How to get cited by ChatGPT: a page-level playbook for B2B SaaS

To get cited by ChatGPT, let OAI-SearchBot crawl your site, publish facts only you can state (pricing, specs, original data) on plain, crawlable pages, and open each section with a short, self-contained answer it can lift. Ahrefs' October 2026 study found ChatGPT leaning toward official, original sources. This guide turns that into a page-by-page checklist and a way to measure it.

Citation vs mention: know which one you are chasing

When ChatGPT answers with search turned on, two different things can happen to your brand. It can mention you, naming your product in the answer text. Or it can cite you, showing one of your pages as a linked source. They often come apart. ChatGPT may recommend your product and link a G2 page as the source, or link your pricing page while recommending a competitor.

This guide is about citations. They matter for three reasons:

• They send traffic. A citation is the only part of the answer a buyer can click to reach you.
• They control the facts. When ChatGPT cites your page, the pricing and feature claims come from you, not from a two-year-old review.
• They can be measured per page. You can see which URLs get cited and fix the ones that should be but are not.

If your goal is being named on shortlists, read our guide to how ChatGPT picks SaaS brands alongside this one. Mentions depend more on what third parties say about you; citations depend more on your own pages.

Demand for this topic is real and spread across many phrasings. We pulled the main variants from Ahrefs this morning:

Notice that "how to find out AI citations for a website" is as common as "how to get cited by ChatGPT." Teams want to know where they stand before they change anything. We cover both, starting with what ChatGPT actually cites.

What ChatGPT cites in 2026: the evidence

The best recent data comes from Ahrefs' study of ChatGPT query fan-out, published October 12, 2026. It analyzed 66.6 million logged-out ChatGPT responses from August and September 2026 and the 15 million searches ChatGPT ran behind them, plus US citation data from March to September. Four findings matter for SaaS teams:

1. Official sources win on facts. On a set of 20,000 US prompts, 84.4% of factual answers cited at least one official source. For comparison prompts the figure was 68.3%.
2. ChatGPT searches brand sites directly. 6.8% of fan-out queries used the site: operator, and 75.9% of the site: searches aimed at the roughly 160 most-searched domains targeted brand sites.
3. Price questions are growing. The share of English fan-out queries containing "price" rose from 6.0% in August to 8.9% in September. "Official" rose from 10.1% to 13.1%. We map these patterns to specific SaaS pages in our guide to query fan-out.
4. Aggregators lost ground. In US answers, Reddit citations fell 88% between March and September (from about 1 in 5 answers to about 1 in 43) and Wikipedia fell 80%. Ahrefs' read is that ChatGPT "seems to be prioritizing original sources over sites that summarize them."

Ahrefs is clear about the limits: it is one model snapshot over two months, free logged-out sessions search less than paid ones, and Ahrefs sells the tool that supplied the citation data. Treat it as direction, not law.

A second data point looks at pages, not domains. A Search Engine Land audit of 15 sites (about 7,500 ChatGPT referral sessions) reviewed the blog posts that ChatGPT sent traffic to. 72.4% of them included an "answer capsule," a self-contained explanation of about 20 to 25 words right after the title or a question-style H2. 52.2% contained original data or owned insight. Roughly 91% of capsules contained no links. The audit is correlational and did not compare against posts that were not cited, so read it as a pattern among winners, not proof of cause.

One more finding changes how you plan. In an earlier Ahrefs study of 15,000 prompts, only 8.0% of ChatGPT's in-text citations ranked in Google's top 10 for the same prompt (data from July 2025). ChatGPT rewrites the buyer's question into its own narrower searches, so a page can be cited without ranking for the head term, and a top-ranking page can be skipped.

The five checks for a citable page

Put the evidence together and a page needs to pass five checks before ChatGPT will cite it reliably. Run them in order, because a failure early on makes the later ones irrelevant.

Steps 1 and 2: fix access, then make facts findable

OpenAI's publisher FAQ is short on this point: "Any public website can appear in ChatGPT search," and to be included in summaries and snippets you should not block OAI-SearchBot. Blocking GPTBot is a separate choice that only affects training. CDN and firewall bot-protection rules can block it even when robots.txt allows it, so check both. Our guide to AI crawlers lists every user agent and the rules to set.

