How to Get Cited by ChatGPT and Perplexity

I get asked how to get cited by ChatGPT at least once a week, and the honest answer disappoints people who were hoping for a trick. There is no hidden markup, no secret file and no separate discipline you can bolt onto a website. The systems read the same web your customers read. They just extract and reuse passages differently, which turns the work into accessibility, clarity and corroboration.

That is good news, because all three are things you can actually check this week.

What Being Cited by AI Answers Actually Means

A citation is a link to your page inside an answer that a machine has written. It is not the same as ranking, and it is not the same as traffic, although the two often move together.

Citation vs Ranking

Ranking is a position in a list of results. A citation is you being named as the source of a fact inside a generated answer. A page can rank first and never be cited, and a page sitting in position eight can be the one the answer quotes, if that page contains the clearest statement of the fact being asked about.

Why AI Answers Cite Some Sources and Not Others

The systems tend to reuse passages that make a clear, self contained, checkable claim. A paragraph that names the thing, gives the number and states the source is far more reusable than a paragraph that talks around the subject. If your best answer is spread across four sections and one image, there is nothing clean to lift.

Make Sure AI Crawlers Can Reach You

This is the most common reason a business is absent from AI answers, and it is usually accidental. A robots.txt rule added years ago, or a firewall setting, silently blocks the crawler that feeds the answer engine.

OAI-SearchBot, GPTBot and How They Differ

OpenAI's crawlers do different jobs and they are controlled separately, which matters more than most people realise. OpenAI states that "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links." It also states in the same crawler documentation that "Disallowing GPTBot indicates a site's content should not be used in training generative AI foundation models."

Read those two together and the choice becomes deliberate. Blocking GPTBot is a decision about training data. Blocking OAI-SearchBot is a decision about whether you exist in ChatGPT search answers at all. Many businesses want to be findable while restricting training, which is only possible if the two are treated as separate decisions instead of one blanket block.

Checking robots.txt and CDN Rules

Open your robots.txt and search for the crawler names. Then check the things people forget: firewall and CDN rules that block unusual user agents, bot protection services set to challenge everything, and JavaScript only rendering that leaves your content invisible to a crawler that does not execute scripts. Make sure crawlers can reach your content before you spend a minute on anything else, because nothing downstream matters if the page is never fetched.

The SEO Fundamentals Are Still the Entry Requirement

AI visibility has been sold as a replacement for SEO. It is not. It is a subset of the same work, and Google says so directly.

Indexed, Eligible and Snippetable

Google Search Central sets the bar plainly: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements." Indexed, eligible, snippetable. If a page is not indexed, nothing else you do will put it in an answer.

Why There Is No Special AI Markup to Add

There is no schema tag, file or markup that gets you into AI answers. Google states that "The best practices for SEO remain relevant for AI features in Google Search (such as AI Overviews and AI Mode). There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." If a vendor is selling you an AI optimisation markup package, ask them to point at the documentation that requires it.

Write Passages That Stand On Their Own

Extraction is the practical difference between AI visibility and classic SEO. A machine may lift one paragraph out of your page and reuse it somewhere you cannot control, so the paragraph has to survive on its own.

Answer First, Then Explain

Put the answer in the first sentence of the section, then explain. A section that opens with three sentences of context before reaching the point is much harder to reuse, because the reusable part is buried and the opening is unusable on its own.

Definitions, Steps and Comparisons

Name the thing you are defining. A weak definition says it is a method used to improve performance. A strong one says account based marketing is the practice of choosing a named set of companies you want as customers and building activity around reaching the people inside them. Steps work the same way: number them, state the outcome first. Comparisons should name the criteria being compared and the trade off, not just list options.

Structure That Helps Machines Extract

Structure is not decoration. It is how a machine works out which passage answers which question.

Headings, Tables and Lists

Use one heading per idea, phrased as the question a person would ask, then answer it beneath. Use tables where the information is genuinely comparable and lists where the information is genuinely a sequence. A table of options with their trade offs is more reusable than the same comparison written as three paragraphs.

Keep the Facts in Text, Not in Images

Numbers inside images are effectively invisible to extraction. If a key figure, a price range or a specification only exists in a graphic, move it into the body text as well. You can keep the graphic, but the fact needs to be written down.

Build Entity Clarity and Trust Signals

An answer engine has to be confident about who you are before it will name you. Ambiguity is expensive here, because a system that cannot tell your business apart from a similarly named one has a reason not to cite either.

Clear Author and Organisation Details

State who wrote the page, what they do and what qualifies them, in the page itself and not only in the footer. Keep the organisation details consistent: the same business name, the same description, the same service list, the same location, wherever they appear.

Consistent Facts Across the Web

Check that your business name, address, phone number, service descriptions and founding details match across your website, your directory listings, your social profiles and any industry listings. The same fact stated consistently in more places is easier to verify, and verifiable facts are what get used.

Earn the Mentions That Make You Citable

Being quotable is a content problem, not a technical one. If everything on your site already exists in the same form on five hundred other sites, there is no reason to cite you specifically.

