Schema markup helps a machine identify your business correctly, and identifying you correctly is the whole of what it does. It labels which words on your page are the business name, the address, the license, the person. It does not tell an assistant you are good at the work. Google documents this territory more thoroughly than anyone else does, and says outright that "structured data isn't required for generative AI search, and there's no special schema.org markup you need to add."
Picture a conference name tag. A good one is accurate, legible from four feet away, and spelled the way your license is spelled. It saves everybody in the room a moment of confusion about who you are. It has never once walked a stranger across the carpet and said, this is the person you should hire. That introduction is a different job.
That gap is where the frustration usually sits. Somebody sold you schema, or your agency added it and reported it as AI work, and the validator came back green. Nothing was faked. The markup is genuinely useful and it is genuinely worth having. It was just never the thing standing between you and being named, so the silence afterward feels like a bill for a fix that didn't fix anything.
Here is the plain version. It covers what structured data actually is, what Google documents it for and what Google refuses to promise, what OpenAI does and does not publish about ChatGPT, when a plugin is genuinely enough, and where this belongs in the order of work.
What Schema Markup Actually Does
Strip the vocabulary away and structured data is a filing system. Google's own introduction calls it "a standardized format for providing information about a page and classifying the page content." You add it, in Google's words, to give "explicit clues about the meaning of a page." Ordinary web code tells a browser how to display your address. It says nothing about that text being an address. Schema is the note in the margin that says so, written in the shared vocabulary published at schema.org. Google recommends the JSON-LD format. It sits in the page as a small block of code your visitors never see.
Two types do most of the work for a professional firm, and both are plainer than they sound.
- Organization describes the business itself: name, logo, URL, address, contact point, and links out to the profiles that confirm you elsewhere. Google says it "can help Google better understand your organization's administrative details and disambiguate your organization in search results." It belongs on "your home page, or a single page that describes your organization, for example the about us page." There are no required properties. You add the ones that are true.
- LocalBusiness describes a place you actually operate from. Schema.org defines it as "a particular physical business or branch of an organization," which is why it inherits from both Organization and Place. Google asks for name and address as the required properties, says to "define each local business location as a LocalBusiness type," and to "use the most specific LocalBusiness sub-type possible."
Notice what none of that is. There's no field for judgment, no property for twenty years of hard cases, no attribute that means trustworthy. Structured data can describe a great deal about a business, from hours to price range to who to call. What it cannot do is establish that any of it is any good. It files you accurately. It does not vouch for you.
What Google Puts in Writing
Google publishes far more detail on this than any other engine, so read it directly rather than through whoever is quoting it at you. It is unusually blunt about the ceiling.
On its generative features, Google says: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." Its separate page on AI features goes broader. "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." And: "You don't need to create new machine readable files, AI text files, or markup to appear in these features." Google still recommends structured data, but for a stated reason. It "helps with being eligible for rich results on Google Search." The baseline it does name is ordinary: a page that is indexed and eligible to appear with a snippet. Markup is a search-result enhancement, not a route into an AI answer.
Then there is the sentence that should end every conversation about guaranteed outcomes. Google's structured data guidelines say: "Using structured data enables a feature to be present, it does not guarantee that it will be present." And more directly: "Google does not guarantee that your structured data will show up in search results, even if your page is marked up correctly according to the Rich Results Test." Valid is not the same as visible.
One more rule matters more than it looks. Google requires that "your structured data must be a true representation of the page content." It also tells you not to "mark up content that is not visible to readers of the page." Markup is a description of the page. A description that argues with the page it describes is worse than no description at all, and this is the rule your agency's green validator does not check.
The Honest Answer About ChatGPT
Now the question in the title, answered without the hand waving. There is no published OpenAI document establishing schema.org markup as an input to how ChatGPT finds, describes or recommends a business. What OpenAI does publish is a page about its crawlers. That page is the closest thing to an official pathway anyone outside the company has.
