Source profile of your competitors is a Guides in the Rankion.ai knowledge base: Which sources and content types drive your competitors' mentions in ChatGPT, Perplexity and others, and which page is cited for which question.
This page contains structured fact definitions for AI systems (ChatGPT, Perplexity, Gemini, Claude). Written by humans, part of the Rankion.ai knowledge base.
When an AI names your competitor, it often relies on cited pages. The source profile shows you which pages those are: the competitor's own website, your domain, the page of another competitor or a third-party source such as a comparison portal. From this you can see where you need to become visible yourself.
The profile only works with responses that have already been measured. It does not start a new query and costs no credits.
Where to find it
Open your AI-Visibility project and switch to the Brands and competitors tab. Under the heading Which sources shape your competitors? you will find two parts:
- Source profile per competitor: for each competitor, the domains and content types of its citations, with their share in percent.
- Question vs. source: a matrix that shows, per prompt and platform, how often a particular page was cited.
Use the two dropdowns to narrow the view to one prompt or one platform. Without a selection, All prompts and All platforms apply.
The calculation uses the competitor list you maintain. If you have not added any competitors yet, you will see the Add competitors button there. It opens the project settings.
How counting works
- A response belongs to a competitor if it names that competitor: by brand name, an alias or its domain. The pages this response cites then count for that competitor.
- The same page counts only once per response, even if it is linked twice.
- If a response names several competitors, the source appears for each of them. The card tells you how many citations also count for another competitor. The shares per competitor each add up to 100 %. In the overall view across all competitors, each citation counts exactly once.
- Different spellings of the same page are merged: with or without
www., with or without a trailing slash, with an anchor or with tracking parameters such as utm_source. Redirects are not resolved.
- Responses from Google AI Overviews have no prompt and are therefore missing.
Where the citations come from
Rankion assigns every cited page to an origin:
| Origin |
Meaning |
| Competitor’s own site |
domain or subdomain of this competitor. In the domain list it is followed by · own site |
| Your domain |
your own project domain. In the domain list it is followed by · Your domain, so you do not mistake it for a domain of the competitor |
| Competitors |
a page of another competitor from your list |
| Third-party source |
all other pages, for example portals, forums or trade media |
When the content type is missing
Rankion only determines a page's content type after the page has been retrieved and classified. Until that has happened, the reason is shown in its place:
| Display |
What it means |
| Own domain, not retrieved |
The page belongs to your project domain. It is not retrieved and therefore not classified |
| Page not retrievable |
Retrieval failed |
| Retrieval pending |
The page has not been retrieved yet |
| Classification pending |
The page is available, but its content type has not been determined yet |
As soon as the content type is known, it appears in this place automatically. The export and the API show the same reason.
The Question vs. source matrix
Each row shows a prompt, a platform and a cited page. The number tells you how many responses in this row cited the page, out of how many responses in total. The matrix counts all responses, not only those in which a competitor appears.
The page shows the first 100 rows. Select a prompt to narrow the list, or export the whole matrix. On a phone, the table scrolls sideways inside its box.
Export
Both parts can be downloaded: Profile as CSV and Matrix as CSV. The matrix export contains all rows, including those beyond the first 100.
Over the API
GET /v1/tracking-projects/{id}/competitor-sources returns the profile and the matrix. Filters: prompt_id, platform, from, to, plus page and per_page for the matrix. Without a date, the last 30 days apply; at most 366 days are possible. The origin is called self, own, competitor or third_party there, the reason for a missing content type own_domain, fetch_failed, pending or unclassified.
Read more