To get cited by ChatGPT and Google AI Overviews, your pages must first be reachable by each assistant's search crawler, then answer real questions directly, in plain text, with specific facts. There is no paid shortcut and no special markup. AI assistants quote pages they can find, read and trust, so the work is making your site all three.
This guide covers what each assistant actually requires, the six changes that make the biggest difference, and how to check whether it is working.
Key takeaways
- Each assistant uses its own crawler. Block it in robots.txt and you cannot be cited there, however good your content is.
- Google says AI Overviews need no extra markup or files beyond normal SEO fundamentals.
- Put the answer in the first sentence of each section. Assistants lift direct answers, not conclusions buried in paragraph three.
- Facts locked in images, PDFs or flipbooks are invisible. Important content has to be text.
- Test your target questions in each assistant every month, the same way you track rankings.
How do ChatGPT and Google AI Overviews choose which sources to cite?
Both work the same basic way: when someone asks a question, the assistant searches an index of web pages, reads the most relevant results, writes an answer and links the pages it drew from. If your page is not in that index, it cannot be chosen.
Where the index comes from differs by assistant:
| Assistant | Where it finds pages | What it needs from you |
|---|---|---|
| Google AI Overviews and AI Mode | Google's normal search index | Indexed and eligible for a snippet in Google Search |
| ChatGPT search | OpenAI's search crawler, OAI-SearchBot | OAI-SearchBot allowed in robots.txt |
| Perplexity | Perplexity's crawler, PerplexityBot | PerplexityBot allowed in robots.txt |
As of November 2026, Google states in its guidance on AI features that there are no additional requirements to appear in AI Overviews or AI Mode. A page must be indexed and eligible to show with a snippet, and that is all. It also says you do not need new AI text files or special structured data. Meeting the requirements does not guarantee inclusion, but failing them guarantees exclusion.
If you want the bigger picture on how these engines differ from a classic results page, read our analysis of whether AI search engines can replace Google.
Step 1: Let AI search crawlers reach your site
This is the most common reason a business never appears in ChatGPT answers, and the easiest to fix. Some sites block every AI bot to keep their content out of model training. In doing so, they also block the crawlers that power search answers.
The two are separate. As of November 2026, OpenAI's crawler documentation lists OAI-SearchBot for ChatGPT search and GPTBot for training foundation models. Each robots.txt setting is independent, so you can block training and still appear in search. Perplexity's crawler documentation describes PerplexityBot the same way: it surfaces and links sites in Perplexity's results, and Perplexity says it is not used to crawl content for foundation models.
A robots.txt that allows search answers but declines model training looks like this:
User-agent: OAI-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: GPTBot
Disallow: /
User-agent: Googlebot
Allow: /
Sitemap: https://www.example.com/sitemap.xml
Two things to check beyond the file itself:
- Your firewall or CDN. Bot protection at the hosting or CDN level can block a crawler that robots.txt allows. Both OpenAI and Perplexity publish IP ranges for their crawlers so you can allow them explicitly.
- Timing. Both OpenAI and Perplexity say a robots.txt change can take up to about 24 hours to take effect.
Whether to allow training crawlers is a business decision, not an SEO one. Blocking GPTBot does not affect ChatGPT search.
Step 2: Answer the question in the first sentence
Assistants extract answers. A section that opens with background and reaches the answer three paragraphs later gives them nothing clean to quote.
Write each section so its first one to three sentences answer the heading completely, then add depth underneath:
- Use the question as the heading. "How long does delivery take to Oman?" matches how people phrase questions to an assistant.
- Lead with the answer. "Delivery to Muscat takes three to five working days" is quotable. "Many factors affect delivery times" is not.
- Keep the answer self-contained. It should still make sense when lifted out of the page with no context around it.
The same structure helps featured snippets in classic search, so nothing is traded off.
Step 3: Put your important facts in text
An assistant can only cite what it can read. Content inside images, scanned PDFs, embedded flipbooks, sliders or text that loads only after a click is often invisible to crawlers.
Google's AI features guidance lists keeping important content available in text form among the fundamentals. In practice, the gaps we see most often are:
- Product catalogues published as a flipbook or a single PDF, so no individual product has a page
- Prices, service areas or opening hours shown only inside an image
- FAQs hidden in accordions that load their content with a script
If a fact matters to a buying decision, it needs its own text on a crawlable page.
Step 4: Use structured data that matches the page
Structured data (schema markup) is code that labels what a page contains: an organisation, a product, an article, a set of questions and answers. It helps any machine read your page without guessing.
Be clear about what it does and does not do. As of November 2026, Google says no special schema is needed for AI Overviews, and that structured data should match the visible text on the page. Schema does not earn a citation by itself. It removes ambiguity on a page that already deserves one.
