Sources
The pages and domains AI engines lean on when they answer questions in your category, as a counted graph you can descend to the citing answers.
When an engine writes an answer it leans on sources. Sources is the counted record of which ones, how often, and which answers they fed. It is the lens that turns "improve your content" into a list of specific pages.
What you'll learn
- What the citation graph counts, and what it deliberately does not
- How to descend from a domain to the answers that cited it
- What to do with a source that is not yours
Citation graph, source, citing answers
The landscape is the citation graph: the domains that appeared as sources across your tracked prompts, with counts. Open one and you get that source: which prompts it turned up on, which engines used it, and how that has moved. Descend again and you are reading the citing answers themselves.
Everything in this lens is counted. There was once a model-written prose graph in the product, free-text sentences between nodes with no count, no weight and no time behind them, and it was retired for exactly that reason. If a node is here, something was counted.
The three kinds of source you will see
- Your own pages. These are the pages doing the work. Know them, keep them current, and do not restructure them casually.
- Pages about you that you do not own. Directories, review sites, marketplaces, press. These are often where an engine gets its opinion of you, and they are frequently out of date in ways you can fix with an email.
- Pages that have nothing to do with you. Category explainers, comparison articles, forums. These are where your rivals get named. A comparison page that lists five vendors and not you is a concrete, addressable gap.
What it is good for
- Finding the page to improve. If one page of yours carries most of your citations, that is your asset. If none of yours appears at all, you have a Structure or Authority problem rather than a content-volume problem. See Readiness.
- Finding the page to get onto. The third kind above, ranked by how often engines lean on it.
- Explaining a loss. A source that stopped appearing often explains a citation that stopped appearing.
What it cannot tell you
A source count is not a traffic number and not an attribution. It says an engine leaned on a page when answering, not that a person read it or clicked it. For arrivals from AI answers, that is Trace, and only the visits Trace can bind to a citation are claimed as coming from an answer we counted.
Try this in Outercite
Ask the assistant "Which pages are the engines citing me on?" then sort the graph by count and look for the first domain that is not yours.
Related
Intelligence
The surface Sources sits inside, alongside Ground, Where you lose, Engines, Gaps and Explore.
Proof
The verbatim sentences from the answers a source fed.
Trace
The other half: who actually arrived from an AI answer.
Last updated on