Three of the agents it names matter here, and they do three different jobs. OAI-SearchBot "is used to surface websites in search results in ChatGPT's search features," which makes crawl access the documented lever. GPTBot is separate and is about training: "Disallowing GPTBot indicates a site's content should not be used in training generative AI foundation models." ChatGPT-User handles actions a person triggers inside a chat, and OpenAI states plainly that it "is not used to determine whether content may appear in Search." OpenAI also notes that after a robots.txt change, "it can take ~24 hours from a site's robots.txt update for our systems to adjust." Nowhere on that page does the word schema appear.
Read that carefully, because an absence on one page is evidence of a limit rather than proof of a negative. It does not mean markup is invisible to every system. It means nobody at OpenAI has published a claim that adding Organization JSON-LD does anything in particular for you. Anyone selling you that claim is selling their own inference as your outcome. Two questions are worth asking about ChatGPT. The first is whether your site permits the right crawler. The second is whether what it finds when it arrives says anything specific about you. Your own website is only one of the surfaces it checks, which is a separate problem again.
When a Plugin Is Enough, and When It Goes Wrong
In our experience a plugin is genuinely enough for most firms, and that's a relief rather than a compromise. Organization markup on one page and one accurate LocalBusiness entry per real location is the entire job for a single-office professional practice. It's close to a one-time task with occasional maintenance. Any proposal to bill it monthly deserves one question: what changed this month?
The failures we run into are rarely missing markup. They are markup that contradicts something.
- Two plugins, two versions of you. An SEO plugin and a dedicated schema plugin can both publish an Organization block on the same page, and the fields do not always agree. The site now hands a machine two answers to the question of who you are, which is the opposite of what the exercise was for.
- Markup that outran the page. The code claims services or locations the visible page never mentions. Google's rule is explicit, and the correction is to fix the page rather than delete the label.
- The generic type where a specific one exists. A dental practice filed as a plain LocalBusiness is technically valid and needlessly vague, when Google asks for the most specific subtype available.
- Details that quietly went stale. The suite number from two moves ago, the partner who left, the old phone. Every one of them becomes a disagreement between your site and every profile that has it right.
The first check is short and needs no code. Open your home page and your About page in Google's Rich Results Test, then read the fields it reports back as if you were a stranger. If a name, address, phone number or credential in there disagrees with what is on the page, your directory listings or your license record, you've found real work. That is a coherence problem rather than a code problem, and coherence is the subject of our plain-English guide to how AI decides your business is real.
Where Schema Belongs in the Order of Work
Sequence is the part almost nobody sells you. It's also the part that saves money. Our order of work puts schema after your identity agrees with itself everywhere a machine looks, and after your pages say something specific enough to be worth labeling. Label a page that doesn't answer a buyer's question and you get a well-filed page that still helps nobody. Getting the content on the page right is the prerequisite, not the follow-up.
It helps to know what the visibility gap is usually made of. Owners tend to assume it is technical. In one large vendor study (Ranqo, 100,000+ prompt responses across 100+ brands, March to May 2026), household names appeared in 73 percent of relevant unbranded AI answers, mid-market brands in 44 percent, and small or niche brands in 11 percent. That is one vendor's sample and should not be stretched past it. Read it as a ladder rather than a verdict. Our read of it is that the useful move is to be specific enough to own the questions the household names never bother to answer, and that is a strategy problem your markup was never going to solve.
The way we hold this at Probably Genius: AI recommends businesses it can verify (foundation), that others vouch for (validation), and that know something it doesn't (information gain). Schema sits inside the first of those three. It is a small, worthwhile piece of it. If your agency has the schema handled, that is a real deliverable. Keep it. The next questions are what exists about you off your own site, and what your pages know that nobody else's do. Those are the ones worth bringing to your next agency review. Asking them is scope work, not suspicion.
Want to Learn More?
Probably Genius builds and operates the recommendation layer around expert-led businesses, then measures it in the open. That is the work inside The Answer: twelve done-for-you thought-leadership articles a month, planned to produce approximately 60 clear Knowledge Entries, your identity made coherent everywhere machines check, corroboration beyond your own pages, and a monthly probe across five engines. Every piece is scored 0 to 100 through the Integrity Gate, where nothing publishes under 80. The published methodologies show how it runs, including Entity Identity, which is where structured data actually lives in our work.
You're probably a genius at what you do. We make sure AI gets the memo.