Original Data and First Hand Experience

The material that gets cited is the material nobody else has: your own survey results, your own benchmarks, your own process documentation, the specific numbers from work you have done, and the practical detail that only comes from doing the job. If you have first hand data, publish it with the method and the sample size attached, because sourced numbers are more reusable than unsourced ones.

Third Party Corroboration

Mentions on other credible sites matter, and not for the link in the traditional sense. A claim that appears in one place is a claim. The same claim appearing in several independent places is closer to a fact, which is what a generative system prefers. The encouraging part is that you do not need a huge link profile for any of this: Ahrefs found that "Content requires just 13 referring domains to rank in an AIO SERP on average."

What Does Not Work

A lot of money is currently being spent on approaches with no published evidence behind them.

llms.txt and Other Unproven Shortcuts

Google's position is unambiguous. "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add." A text file convention like llms.txt is a proposal, not a requirement of any major AI search system, and its practical effect on citation is unproven. Treat it as an experiment at most. If you want to understand what it is, what llms.txt is and why it is not required covers the detail, but do not make it a priority ahead of crawler access and clear writing.

Keyword Stuffing and Thin Pages

Stuffing keywords does not help here any more than it helps in normal search. Neither does publishing quantity over substance, because a hundred thin pages give a system nothing quotable. Extraction rewards clarity and specificity, and neither can be faked at volume.

Measuring Whether It Is Working

There is no reliable dashboard that reports your AI citations, so measurement is manual and approximate. Do it anyway, because the alternative is guessing.

Tracking Mentions and Referral Traffic

Run your priority questions through the answer engines monthly and note if you are mentioned and if you are linked. Track referral traffic from AI domains in your analytics, and treat it as a minimum rather than a total, because many AI referrals arrive unattributed. Track branded search demand as well, since being named in an answer increases the number of people who search for you directly.

What to Compare Month to Month

Compare the same questions, the same way, on the same day of each month: which of your pages are cited, how many, and whether the pages being cited are the ones you want representing you. Growth here is slow and uneven, which is exactly why a consistent monthly record matters more than a single reading.

A Practical 30 Day Starting Plan

Week one is access. Check robots.txt and your CDN or firewall rules for every AI crawler you care about, decide deliberately on OAI-SearchBot and GPTBot, and confirm your key pages are indexed and eligible to show a snippet.

Week two is extraction. Take your three most important answer worthy pages and rewrite the opening section of each so the answer is in the first sentence and each passage makes sense lifted out of context.

Week three is structure and entity. Give every section a heading phrased as a real question, move any key facts out of images and into text, and make your author and organisation details consistent across your site and your main external listings.

Week four is measurement and assets. Start the monthly citation check, and identify one piece of original data or first hand experience you could publish with a method note attached.

The audience for this is real and it is growing. Pew Research Center found that "Today, 34% of U.S. adults say they have ever used ChatGPT, according to a Pew Research Center survey. That includes a 58% majority of adults under 30." Those users are asking the questions your service answers. AI visibility and generative engine optimisation is the discipline of being findable in those answers, and read the AI search visibility guide if you want the wider picture before you start. I can check today whether AI crawlers are being blocked on your site and whether your key answers are extractable.

Frequently Asked Questions

How do I get my website cited by ChatGPT?

Start with access, then clarity. Confirm that ChatGPT's search crawler, OAI-SearchBot, is allowed in your robots.txt, because OpenAI states that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers. Then make the content easy to reuse: indexable, in plain text, structured with clear headings, and written so a single passage makes sense on its own.

Do I need an llms.txt file to appear in AI answers?

No. Google states directly that you do not need to create new machine readable files or AI text files to appear in AI Overviews or AI Mode, and that there is no special schema markup to add. A text file like llms.txt is a proposed convention, not a requirement of any major AI search system, so treat it as an experiment rather than a priority.

Is AI search optimisation different from SEO?

The fundamentals are the same and Google says so explicitly: the best practices for SEO remain relevant for AI features, and there are no additional requirements to appear in AI Overviews or AI Mode. What changes is emphasis. Being extractable, factual and unambiguous matters more, because a machine may reuse one passage out of context rather than sending a user to read the whole page.

How long does it take to see AI citations?

There is no published timeline, and any agency quoting a fixed one is guessing. The practical sequence is to fix crawler access, fix extractability, then measure mentions and referral traffic month over month. Crawling itself varies widely, and Google notes that recrawling can take anywhere from several days to several months depending on how often a page needs refreshing.

Should I block GPTBot?

That is a business decision about training data, not a search visibility decision. Disallowing GPTBot signals that your content should not be used to train generative AI foundation models. Disallowing OAI-SearchBot is different: it removes you from ChatGPT search answers. Many businesses want to be in search results while restricting training, which is why the two need to be treated separately.

Are AI answers worth the effort for an Australian small business?

The audience is real and growing. Pew Research found 34% of U.S. adults had used ChatGPT and 58% of adults under 30 had, and AI features now sit inside mainstream search results. The work involved overlaps heavily with good SEO and clear writing, so the effort is rarely wasted even where citations take time to appear.