A SaaS company runs a free AI visibility checker and gets an 82 out of 100. Schema markup: present. Structured data: valid. Entity consistency: green across the board. Three months later, someone asks ChatGPT to recommend a tool in their category, and the brand never comes up. The score was real. The absence of citations was also real. Understanding how both can be true is the difference between fixing a checklist and fixing the actual problem.
Key Takeaways
- A free AI visibility checker tests retrieval presence, meaning schema validity, entity consistency, and crawlability, not whether a model already trusts or recalls the brand.
- Schema gaps are markup bugs fixable in an afternoon; entity gaps are trust fragmentation across independent sources and close far more slowly.
- Citations compound once they begin, so a passing audit score with no citation history has simply not entered that cycle yet.
What a Free AI Visibility Checker Actually Scans
Run a free checker on any domain and it will crawl a fixed set of things. It checks whether Organization, Product, and FAQPage schema exist and validate against schema.org definitions. It checks whether the same schema type appears consistently across pages describing the same entity. It checks for broken JSON-LD, missing sameAs links, and orphaned pages with no internal path back to the homepage.
None of this touches what a model has already learned about the brand. It touches what a model, or the retrieval layer sitting in front of it, can find and parse right now. That is a narrower and more mechanical question than most audit reports let on, and it is closer to what an AI Visibility Audit or Trust Signal Review is built to check in the first place.
Schema Gaps vs Entity Gaps: Two Different Problems
A schema gap is a markup problem. The Product schema is missing a price field, the FAQPage markup does not match the visible FAQ text, the Organization schema has no logo property. These are fixable in an afternoon by whoever owns the CMS.
An entity gap is a different kind of problem. It asks whether a model, cross-referencing multiple sources, can resolve “this brand” as one consistent thing across Wikipedia, Wikidata, Crunchbase, LinkedIn, and review sites. If the company name is spelled two ways, a founder is credited on one site and not another, or the Wikidata entry lists the wrong industry, the entity is fragmented even when every individual schema tag validates.
A checker can flag a broken sameAs link. It cannot tell you that three of your five sameAs targets disagree with each other about your headquarters. That disagreement is what erodes trust, and trust is a graph of agreement across independent sources, not a score a single audit can compute.
Why a Passing Score Still Misses Citations
Passing a schema and entity audit confirms one thing: a model that decides to look at your site right now will be able to parse it cleanly. It says nothing about whether ChatGPT, pulling from across the open web, has already encountered your brand often enough to recall it unprompted. It says nothing about whether Gemini’s cross-reference against Search, YouTube, and Scholar turns up enough independent coverage to treat you as a credible answer. Claude leans on high-authority publications and documentation, so a technically perfect product page on your own domain may not move its answer at all.
Citations also compound. Once a model cites a brand in one answer, the next citation becomes more likely, because the retrieval layer has already surfaced that source and found it reliable. A brand with a fresh audit score of 90 and zero citation history has not started that cycle. A brand with a messier score but eighteen months of accumulated citations has momentum the audit has no way of seeing.
growth in AI brand presence for NewDay USA, tracked weekly across five engines over six months.See the case study
That kind of movement only shows up when it is tracked over weeks and months on the engines that matter, not read off a single audit run once and filed away.
A passing score means the door is open. It does not mean anyone has walked through it.
Training Data Presence vs Retrieval Presence
AEO has two layers, and a free checker only ever tests one of them. Retrieval presence is earned technically: structured markup, current content, crawlable pages, clean entity links. It can improve within a single sprint, which is exactly why it is the easiest thing to package into a scored report.
Training data presence is earned slowly, through wide coverage across independent sources over months or years. It behaves more like reputation than a technical fix. A brand can have flawless retrieval presence and still be near-invisible in a model’s training data, because no one outside the brand’s own domain has written about it enough for the pattern to stick.
A checker score sits entirely on the retrieval side of that split. That is useful information. Treating it as the whole picture is how a team ends up staring at a green audit and a blank answer at the same time.
What to Do Next
Start by treating the checker’s findings as two separate to-do lists, not one score. Fix the schema bugs: missing fields, invalid JSON-LD, orphaned pages. That work is mechanical and worth doing on its own terms, because a model that cannot parse a page will not cite it no matter how strong the brand’s reputation is elsewhere.
Then open a second list for entity work, and expect it to move slower. Check whether the Wikidata entry, Crunchbase profile, and LinkedIn company page agree on the basics: name, industry, headquarters, founding date. Where they disagree, correct the outlier rather than the majority. This is the work that builds the trust graph a schema fix cannot touch, and it overlaps heavily with what a Digital PR or Entity Optimization effort is actually trying to accomplish.
Finally, stop measuring success with a single audit score. Track Share of Citation across ChatGPT, Gemini, Claude, Perplexity, and Grok on a recurring basis, and keep that number separate from Share of Voice in any report that reaches a founder or CMO. A brand can dominate mentions across the web and still register close to zero when a model is asked to name a source. The gap between those two numbers is usually the real story.
FAQs
What Does a Free AI Visibility Checker Actually Measure?
It measures retrieval presence: whether schema markup exists and validates, whether entity signals like sameAs links are consistent, and whether pages are structured and crawlable. It does not measure whether a model has already learned about the brand or cites it in practice.
Why Does a High Schema Score Not Guarantee Citations?
Citations depend on training data presence and on agreement across independent sources, both of which sit outside a schema audit’s scope. Citations also compound once they start, so a brand with no citation history has not entered that cycle regardless of how clean its markup is.
What Is the Difference Between a Schema Gap and an Entity Gap?
A schema gap is a markup problem, such as a missing field or invalid JSON-LD, and is usually fixable within hours. An entity gap is inconsistency in how a brand is identified across independent sources like Wikidata, Crunchbase, and review sites, and it takes longer to resolve because it depends on outside parties.
How Is Share of Citation Different From Share of Voice?
Share of Voice measures how often a brand is mentioned across the web overall. Share of Citation measures how often AI models pull that brand into an answer and name it as a source. A brand can have high Share of Voice and near-zero Share of Citation at the same time.
