Co-cited sources
Find the publications the model already trusts on your subject. You get the domains cited alongside your brand, or instead of it, ranked by frequency and question type.
Authority, the fourth rung
Find the pages cited instead of yours
Every answer is split into one row per cited URL. You get the sources engines trust on your topics, and the short list of your pages that earn a citation.
Key takeaway
When an engine answers a question about your category, it cites a set of sources. The ones that are not yours are not a ranking problem. They are a list of pages to be mentioned on, ordered by how often the engine reaches for them.
Two lists, and the space between them
Citations are extracted per probe, so every count traces back to a specific answer you can open and read.
Your own pages cited
See which of your URLs actually earn citations, on which prompts and from which engines. Owned coverage is flagged separately from earned, so the two never average into one flattering number.
Linked against unlinked mentions
Find out which of two problems you have. Mentions are counted both ways, because an unlinked mention is a reputation signal and a linked one is a retrieval signal, and one metric would hide which you are facing.
The outreach list
Work a ranked outreach list instead of a scraped domain dump. It is the sources cited on your topics that never cite you, and every row traces back to the probes it came from.
What sits behind each citation
Extracted per probe rather than aggregated up front, so any number on the page can be opened down to the answer that produced it.
One row per probe and URL
The citation array on every probe result is unnested, so a single answer citing eight sources produces eight rows rather than one blended record.
Engine and probe type
Each citation carries the engine that made it and whether the probe ran with live web search or without, which separates a retrieval citation from a training-data one.
Owned domain detection
Cited domains are matched against your tracked property domains, so owned pages are identified automatically instead of maintained as a list you have to keep current.
Position within the answer
Where your citation sat among the sources for that answer. Being the eighth source of eight is not the same result as leading it, and averaging position hides the difference.
The prompt that caused it
Every citation links back to the prompt that produced it, so you can see which question types earn citations and which never do.
Citation match after a change
When you follow a recommendation, the target URL is watched for appearance in later citations. A citation match alongside a mention-rate lift is what strengthens a winner verdict.
Why this is the rung most brands skip
Citations are where AI visibility stops being about your own site.
Authority is borrowed before it is owned
Move faster by getting cited elsewhere than by publishing again. Engines cite sources they already trust on a subject, and the co-cited list names exactly which ones count in your category.
It explains a crawled-but-never-cited page
Stop rewriting a page that retrieval already reaches. If it is fetched regularly and never cited, authority is the constraint, which is a different quarter's work from a robots fix.
Leading indicator, stated as one
Read citation rate as the leading indicator it is. It ships with its noise level and is never converted into attributed revenue, which keeps it in the same regime as the rest of AI visibility and SEO.
Against the usual approach
Most tools report a mention count and stop there.
Co-cited third-party sources
TrustData
GA4
Other tools
SometimesYour own pages cited, separated out
TrustData
GA4
Other tools
RarelyLinked against unlinked mentions
TrustData
GA4
Other tools
Every citation traceable to a probe
TrustData
GA4
Other tools
Citation match after a content change
TrustData
GA4
Other tools
| Feature | TrustData | Google Analytics 4 | Other tools |
|---|---|---|---|
| Co-cited third-party sources | Sometimes | ||
| Your own pages cited, separated out | Rarely | ||
| Linked against unlinked mentions | |||
| Every citation traceable to a probe | |||
| Citation match after a content change |
FAQ
Direct answers.
AI citation tracking records which URLs an AI engine references when it answers a question. TrustData unnests the citation array on every probe result into one row per cited URL, so a single answer citing eight sources produces eight rows. Each carries the engine, whether the probe ran with live web search, the prompt behind it, and whether the domain is one of yours. That gives two views: the third-party sources cited on your topics, and your own pages that earn citations.
A mention is your brand named in the answer text. A citation is a URL the engine references as a source. You can have either without the other, and they mean different things. An unlinked mention says the model knows you and is willing to name you, which is a reputation signal. A citation says the engine went to a page and used it, which is a retrieval signal. TrustData counts both separately, because a brand with many mentions and no citations has a very different problem from one with the reverse.
Treat it as a ranked outreach brief. The domains on it are the ones an engine already trusts on your subject, and the ones cited most often on your question types are where a placement moves the needle furthest. Because each row traces back to the probes that produced it, you can see which specific questions a source is winning before you approach it, rather than pitching blind from a domain-authority score.
No, and TrustData does not report it that way. Citation rate is a leading indicator, published with its noise level. The sixth rung of the readiness ladder reports AI Search sessions and their conversion rate as proxy signals, labelled emerging. Causal claims are reserved for paid media, where holdout tests can carry them. A tool telling you a ChatGPT citation produced a specific amount of revenue is overstating what the data supports.
Because sources in an AI answer are not equal. Leading the source list for a comparison question is a materially different outcome from being the last of eight, and averaging the two produces a number that describes neither. Position is recorded per citation so that improvement within the source list is visible even in a period where your citation count did not change.
14-day free trial
See who gets cited on your topics
14-day free trial. Citation tracking is part of AI Visibility, included in every plan.