to tens a month.
Read storyQuestions about the checker. Answered plainly.
Still unsure? Ask us directly.
What is an AEO, GEO and SEO checker?
It is a readiness assessment. It scores whether AI assistants and search engines can fetch your pages, lift a clear answer out of them, and find independent sources that agree about you. This one asks 18 questions, scores 11 of them across five weighted pillars out of 100, and publishes every weight, band and source it uses. It does not crawl your site, so the score reflects what you report.
How is the score calculated?
Crawlability and access is worth 25 points, answerability 25, corroboration 20, proof and entity clarity 15, and measurement and ownership 15. Each question carries a fixed number of points inside its pillar, and the full table of questions and answer values is published on this page. The browser does the arithmetic with those exact numbers, so you can reproduce the result by hand.
Why is access weighted most heavily?
Because it gates everything else. Google will only use a page in its AI features if that page is indexed and eligible to appear with a snippet, OpenAI says a site that disallows OAI-SearchBot will not appear in ChatGPT search answers, and Perplexity asks site owners to allow PerplexityBot. A well written page that no engine can fetch contributes nothing to any answer.
Does the score predict traffic or revenue?
No. It is a readiness score and nothing more. It describes how ready your site is to be retrieved, selected and corroborated. It cannot see your analytics, your pipeline or how any engine answers a question today, and we do not publish a conversion rate or a revenue multiple against it.
Is structured data required to appear in AI answers?
No. Google states that structured data is not required for generative AI search and that there is no special schema.org markup you need to add. Markup still earns points here, because it makes your facts machine-readable, keeps rich results working and helps every source describe you the same way, which is the most common cause of an engine getting your company wrong.
What happens to my answers?
They go to our team so a strategist can read the result before any conversation, along with the page you came from. Nothing is sold or shared. If you would rather talk it through without filling the form, book a slot and bring the score.
Rather have someone read it with you? Twenty minutes, your score, no pitch.
- A senior operator, not an SDR
- We run your category questions before the call
- Slots across US, India and GCC time zones
Prefer email? growth@lemniscategrowth.com