Retrieval is the step most teams skip. Because ChatGPT runs site: searches on brand domains and asks more price and "official" questions, the facts it looks for need to live on plain, crawlable HTML pages with clear titles. For a B2B SaaS company, that usually means:

1. A pricing page with real numbers or a clear pricing model in text, not only in an image or a calculator.
2. One page per major integration, named the way buyers search ("[product] Salesforce integration").
3. A security and compliance page listing certifications, data residency and the date of the last audit.
4. Docs and changelog pages that are public and indexable, so feature questions resolve to you.
5. A comparison page for each main competitor that states the differences factually.

Steps 3 and 4: write passages worth quoting

Once ChatGPT finds a page, it needs a passage it can lift. The pattern from the Search Engine Land audit is easy to apply: under the title and each question-style H2, put a 20 to 25 word answer that stands on its own, with no links and no "as mentioned above." Then expand.

Compare two openings for a section titled "Does Acme integrate with HubSpot?":

• Hard to quote: "Integrations are central to how modern teams work, and we've invested heavily in our ecosystem over the years."
• Easy to quote: "Yes. Acme has a native two-way HubSpot integration that syncs contacts, companies and deals every 15 minutes on all paid plans."

(Acme is a made-up example.) The second version answers the question, names the specifics and can be cited without editing.

Originality is the other half. Ahrefs' data points to ChatGPT preferring the source over the summary, and the Search Engine Land audit found original data in about half of cited posts. For a SaaS company, the things only you can publish include product facts, usage benchmarks from your own data, results from tests you ran, and clear opinions from people who do the work. A "what is X" post that restates the top five results gives ChatGPT no reason to cite you over them. Our LLM SEO guide covers how models weigh sources more broadly.

Two quick rules keep pages citable over time. Put a visible "last updated" date on factual pages and update it when facts change. And keep one canonical page per fact, so old blog posts with outdated pricing do not compete with your pricing page.

Step 5: find out which of your pages get cited

You can measure citations at the URL level with two methods. Use both, because each misses things the other catches.

Referral data in GA4

OpenAI says ChatGPT "automatically includes the UTM parameter utm_source=chatgpt.com in referral URLs." Build a GA4 exploration with session source = chatgpt.com and landing page as the dimension. That gives you the pages buyers clicked through from a citation. It undercounts, since many people read the answer and never click. Our guide to tracking AI traffic in GA4 shows how to set up the channel properly.

A logged prompt set

Write 30 to 50 prompts your buyers would ask, split across factual ("what does [brand] cost"), comparison ("[brand] vs [competitor]") and category ("best [category] for mid-market") questions. Run each with search on, several times, and log every cited URL. You will see three lists: your pages that get cited, your pages that should be but are not, and third-party pages cited in your place. AI answers vary from run to run, so compare totals over several runs rather than reacting to one. Our guide to AI visibility tracking covers prompt design and sample sizes.

For Google's AI Overviews and AI Mode, the rules are simpler. Google says a page needs to be indexed and eligible for a snippet, and "there are no additional requirements" or special optimizations, per its AI features documentation. The same citable pages tend to serve both.

A 30-day plan to earn your first citations

1. Week 1: baseline. Check OAI-SearchBot access in robots.txt and your CDN. Build the prompt set and log today's cited URLs. Set up the GA4 chatgpt.com report.
2. Week 2: fact pages. Fix or create the pricing, integration, security and comparison pages your factual prompts need. Put the facts in plain text.
3. Week 3: capsules. Rewrite the opening of your top 10 pages and each question-style H2 so the first 20 to 25 words answer the question.
4. Week 4: original content and re-measure. Publish one piece built on your own data or testing. Rerun the prompt set and compare cited URLs with week 1.

Expect the factual prompts to move first, because they map to pages you control. Comparison and category prompts take longer, since ChatGPT still leans more on third-party sources there. That is where earning brand mentions on review sites and industry press takes over.

FAQ