Glotier tells you whether ChatGPT, Gemini and Perplexity name your product when buyers ask what to use in your category. It puts your buyers' real questions to all three as live web queries, records whether you were named and by which of them, which competitors were named instead, and which pages each answer was built from.

How does Glotier check AI visibility?

It runs one live web search per question. ChatGPT and Gemini read the identical results, and Perplexity runs its own retrieval because it does not accept results handed to it. Whether your product is named is decided by a literal presence check on the answer text rather than by asking a model for its opinion, so the number cannot be inflated by a model being agreeable. You can also connect your own ChatGPT or Claude key and have the questions run through your account instead.

Why ask three assistants instead of one?

Because they disagree. In our own study, across the 39 of 40 buying questions where all three answered, all three named the same product only 30% of the time and 47% of products were named by exactly one assistant. Checking a single assistant shows you 58% to 65% of the products named across the three, so a founder who checks ChatGPT, sees their name and relaxes has measured one engine on one phrasing.

Is Glotier free to use?

The check is free with no signup. Paid plans start at $9.99 a month (Fast, one product), then Solo at $39, Studio at $99 and Agency at $249. Every paid plan covers the daily check and the content the agent writes.

Does Glotier make up data about my product?

No. Every result comes from observed data: the live search results, the answers the assistants actually returned, and the sources they cited. It does not invent metrics, ratings or engine claims, and when a measurement has not run it says so rather than printing a zero.

Anyone selling something buyers research before choosing, where the category has enough alternatives that an assistant has to pick a shortlist: software, tools, e-commerce, and services. Founders, marketers and agencies all use it, and the work is the same at any size. It is not useful for local or in-person businesses, where maps and reviews decide the outcome instead.

Method · 2026

How to get cited by ChatGPT

When ChatGPT answers with search on, it reads a small set of live pages and names some of them as sources. Getting cited is being one of those named pages. It is two problems in sequence: being in the set it retrieves, then being the clearest thing to quote once you are there. Everything below is the method for both, and every number in it is one we measured on our own pages rather than a rule of thumb.

First, three words that get used as if they mean the same thing and do not. Retrieved is a page the model pulled and read. Cited is a page it named as a source, the smaller visible set. Recommended is your product named in the answer itself. This page is about the middle one, citation, and it is where most of the winnable work is.

1. Be retrievable, or nothing else matters

An assistant can only cite a page it can fetch. That means two things: the page has to rank for the question a buyer asked, because the assistant searches the same web you do, and it has to let the AI search crawlers in. Most sites that block AI do it by accident, with a broad rule that catches the search crawler along with the training one. The numbers below are from every domain an assistant actually cited across our own checks.

What we measured, 31 July 2026

Of 259 domains AI cited in the AI visibility category on 31 July 2026, 235 allowed every AI crawler.
18 of 244 of those sites block GPTBot but allow OAI-SearchBot, leaving ChatGPT search while opting out of training.
Only 5 of 259 cited domains block an AI search crawler by name, and all four are large publishers.

Sample: every domain an assistant actually cited across our own checks, not a top-100 list. Blocked means a whole-site disallow for the matching group, so partial rules are not counted and these figures understate blocking. 0 domain did not answer and is excluded rather than counted as open. How we measure

2. Own one sub-question, do not spread thin

An AI answer is not a ranked results page. Across 13 of our own Google AI answers carrying 168 citations, no domain was ever cited twice in the same answer: 13 of 13 answers used a different domain for every source. An assistant gathers one best page per sub-question from wherever it lives, so you get one slot at most. The way to appear more often is to be the best page for more distinct sub-questions, not to pile more pages onto one broad term. That is the opposite of classic SEO, and it is why ten thin pages on the same keyword do nothing here.

3. Be where ChatGPT already reads

When we measured which sources AI built its buying answers from, community threads beat review directories by about 21 to one, and Reddit alone was a source for 57% of the answers in our own category. A genuine, helpful presence in the threads where your buyers already ask is worth more than another listing on a directory the assistants rarely quote. Earn it honestly: answer real questions, do not spam your link.

4. Write sentences an assistant can lift whole

The unit that gets quoted is a sentence, not a page. Anthropic documents its citation payload as up to 150 characters of your text, and OpenAI anchors each citation to a span of the answer. A claim spread across three paragraphs that depend on each other has nothing that survives extraction. So every claim worth citing should live in one self-contained sentence that names its own subject and carries its own number. Write for the quote, not just the read.

5. Keep it fresh, because sources churn

A page cited today is not cited forever. Across 20 of our own questions asked on more than one day, only 40.9% of the sources present on the first day were still there on the last, eleven days later. Fewer than half survived, and that is measured at the domain level, so the real churn is higher. A page you wrote once and never touched slides out of the answer set. Update the ones that matter on a schedule.

6. Measure across all three, not one

ChatGPT is one assistant, and it shows you part of the picture. When we asked the same buying questions of all three, a single assistant would have shown you 58% to 65% of the products the full panel named. Checking only ChatGPT means missing a third or more of who your buyers are being pointed to. Measure the same frozen questions on a schedule, on all three, so a change is a signal and not a reworded prompt.

The proof, on our own site

This is not theory we sell without eating. In one week our own page beacon counted 6 real human visitors sent by an AI answer, 5 of them to pages built from our own measurements and 0 to a page that sells. The assistants sent people to the pages that carried a number worth quoting, not to the pages that carried an offer. Publish something worth quoting, on a page an assistant can quote in one sentence, and the citations follow.

What does not work

An llms.txt file on its own. It lists your pages; it does not make any one of them the clearest answer to a question, which is what gets cited.
Keyword stuffing. The assistant reads for meaning and quotes a sentence. A page dense with the phrase and thin on a real, specific claim gives it nothing to lift.
A visibility score with no sources. A number you cannot trace to the answers and pages behind it tells you nothing you can act on.

See whether ChatGPT cites you right now

Run the free check on any site: it asks ChatGPT, Gemini and Perplexity your buyers’ real questions and shows whether you are named and which pages each answer cited, in about a minute, no signup. When you want it watched over time, Glotier starts at $9.99 a month for one product, with no sales call.

Cancel any time. Not for you? Email us within 7 days of a charge and we refund it in full.

Run the free check →Compare GEO tools

Common questions

What does it mean to be cited by ChatGPT?

When ChatGPT answers with search on, it fetches a handful of live pages and names some of them as sources with links. Being cited means your page is one of those named sources, not just somewhere in its training data.

Is being cited the same as being recommended?

No. Cited means your URL is listed as a source. Recommended means your product is named as an answer. You can be cited without being recommended, and the fix for each is different.

Does an llms.txt file get me cited?

On its own, no. Assistants cite pages they retrieved for a specific question and could quote. A file that lists your pages does not make any single page the clearest answer to a buyer's question.

How do I know if ChatGPT cites me?

Ask it your buyers' real questions and read the sources it returns. Glotier runs this as a free check across ChatGPT, Gemini and Perplexity and shows the source list for each answer.