Journeys
Where a page leads, as a flow rather than a list.
A funnel asks whether a path you chose was walked. A journey asks what people actually did, without you naming it first.
GET /api/stats/journeys?site=1&range=30d&depth=4&start=/docs
It returns nodes and links (a flow) from an entry page outward, to whatever depth you ask for.
The two controls
Start. Anchor the flow to an entry page, chosen from your real entry paths. Leave it off to see every entry at once, which is a wider and usually noisier picture.
Depth. How many steps forward to follow. Three or four is where it stays readable; ten is a diagram nobody can read, and the answer it gives is rarely different.
Reading it
Follow the widest link out of your busiest entry. That is what most readers do next, and it is frequently not what the page was designed to make them do.
Then look for the exits. A step where a large share leaves is either the end of a job finished (which is fine) or the point where the page ran out of what someone came for. The path list next to the flow tells you which, because a session that ended after converting looks different from one that ended after ten seconds.
Journeys and agents
Switch the audience to agents and the flow becomes crawl order, which pages a crawler reaches from which, and how deep it goes. That is a genuinely different question from the human one, and worth asking on a documentation site: a crawler that never gets past your index is a crawler that never read the pages you wrote.
Crawl flow has no sessions in the human sense, so the numbers are fetches rather than people, and the screen says which it is showing.