Itemise the question, aggregate the crawl
A live-fetch token identifies the request as an on-demand retrieval rather than a background crawl. Its path shows what was requested, but neither the token nor vendor verification proves human initiation, reading or comprehension. A crawler builds an index or dataset over time, so each individual path is usually noise rather than an actionable research trail.
illun therefore records live requests individually and aggregates other recognised kinds per minute inside the customer’s process. Failed live identity claims are aggregated too, because an unverified path should not be presented as a genuine assistant journey.
Different shapes, honest comparison
The dashboard names both categories and keeps their collection fidelity visible. It does not draw a chart that visually equates one live request with one crawler count, and it never calls either category a visitor or user.
That distinction is part of the product model, not a presentation preference. It protects the unit—requests from identifiable agents—and keeps a useful retrieval from disappearing inside crawler volume.