AI Citation Tracker
Track which sources AI engines cite — and whether yours is one of them
Being named in an AI answer and being cited as its source are two different things. This tracks the second one, across ChatGPT, Perplexity, Gemini and Copilot.
Free tier: 60 prompts, no credit card.
What is an AI citation tracker?
An AI citation tracker measures how often AI answer engines attribute an answer to your domain — which of your URLs get used as sources, how often, on which questions, and which other domains are cited in your place. It records the answer's sources, not just its text.
That makes it different from a visibility or mention tracker, which records whether your brand name appears in the answer. The two can move independently: an engine can recommend you warmly while citing a review site rather than your own pages, and it can cite your documentation while recommending a competitor by name.
If you have ever wondered why a strong AI mention rate produced no traffic and no obvious way to intervene, this is usually the missing half of the picture.
Mentioned, cited, both, or neither
Every prompt puts your brand in one of four states. Each one calls for different work.
Cited and mentioned
The engine names you and attributes the answer to your page. You own the question. The job is to keep the source page current, because this is the state most easily lost to a fresher competing source.
Mentioned but not cited
The engine knows your brand but is building the answer from someone else's page — a review site, a listicle, a competitor's comparison. Your reputation is doing the work while another domain collects the attribution. Fix by making your own pages the better source for that specific question.
Cited but not mentioned
Your page is good enough to be a source, yet the answer recommends someone else. This usually means your content is informative but not positioned — it explains the category without making a case for you. Often the cheapest win available.
Neither cited nor mentioned
You are absent from the question entirely. Before writing anything, look at which domains are cited — appearing on one of those is frequently faster than trying to win the question from your own site.
Citations and mentions are different metrics
Both matter. They are measured differently and improved differently.
| Brand mention | Citation | |
|---|---|---|
| What is recorded | Your brand name appears in the answer text | A URL or domain is attributed as the answer's source |
| Where it comes from | Often the model's existing knowledge of your category | Usually live retrieval of a page the engine can reach and trusts |
| What moves it | Reputation, category presence, third-party coverage | Crawlable, specific, well-structured pages that answer the exact question |
| Who owns the fix | PR, partnerships, positioning | Content and technical SEO |
| Speed of change | Slow — tied to how the model understands your market | Faster — a new page can be picked up as a source without a model update |
| Visible to the user | Always — it is in the sentence | Depends on the engine; Perplexity shows sources plainly, ChatGPT often does not |
The practical reason to care about the citation column: it is the one you can act on this quarter. Mentions follow how a model understands your market, which is slow. Citations follow what is retrievable right now.
What a citation tracker records
Four levels, from the blunt number to the actionable detail.
Your citation rate
The share of sampled answers that attribute anything to your domain, broken down by prompt and by engine. This is the headline, and on its own it is the least useful number on the page — it tells you the size of the gap but nothing about its shape.
Which of your URLs get cited
Citations concentrate. Most domains find that a small number of pages earn nearly all of their attributions, and they are rarely the pages the marketing team considers important. Knowing which pages already work tells you what to write more of.
The competing source list
For every question where you are not cited, which domains are. Aggregated across a full prompt set, a handful of domains usually dominate a category — the sources that effectively decide who gets recommended.
The gap between citation and recommendation
Questions where you are cited but a competitor is named, and vice versa. This is where the cheapest improvements live, because the engine has already accepted one half of the case for you.
Why citations are the more actionable half
A mention rate tells you how a model currently understands your market. That understanding is built from an enormous amount of text and does not shift because you published a page last Tuesday. It is a real metric, and a slow one.
Citations behave differently. When an engine retrieves sources to answer a question, it is reading what is available now. A page that did not exist last month can be cited this month. That is why citation work has a feedback loop short enough to manage, and why it is usually where a first quarter of effort should go.
- You can identify the exact question you are losing, not just a category
- You can see which specific page won it instead of yours
- You can publish or restructure a page and re-measure the same prompts
- You can tell whether the change worked, rather than inferring it
None of this makes mentions unimportant — a brand nobody has heard of will struggle to be recommended no matter how citable its pages are. But mentions are a reputation programme, and citations are a content and technical programme.
AI citation tracking FAQs
What is the difference between an AI mention and an AI citation?
A mention is your brand name appearing in the answer text. A citation is a URL or domain being attributed as the source the answer was built from. They are independent: you can be recommended by name while the answer cites a review site, or have your page used as a source while the answer recommends someone else. Tracking only mentions hides which of the two is actually failing.
Can I be cited by an AI engine without being mentioned?
Yes, and it is more common than people expect. It typically happens when your content explains a topic well without making a case for your product — a good documentation page or blog post gets used as the factual basis while the recommendation goes to a competitor. It is one of the most fixable situations in this whole discipline, because the engine has already accepted your page as trustworthy.
Which AI engines actually show their sources?
Perplexity is the most transparent and attributes sources openly in the answer. Copilot generally surfaces links, drawing on the Bing index. Gemini shows sources inconsistently depending on the question. ChatGPT cites when it retrieves but often answers from what the model already knows, with no links at all. Because of that spread, a citation figure only means something when you know which engine produced it.
How do I find out which of my pages AI engines cite?
You have to capture the sources from the answers themselves, at enough volume for the pattern to be stable. Server logs and analytics will not tell you — a citation is not a visit, and most cited pages get attributed far more often than they get clicked. Run a prompt set across the engines, extract the cited URLs from every answer, and aggregate by page.
Do AI citations send referral traffic?
Some do, and the volume is usually modest relative to how often a citation occurs. Perplexity and Copilot link out, so those citations can produce measurable referrals; ChatGPT frequently answers without any link to click. The honest position is that a citation's value is mostly in influencing the recommendation, not in the click — which is exactly why it needs measuring directly rather than being inferred from analytics.
Go deeper
This page is the overview. Each of these covers one part of it properly.
Find out who is being cited in your place
Run 60 prompts free and read the sources behind every answer you are missing from.