AEO checker

The seven layers the tool checks

Citability by AI engines does not depend on a single factor but on the combination of several technical layers worth reviewing separately. This tool examines each one and reports its state independently, so you know exactly where the problem is when the score is low.

Layer one: Schema.org structured data. AI engines read JSON-LD to resolve entities and extract structured information. We look for the presence of the types that most impact AEO: Organization (brand identification), FAQPage (pages with FAQ are 3.2 times more likely to appear in AI answers), Article with author (for editorial content), Service or Product (for commercial pages), LocalBusiness (for local businesses), BreadcrumbList (for navigation) and HowTo (for tutorials). Each one adds to the score.

Layer two: basic meta tags. A well-formed title, a useful meta description, Open Graph tags that control how the site looks when shared, hreflang if there are versions in several languages. These elements seem basic but are missing or wrong in many sites, and AI engines use them to understand what each page is about.

Layer three: semantic structure. A single H1 per page (not multiple), a coherent hierarchy of H2, H3, H4 that reflects the organization of the content. When the headings are poorly structured, the AI has difficulty identifying extractable sections and opts for other sources.

Layer four: detectable FAQ sections. Beyond having FAQPage schema, the checker identifies whether the page has patterns that suggest clearly delimited questions and answers. Explicit FAQs are the format generative models extract most.

Layer five: author identification. For articles and editorial content, AI models give more weight to sources with identifiable authorship. We check author meta tags and the presence of author inside the Article schema when it applies.

Layer six: freshness signals. Publication date and, above all, a visible modification date. Pages not refreshed in the last six months are three times more likely to lose AI citations. We check that these dates are present and reasonably recent.

Layer seven: bot configuration. Although we cannot directly read your robots.txt from the client, we try to detect it and, if accessible, identify whether bots critical for AEO (OAI-SearchBot, PerplexityBot, Google-Extended) are allowed or blocked.

How the score is calculated

The score combines the results of the seven layers with different weights according to their impact on AEO. Structured data weighs more than the presence of basic meta tags, because the difference they make in citability is greater. FAQs have a significant weight because they are the format with the largest documented lift. Freshness weighs moderately. Bot configuration, when it can be verified, can neutralize everything else if it is wrong: a site that blocks OAI-SearchBot can have a high score in other layers and still be invisible in ChatGPT.

The tool does not compensate some layers with others linearly. If a critical layer is broken (for example, there is no Organization schema), the score is penalized more strongly than if only something optional is missing. That reflects reality: AI engines harshly penalize the absence of the basics, while tolerating absences in secondary layers as long as the critical ones are present.

Honest limitations: what the tool does not detect

Three limitations worth being clear about. First, it does not evaluate content quality. A site can have perfect structured data, impeccable hierarchy, marked FAQs and an identified author, and still not be cited because the content is empty, copied from other sites or sounds AI-generated. That evaluation requires human reading or using our filler detector tool as a complement.

Second, it does not measure real appearance in AI engines. The tool checks whether your site has the technical conditions to be cited, but it does not query ChatGPT, Perplexity or Gemini to see whether it really appears. That measurement requires the manual prompt method we describe in our article on how to appear in AI engines.

Third, it does not audit domain authority or backlinks. The external signals (brand mentions, links from reliable sources, presence in directories) are critical for AEO but are not visible from the HTML of an individual page. A complete AEO audit includes these external dimensions; our tool covers only what is verifiable from the page itself.

How to interpret the result and what to do next

The tool's result is the prioritized list of what to fix. The practical rule for using it is to follow the order of impact: first address the critical layers with high weight (structured data, bot configuration), then the medium ones (FAQs, authorship, freshness), lastly the details (semantic structure, secondary meta tags). Solving the first three usually raises the score more than 20 points and represents the base on which everything else works.

A good practice is to run the tool before any change, note the baseline score, apply the priority adjustments and run it again after a week (time for the changes to be reflected and the engines to reindex). The change in the score is the direct measure of the impact of the work done. Like any technical optimization, the progress is cumulative: three or four iterations spaced over time build a sustained improvement more than a single big attempt to fix everything at once.

What makes the difference between a "present" schema and a "useful" schema

The tool detects the presence of Schema.org schemas but it is worth making explicit what makes them really effective for AEO, because many sites have declared schemas that technically validate but contribute little. The difference is in three factors the tool cannot judge automatically and that are worth reviewing by hand when the checker reports presence.

First, completeness of optional fields. A minimal Organization schema with only name and url validates and adds to the score, but a complete Organization schema with name, url, logo, sameAs (links to official profiles), contactPoint and address resolves the entity much better. The same happens with Article: the basic one has headline and datePublished; the effective one adds author with a well-structured Person object, dateModified, publisher, mainEntityOfPage and image. Each extra field reduces the ambiguity when the AI engine decides whether to cite you or cite someone else.

Second, coherence between what is declared and what is visible. If the schema declares an address on Calle 50 but the visible page shows an address in Costa del Este, the AI engine detects the conflict and discounts trust. If the FAQPage schema says the page has 8 questions but only 6 are visible in the HTML, the same. Coherence is invisible to a superficial technical validation but critical for real citability. The practical rule is: any data in the schema must be verifiable by looking at the visible page.

Third, linking between entities. Schemas are much more powerful when they link to each other through @id and cross-references. An Article that references its Organization through publisher, that Organization that has sameAs pointing to its official profiles, and all within a coherent @graph, gives the AI engine a network of verifiable relationships that an isolated schema never reaches. Sites serious about AEO use @graph; sites that only add schemas as technical compliance do not.

