AI Visibility Audit across 46 criteria
An AI visibility audit checks whether ChatGPT, Perplexity, Gemini, and Google AI Overviews can understand and cite your site — and why, right now, they probably can't. LightSpot audits any URL against 46 measurable criteria — 21 SEO and 25 GEO (Generative Engine Optimization) — and returns an actionable diagnosis in minutes: a score, a letter grade from A to F, and the 3 fixes that matter most for getting cited. Every page is judged for what it actually is, and the 46 criteria behind the score are public.
Start free trial
What is an AI visibility audit?
An AI visibility audit is an automated analysis that checks whether a site's content is structured, sourced, and accessible enough to be picked up by generative search engines. Unlike a standard SEO audit, it measures citability — not just ranking in Google's blue links. LightSpot runs this analysis through 46 measurable criteria: 21 SEO checks for technical foundations, and 25 GEO checks for everything that determines whether a passage on your page can be cited by an AI — answer structure, structured data, AI crawler accessibility, brand signals.
How to audit your site's AI visibility
Auditing AI visibility follows a repeatable method. 1. You enter a URL — LightSpot crawls the site's key pages, respecting robots.txt. 2. The 21 SEO criteria are checked through code analysis (tags, structure, Core Web Vitals via Google PageSpeed Insights). 3. The 25 GEO criteria are evaluated — the semantic checks by our AI engine, the technical checks (structured data, robots.txt, AI crawler accessibility) by code analysis. 4. Every criterion gets a pass, fail, or skip verdict — never a crash — and the report is always produced, with a score, a grade, and the top 3 priority actions.
Every page judged for what it actually is
A product page has no business quoting three experts, and a category page doesn't need a thousand words of editorial prose. So LightSpot classifies every audited page — editorial, transactional, listing or category, corporate, home — from its URL and its structured data, then refines that call with our agents. Criteria that make no sense for that page type are marked not applicable and drop out of the score's denominator: they count neither as a pass nor as a failure. A criterion's weight, on the other hand, never moves — it either applies or it doesn't. Two very different pages stop being measured with the same ruler, and the score stops punishing a page for something it was never meant to do.
Why ChatGPT doesn't cite your site
Most sites aren't cited by ChatGPT or Perplexity because they're built for Google rankings, not for extraction by a generative AI: no direct answer at the top of a section, no structured data, no sourced statistics, stale content. A strong Google ranking guarantees nothing when it comes to AI citation. LightSpot pinpoints exactly which of these factors are missing on your site, page by page, and ranks them by impact instead of handing you a generic list of 200 warnings.
Why content structure determines AI citation
According to the Princeton/ACM SIGKDD 2024 study on AI citation optimization, structuring a section as a question and answer (an H2 phrased as a question, followed by a self-contained answer) increases AI visibility by 27%. Citing at least two recognized institutions in the same piece of content pushes that figure to 115% — while the single strongest factor is citing a named expert with their title, at +41% according to the same study. These aren't formatting tricks: they're the signals generative engines use to judge whether a passage is trustworthy and extractable. LightSpot's methodology (46 criteria) checks for their presence, section by section.
A score, a grade, and the 3 actions that matter most
Every LightSpot audit produces an overall score out of 100, a letter grade from A to F, and a page-by-page breakdown — but above all, the 3 highest-impact fixes, the ones that move AI visibility the most, instead of a 200-item warning list nobody knows where to start with. Every criterion is marked pass, fail, or skip, and never silently fails: if a check can't run, it's marked skip and the report is still produced with whatever could be measured.
Online stores: Google Shopping compliance
If your site is detected as a store — Shopify, WooCommerce, or simply product data in the markup — twelve extra merchant criteria switch on. They start with your mandatory pages: a return policy stating an explicit window, terms of sale, a privacy policy, shipping costs, and two ways to reach you. They then check your JSON-LD product data: name, image, description, price, availability and condition in the official vocabulary, a product identifier (a correctly formatted GTIN, or an MPN with the brand), and above all a price that matches between the structured data and the price shown on the page. Finally they make sure Google Shopping's crawlers aren't blocked and your product pages stay indexable. These are exactly the points that get a product rejected from Google Shopping, and the ones that make a product understandable to an AI engine. On a site that isn't a store, this block never appears.
Our 46 criteria are public and versioned
The 46 criteria that make up the score are published on GitHub, at github.com/nmarijane/geo-criteria: every criterion, its pass condition, and the score formula, in plain language, readable without an account. That reference is generated from the same source as the audit engine, and our test suite blocks any release where a criterion changes without the reference being regenerated — so it cannot drift away from the product. Merchant criteria and accessibility checks, each scored on its own scale, are documented separately. You can check what we measure before taking our word for it, argue with a rule, or use it to walk a client through a report.
A shareable report, built for agencies
Every audit generates a public report link: ideal for an agency presenting a diagnosis to a prospect, or for sharing a result on LinkedIn. The public report shows the score, the grade, and the headline findings; the full breakdown — criterion by criterion, page by page — is reserved for connected accounts. Without an account, you already see enough to grasp the scale of the problem; with one, you get the full action plan and access to the correction autopilot.
Frequently asked questions
How do I audit my site's AI visibility?
Run a LightSpot audit on your site's URL: it checks 46 measurable criteria — 21 SEO and 25 GEO — in minutes and returns a score, a letter grade from A to F, and the top 3 priority fixes. No installation required — the audit runs remotely against your site's pages.
What criteria matter for auditing my website's AI visibility?
LightSpot evaluates 25 GEO criteria: citable passage structure, direct-answer presence, structured data, AI crawler accessibility (GPTBot, PerplexityBot...), brand signals, and content freshness — alongside 21 standard SEO criteria (tags, structure, Core Web Vitals) covering the technical foundations search engines still rely on.
Where can I read the full list of 46 criteria?
The reference for the 46 scoring criteria is public on GitHub, at github.com/nmarijane/geo-criteria: every criterion, its pass condition, and the score formula. It's generated from the same source as the audit engine and versioned on every change, so it can't describe anything other than what the engine measures. Merchant criteria and accessibility checks are documented separately.
Does the audit cover e-commerce sites?
Yes. When your site is detected as a store, twelve merchant criteria are added: your mandatory pages (return policy with an explicit window, terms of sale, privacy policy, shipping costs, contact details), your JSON-LD product data (name, image, description, price, availability and condition in the official vocabulary, a product identifier, a price matching the one shown), and Google Shopping's crawlers reaching indexable product pages. We check the GTIN's format, not its check digit; with no GTIN, an MPN with the brand is enough.
How do I get cited by ChatGPT?
By structuring your content the way a generative AI extracts information: a direct answer at the top of each section, sourced and attributed data, structured data in place, recent content. LightSpot measures these criteria, and the autopilot can apply some fixes directly to your CMS.
How do I track and improve my citations on AI engines?
LightSpot's citation tracking probes ChatGPT, Perplexity, Gemini, and Claude every week with prompts tailored to your business, detects which citations you gained or lost, and alerts you by email — the report's Impact section then shows exactly how your AI visibility is trending over time.
What is GEO (Generative Engine Optimization)?
It's the practice of optimizing content to be cited by generative search engines — ChatGPT, Perplexity, Gemini, Google AI Overviews — where classic SEO instead optimizes for ranking in Google's blue links. LightSpot measures it through 25 dedicated GEO criteria, alongside 21 standard SEO ones.
Other features