What Is LLM Citation Tracking? How to Monitor What ChatGPT, Gemini and Claude Say About Your Brand

LLM citation tracking is the practice of asking AI assistants such as ChatGPT, Gemini and Claude the questions your buyers ask, at scale and on a schedule, and recording whether they mention your brand, cite your website as a source, and what they say about you compared with competitors. It is AI brand monitoring for the answers people now read instead of a list of links.

Search Console tells you how you rank on Google. It cannot tell you whether ChatGPT recommends your competitor when a buyer in Cape Town asks for the best option in your category. LLM citation tracking fills that gap.

In short

• LLM citation tracking measures two things: mentions (your brand is named) and citations (your page is linked as a source).
• The audience is large: OpenAI said ChatGPT reached 900 million weekly active users in February 2026, and Google said AI Overviews had over 2.5 billion monthly active users in May 2026.
• AI answers vary from run to run, so tracking needs a fixed question set, repeated runs and trends, not screenshots.
• AI share of voice is your brand mentions divided by all brand mentions in a set of answers, times 100.
• Tracking is the measurement. Generative engine optimisation (GEO) is what you do about the results.

What is LLM citation tracking?

A large language model (LLM) is the technology behind assistants such as ChatGPT, Gemini and Claude. When these assistants answer a question, they can do two different things with your brand:

Term	What it means	Example
Mention	The answer names your brand	"Popular options include Kestrel Insurance and two others."
Citation	The answer links your page as a source	A source card pointing to your pricing page
Third-party citation	The answer cites someone else's page about you	A review site or news article that names you
Sentiment	How the answer describes you	"Known for fast claims" versus "mixed reviews"

Citations exist because many assistants now search the web while answering. Anthropic says that when Claude uses web information, "it provides direct citations" so users can check the sources. OpenAI's crawler documentation says its OAI-SearchBot is used "to surface websites in search results in ChatGPT's search features." Google's AI Overviews and AI Mode also show links to the pages they draw on.

Tracking all of this over time, against your competitors, is LLM citation tracking.

Why does AI brand monitoring matter for South African brands?

Because the audience is already there. TechCrunch reported on 27 February 2026 that ChatGPT had reached 900 million weekly active users. At Google I/O on 19 May 2026, Sundar Pichai said "AI Overviews now has over 2.5 billion monthly active users", and that AI Mode had passed 1 billion.

Closer to home, CitedAI ran 96 Google searches from South Africa across 12 sectors on 11 October 2026. Google showed an AI Overview on 92 of them, or 95.8%, according to its study of AI Overviews in South Africa.

For example, a buyer asks "Which medical aid is best for a young family in Gauteng?" or "Who installs reliable solar in Durban?" If the answer names three competitors and not you, you lose the sale before the buyer ever sees your website, and your analytics show nothing.

How does LLM citation tracking work?

1. Build a question set. List the questions buyers actually ask, both unbranded ("best car insurance for young drivers") and branded ("is Kestrel Insurance any good?"). CitedAI's guide to choosing the prompts to track covers how.
2. Run the questions across assistants. Typically ChatGPT, Gemini, Claude and Google's AI Overviews, in the way real users reach them.
3. Repeat the runs. The same question can return different brands on different days. As CitedAI's guide to measuring GEO puts it, "One answer is an anecdote." Several runs give you a rate.
4. Parse the answers. Record which brands are mentioned, in what order, with what sentiment, and which domains are cited.
5. Compare with competitors. Your numbers only mean something next to theirs.
6. Track the trend. Monthly or weekly, so you can see whether changes to your site and PR move the numbers.

Even large vendors treat the output as directional. Ahrefs says its Brand Radar runs prompts through "the free, publicly available web interfaces" of the assistants it covers, and that its metrics are "directional indicators, not exact traffic counts." That is the right way to read any AI visibility number.

What should you measure?

Metric	What it tells you
Mention rate	Share of answers that name your brand at all
AI share of voice	Your mentions as a percentage of all brand mentions in the set
Citation share	How often your domain appears among the cited sources
Position	Whether you are named first, in the middle or last
Sentiment and accuracy	Whether the description is positive and factually right
Cited third-party domains	Which review sites, directories and publications the assistants rely on in your category

CitedAI defines AI share of voice as "the percentage of all brand mentions in a set of AI answers that belong to your brand." The last row is often the most useful: if assistants keep citing one comparison site, getting listed there accurately may do more than rewriting your homepage.

What do you do with the results?

That is the work of generative engine optimisation. A few evidence-based starting points:

• Get the SEO basics right. Google says there are "no additional requirements to appear in AI Overviews or AI Mode," and that pages must be indexed and eligible to show with a snippet. See its guidance on AI features and your website.
• Let the right crawlers in. OpenAI recommends allowing OAI-SearchBot in robots.txt if you want to appear in ChatGPT search.
• Make pages easy to quote. The GEO study by Aggarwal and colleagues, accepted to KDD 2024, found that optimisation methods could boost visibility in generative engine responses by up to 40%, with results that varied across domains. It was a controlled study, not a guarantee.
• Fix what is wrong. Correct outdated facts on your own site and on the third-party pages assistants cite.

CitedAI's articles on how ChatGPT recommends brands and GEO for SA brands go deeper. Our post on how AI agents are changing SEO and GEO covers the execution side.

How do you start tracking?

Start small. Write 20 buyer questions, run them in three assistants once a month, and log mentions and cited domains in a spreadsheet. That is enough to see where you stand.

When the manual version becomes a chore, use a tool. CitedAI, built by AmbitX, tracks what ChatGPT, Claude and Gemini say about a brand and its competitors; see the CitedAI product page. Whichever tool you choose, ask how many runs sit behind each number and which questions are in the set. If you are hiring help, our guide to choosing an SEO agency in South Africa covers what to ask about AI visibility.