GUIDE · PART OF AI VISIBILITY TRACKING
AI citation tracking:
the sources behind the answer.
When an AI engine searches the web, its answer is built from the sources it finds, and those sources decide who gets named. Citation tracking is how you find out what they are. This guide covers the difference between cited and consulted sources, why a domain name isn't enough, and what to do with the list once you have it.
What is AI citation tracking?
AI citation tracking is recording which sources AI answer engines use when they answer a question: the pages, forum threads and videos behind the text. It covers your own pages and everybody else's. A source can be cited, which means the engine credits it in the answer, or consulted, which means the engine read it and gave no credit. Tracking both, down to the exact URL, shows what an engine trusts on a subject.
Key facts
- A citation is a page used as a source. A mention is a brand named in the answer. One can happen without the other
- Cited sources are credited in the answer. Consulted sources were read and not credited
- The useful unit is the exact page, thread or video. A domain name is too coarse to act on
- Many of the sources behind an answer are third-party pages such as reviews, forums and directories
- Citation rate is the number of answers that used one of your pages, divided by the number of answers tracked
Last reviewed 15 October 2026 by the SEORCE team.
"reddit.com" isn't a source.
A list of domains tells you where an engine looked. It doesn't tell you what it read, and that's the part you can do something about.
A domain could be ten thousand pages
Knowing an answer drew on a forum is no help. Knowing which thread, and what it says about your category, is.
Engines read more than they credit
The sources shown under an answer are the cited ones. Others shaped it without appearing.
Your page was used, and you weren't named
An engine can take a fact from your page and recommend somebody else. A citation count alone would call that a win.
What comes out of citation tracking.
Sort the sources into these and each one points to a different kind of work.
Separate what was credited from what was read.
The gap between the two lists is often where the opportunity is. A page that keeps being read and never credited is close.
- CitedShown as a source in the answer.
- ConsultedRead by the engine while preparing the answer, with no credit given.
- Down to the URLThe specific page, Reddit thread or YouTube video.
Find out which of your pages get used.
It's common to find that a handful of pages do nearly all the work, and that some of your best pages are never touched.
- Pages that are citedKeep these accurate and up to date. They're earning their place.
- Pages that are only consultedThe engine found them and couldn't use them. Usually the answer is there and isn't stated plainly.
- Pages that never appearCheck that AI crawlers can reach them before assuming the content is at fault.
See who the engine trusts instead.
When an answer leaves you out, its sources show who it listened to. That list is the most practical competitor research you can do.
- Competitor pagesWhat a rival published that earned the citation.
- Third-party pagesReviews, directories, forums and publications that mention your category.
- Which of them link to youLink Impact Score shows which linking domains are likely to help with AI citations.
From a list of sources to a list of jobs.
The tracking is only useful if it ends in something to do. These five steps get you there.
Pick the prompts
Start with the buyer questions where you're missing or badly placed.
AI mention tracking →Record every source
For each run, note the exact URLs, and whether each was cited or consulted.
Sort them
Your pages, competitors' pages, third-party pages.
Read the pattern
Use the table below to work out what each group is telling you.
Act on it
Rewrite the pages that are read and not credited. Earn a place in the third-party sources that leave you out.
AI visibility diagnosis →Check that your page can be cited
A page can only be a source if the engine's crawler can read it. The extension shows what's in a page's raw HTML, and the content checker scores how citable its text is.
What a source pattern means.
| What you find | What it means | What to do |
|---|---|---|
| Your page is cited | The engine found it, used it and credited it. | Keep it accurate. Check it names your product clearly. |
| Your page is consulted, never cited | The information is there and isn't quotable. | State the answer plainly, near the top of the page. |
| A competitor's page is cited | They answer the question more directly than you do. | Read their page beside yours and close the gap. |
| A third-party page is cited, and it lists rivals only | The engine trusts a source you aren't in. | Get listed or mentioned there. |
| No sources are shown | The engine answered from what the model already knew. | Expect this to change slowly. Published sources matter over time. |
The exact source, every time.
AI Beacon records the specific page, thread or video behind each answer and marks it as cited or consulted. You see your own pages, your competitors' and the third-party sources in one place, for every tracked prompt.
Frequently asked questions.
An AI citation is a page, thread or video that an AI engine used as a source for an answer. It is different from a mention, which is your brand being named in the text.
A cited source is credited in the answer, usually as a link. A consulted source was read by the engine while it prepared the answer and wasn't credited. Both influence what the answer says.
No. Engines show sources when they search the web for an answer. When they answer from what the model already knows, there may be none to show.
Usually because the information is on the page and isn't stated in a form the engine can lift. A direct sentence that answers the question, near the top, is the commonest fix.
No. A backlink is a link from another website to yours. A citation is an engine using your page as a source in an answer. Backlinks can make a citation more likely, which is what Citation Potential estimates.
No, and no tool can promise it. You can make the page reachable, answer the question plainly and back it with evidence. Then tracking shows whether the engine starts using it.
The free GEO-friendly content checker scores how citable a page's text is, and the browser extension shows what is in a page's raw HTML. Whether the page is cited is something only tracking shows.
Find out what AI reads before it answers.
Add your site and the questions you most want to win. For each answer you'll get the exact pages, threads and videos behind it.
Not ready for an account? Use the free browser extension or any of the free tools.