For most businesses, the useful set is:
| Schema type | Where it goes | What it clarifies |
|---|---|---|
| Organization | Homepage, once | Your official name, website and logo |
| LocalBusiness | Location pages | Address, phone and hours for each branch |
| Article | Blog posts and guides | Headline, author and publication date |
| FAQPage | Pages with a question-and-answer section | The exact questions and answers on the page |
| Product | Product pages | Name, specifications and availability |
Never mark up something the visitor cannot see, and never put a rating in schema you cannot verify that day.
Step 5: Be specific, consistent and checkable
Vague claims give an assistant nothing to repeat. Specific, verifiable facts do. A line like "ISO 9001 certified, delivering to 12 countries" can be quoted. "Industry-leading quality" cannot.
Consistency matters as much as specificity. If your website, Google Business Profile and directory listings give different phone numbers, addresses or opening hours, you give every system a reason to doubt all of them. Google's AI features guidance specifically mentions keeping Business Profile information up to date. Our Google Business Profile service covers keeping those details aligned across every listing.
Every number you publish also has to be true on the day it is written. A figure an assistant repeats and a customer then disproves does more damage than no figure at all.
Step 6: Give other sites a reason to mention you
Assistants form a picture of a business from more than its own website. Reviews, industry directories, trade press and genuine editorial mentions all describe you in other people's words.
You cannot control those sources, but you can earn them: publish original data, contribute expert answers to trade publications, and keep directory listings accurate. This is the same authority-building that helps classic rankings, covered in our guide to domain authority and SEO. Avoid paid link schemes. They put your organic visibility at risk and are exactly the kind of signal any trust-based system is built to discount.
How do you check whether AI assistants cite you?
Test it directly, every month, with the questions your customers actually ask.
- List 5 to 10 target questions, phrased the way a real customer would ask them, such as "Which accountants in Lisbon handle company formation for foreign founders?"
- Ask each question in ChatGPT, Perplexity and Google, using a private or logged-out session where possible.
- Record the result for each one: cited with a link, mentioned without a link, or absent.
- Note which competitors are cited instead, and look at what their pages do that yours do not.
- Repeat on the same date each month and track the trend.
Answers vary between sessions, so treat one result as a sample, not a verdict. The trend across months is what tells you whether the work is landing. Google also counts traffic from AI Overviews and AI Mode inside the Search Console Performance report under the Web search type, as of November 2026, so you will not see a separate line for it there.
Frequently Asked Questions
Can I pay to have my business recommended by ChatGPT?
No. ChatGPT's search answers cite pages its search crawler can reach and judges relevant to the question. Visibility comes from being crawlable, answering questions directly and being described consistently across the web. Anyone selling guaranteed placement inside AI answers is promising something they do not control, so treat that offer with caution.
Does schema markup help me appear in Google AI Overviews?
Not on its own. As of November 2026, Google says AI Overviews need no special schema beyond normal Search requirements: the page must be indexed and eligible to appear with a snippet. Structured data still helps machines read your page accurately, as long as it matches the visible content. Think of it as removing doubt, not as a ranking switch.
Will blocking GPTBot stop my site appearing in ChatGPT?
No. OpenAI uses separate crawlers for separate jobs. GPTBot collects content that may be used to train its models, while OAI-SearchBot powers ChatGPT search. Each setting in robots.txt works independently, so you can block GPTBot and still allow OAI-SearchBot. If OAI-SearchBot is blocked, OpenAI says your pages will not be shown in ChatGPT search answers.
Do I need an llms.txt file to get cited by AI assistants?
Not for Google. As of November 2026, Google's guidance says you do not need to create new machine-readable files or AI text files to appear in AI Overviews or AI Mode. Your time is better spent on the fundamentals: crawler access, text-based content, direct answers and consistent business information. Those help every assistant at once.
How long does it take to start appearing in AI answers?
There is no fixed timeline. Crawler access changes can take effect within about a day, according to OpenAI and Perplexity. Being cited depends on your content, your competitors and how much other sites trust you, and none of that moves on a schedule. Track your target questions monthly and judge progress by the trend, not by a single test.
Is answer engine optimisation different from SEO?
They overlap heavily but are not the same. Both depend on a crawlable, well-structured site. Answer engine optimisation adds a focus on direct, self-contained answers, specific facts and consistent information across every source, because an assistant quotes and summarises rather than listing ten links. A site built for both is easier to rank and easier to cite.
Next step
If you want to know where your business stands today, start with the questions your customers ask and check who gets cited. Our answer engine optimisation service runs that baseline, fixes what blocks citation, and tracks AI visibility every month alongside your rankings. You can also contact our team with the questions you most want to be found for.