How the importance of each layer varies by type of site

Although the score combines all the layers with fixed weights, in practice the relative importance varies according to the type of site you are evaluating. It is worth nuancing the interpretation according to the context.

Institutional and professional services sites. Here Organization, LocalBusiness, Service and FAQPage weigh more. Article and BlogPosting are secondary if the site does not have an active blog. The critical thing is that the AI engine can resolve the company's identity, its location, its services and the answers to the questions a potential client would ask.

Editorial sites and blogs. Here Article, BlogPosting, author with a well-structured Person, datePublished and dateModified weigh more. Organization is still necessary but secondary. Freshness is critical because editorial content loses value fast if it is not maintained.

Online stores. Product schema on each listing is priority one. Organization, Review and AggregateRating follow in importance. FAQPage on information pages helps. Freshness applies differently: the catalog changes constantly, so it is evaluated by the update of individual products rather than by the global dateModified of the site.

Portfolio or personal sites. Here Person with sameAs to professional profiles (LinkedIn, GitHub, official profiles) weighs more than Organization. CreativeWork may apply for published projects. Authorship is implicit across the whole site, which simplifies some layers but makes the consistency of the personal identity across all the schemas critical.

How this tool compares with similar ones

There are other tools on the market that evaluate similar dimensions, and it is worth situating ours honestly so you know when to use it and when to turn to more complete alternatives.

Official Google validators. Rich Results Test and Schema Markup Validator are the reference for validating JSON-LD syntax. Our tool does not replace that rigorous validation: if a page fails in Schema Markup Validator, it will not appear in rich results, regardless of what the AEO checker says. The difference is one of purpose: the official validators ensure technical correctness; our tool evaluates AEO suitability, which is an additional layer with different criteria.

Commercial AEO tools. HubSpot's AEO Grader, GEO Score, Snezzi and Profound are paid or freemium products that measure something different: how AI engines currently perceive you, not whether your site is technically prepared. Our tool is complementary, not a competitor: it makes sense to use ours to ensure the technical base, and then the commercial ones to measure the real result in AI engines. The two dimensions are needed: being technically ready without real presence in engines does not work, and appearing a lot in engines without a solid technical base is fragile and falls with any change.

Complete SEO crawlers. Screaming Frog, Sitebulb and similar cover deep SEO audits with an analysis of all the pages on the site. Our tool analyzes a single URL at a time, which makes it fast and useful for specific checks but less complete for massive audits. If you need to audit 500 pages, use a crawler; if you need to verify the AEO state of a key page in 30 seconds, this is the right tool.

CORS limitations and how we handle them

Some sites block access to their HTML from external browsers through CORS header configuration, a legitimate security practice that prevents tools like this from reading their content. When we detect a CORS block, we automatically try a public backup proxy. If both methods fail, the tool tells you honestly that it could not complete the analysis, without inventing results or giving false confidence.

If your own site fails repeatedly in the analysis, that is a sign that your CORS configuration may also be limiting legitimate AI engine bots. It is not a guarantee, but it frequently correlates with broader indexing problems. It is worth reviewing the CORS headers if this happens, or trying the tool with the site in public mode from an incognito window to verify that nothing is blocking external access.

Frequent errors we detect in Panamanian sites

After analyzing several dozen Panamanian sites with this tool and similar ones, there is a repeated pattern of errors worth mentioning so they serve as an informal checklist. They are not the only ones but they are the ones that appear most frequently and the ones that lower the score fastest.

Duplicate schemas with contradictory data. Sites that migrated from one content management system to another usually accumulate old and new schemas on the same page. The tool reads them all and, if the data does not match, the AI engine detects the conflict and lowers trust in the whole site. The recommended review is to make sure there is only one source of truth per entity: one Organization, one Article, etc., without duplicates with different fields.

FAQPage schema with empty or too-short answers. Many sites add FAQPage schema but the answers inside are loose one-line phrases. The AI cannot extract useful answers from that and prefers other sources. Effective answers have at least two or three sentences with concrete information, ideally with figures or verifiable references.

LocalBusiness without geo coordinates or openingHours. Local businesses declare LocalBusiness with name and address, but omit critical fields like geo (latitude/longitude), openingHoursSpecification, telephone with an international format. For local searches in AI engines, those fields are the difference between being cited as a verifiable nearby business or being ignored for insufficient information.

Article without an updated dateModified. Many blogs publish Article with a correct datePublished but never touch dateModified. Since AI engines reward freshness, having a dateModified that equals the datePublished and dated two years ago sends a signal of abandoned content. Any legitimate content review should update dateModified, and it is worth reviewing the pillar pages at least quarterly with this logic.

Schemas on pages but not on the home. Sites that added schemas on specific pages but forgot the home, where the AI engine usually enters first. The home must have Organization (and LocalBusiness if it applies) as a minimum base, without exceptions. Anything else on internal pages adds on top of that base, but without the base, the internal pages work in isolation.

The checker as a routine, not a one-off test

The most useful way to incorporate this tool into the workflow is not to use it once but to turn it into a light routine. We recommend three moments. First, before publishing important new pages: validate that the schema and the structure are right before exposing the page to the public and the engines. Second, after structural changes to the site: refactorings, theme updates, migrations. Any major change risks breaking schemas that were working. Third, as a quarterly check of pillar pages: the three to five pages that receive the most traffic deserve a periodic verification to detect regressions. The tool being free is precisely what allows making it a routine with no additional budget.