A handful of free tools launched over the past few months with a simple pitch: type in your website, and in under a minute they'll tell you whether ChatGPT, Gemini, or Perplexity would ever recommend you. Names like AgentSiteScore, Skore, BulletProof's AEO checker, and a few others built specifically for real estate agents now sit alongside general-purpose AEO checkers anyone can run on any site.
They're worth running. Most take less time than making coffee, and none of the ones we looked at ask for a credit card. But the number they hand back only means something once you know what it's actually measuring. This post walks through what these tools check, why each part matters, and where the checks run out of things they can tell you.
What "AI visibility" actually means for a real estate agent
Being visible to AI search means an AI system, when asked a question a buyer or seller might type, can find your page, read it cleanly, and use it to construct part of its answer. That's a narrower thing than "ranking well on Google."
A page can rank fine in traditional search and still be invisible to AI tools, because the two systems don't work the same way. Google's classic ranking sends a person to a page to read it themselves. An AI answer engine has to extract facts from the page and restate them, sometimes without sending the reader to the page at all. A page built for a human scanning it top to bottom isn't automatically built for a machine trying to pull out your name, your service area, and your specialty as clean, quotable facts.
That's the gap these visibility tools are trying to measure. Most of them check three separate things, and it's worth knowing which one is failing when a score comes back low.
Check one: can an AI crawler actually reach your pages
Before an AI system can cite anything on your site, something has to have fetched and read the page. That happens through crawlers, and each major AI company runs its own with its own name and its own job.
OpenAI's GPTBot, for instance, gathers content to train ChatGPT's underlying model over time. That's a different job from OpenAI's search-specific crawler, which fetches a page in the moment to help ChatGPT answer a live question. Anthropic runs the same split for Claude: one crawler for training, a separate one, Claude-SearchBot, for live search answers. Blocking the training crawler doesn't stop your site from being cited in a live answer today. Blocking the search-specific crawler does, according to a 2026 crawler-optimization breakdown from Vemetric.
This is the single most common failure these visibility tools catch, and it's rarely intentional. A security plugin, a CDN's default bot-blocking rule, or an old robots.txt line copied from a template can quietly block every AI crawler by name without anyone realizing it. The fix is almost always the same: open your robots.txt file, check whether it names any AI crawler as disallowed, and remove that line if the block wasn't a deliberate choice.
Having a robots.txt file at all isn't the check that matters. What it allows or blocks is.
Check two: does your site hand over facts, or make an AI system guess
Once a crawler can reach a page, the next question is whether it can pull a clean fact out of it. This is where structured data comes in: a layer of code on the page, invisible to a visitor, that states facts in a format both search engines and AI systems can read directly instead of inferring from paragraph text.
For a real estate agent's site, the practical version of this is markup that states, in a machine-readable way, who you are, what area you serve, what kind of properties you specialize in, and how to reach you. Without it, an AI system has to infer those facts from prose, which it can sometimes do, but not reliably, and not as a fact it will state with confidence.
The same logic applies to FAQ content. A visible FAQ section with FAQPage structured data behind it gives an AI system a pre-packaged question-and-answer pair it can quote directly. A visible FAQ section without that markup, or with vague one-line answers that don't actually resolve the question, checks the box of "having an FAQ" without doing the job an AI system needs it to do.
Content quality and technical structure blend together here. A visibility tool can confirm the markup exists. It generally can't judge whether the answer inside that markup is specific enough to be worth quoting. That's a separate problem from whether the code is there at all.
Check three: does your name match everywhere else you appear
The third thing several of these tools check, and the one that surprises agents most, is consistency of your basic facts across the web, not just on your own site. Your name, brokerage, phone number, and service area on your website get compared against your Google Business Profile and, for some tools, against how you appear on portals like Zillow or Realtor.com.
This matters because AI systems, like search engines before them, treat agreement across independent sources as a confidence signal. When your website says one brokerage and your Google Business Profile says another, or your service area is "Greater Toronto" in one place and a specific list of five neighbourhoods in another, an AI system has less basis for stating either version as fact. A piece on Google Business Profile mistakes from nomorezerodays.org frames the same inconsistency problem: mismatched details across listings reduce confidence and, separately, cost trust with the people reading them directly.
Fixing this one is usually a matter of picking one version of your name, one phone number, and one description of your service area, then updating every place you appear to match it exactly, down to abbreviations like "St." versus "Street."
What a passing score does not tell you
Passing all three checks means the door is open. It does not mean an AI tool has decided to walk through it and recommend you specifically.
Whether an AI system actually names you in an answer depends on a fourth thing none of these audit tools can fully measure: how your content compares to everyone else's content answering the same question. If a buyer asks an AI tool to recommend an agent for condos in a specific neighbourhood, and three other agents in that area have pages that name the neighbourhood, the price range, and the property type directly, while yours says "serving the greater metro area," the technical checks might all pass on your site and it still won't be the one that gets named.
That gap explains why a brand-new site with almost no content can still pass a technical AI visibility check. Crawlable and well-structured is real and necessary. It isn't the same as having enough substance to be worth quoting over a competitor's page.
Reading a low score without overreacting
A low AI visibility score is useful information, not a verdict on the whole website. The right response is to find out which of the three checks failed, not to assume the entire site needs to be rebuilt.
If the failure is crawler access, that's typically a single settings fix: a blocked user agent in robots.txt or a firewall rule catching bots it shouldn't. If the failure is structured data, that's a markup problem that can often be added to an existing site without touching its design. If the failure is consistency across listings, that's an audit-and-update task across a handful of profiles, not a rebuild.
Where a full rebuild starts to make sense is when the underlying pages themselves don't say anything specific enough to be worth citing in the first place, no matter how clean the crawler access or the markup is. At that point the problem isn't technical visibility. It's that the content has nothing distinct to extract.
The takeaway
These free scoring tools are a fast, no-cost way to find out which of three separate things might be holding an agent's site back from AI search: crawler access, structured data, and cross-platform consistency. Run one, read past the headline number to the specific checks it lists, and fix what's actually broken before assuming the whole site is the problem.
If the checks all come back clean and an AI tool still never mentions you for the questions that matter in your market, that's a content and competitiveness question, not a technical one, and it's worth reading through our rebuild readiness checklist or looking at what a rebuild can and can't fix before deciding what to do next. Saige's own free rebuild preview works differently from a scoring tool: instead of grading your current site, it generates a rebuilt version of it, structured for both search engines and AI systems from the start, so you can see the fix rather than just the diagnosis. If you have questions about how that works, the FAQ page covers the mechanics in more detail.
