Skip to content
Analytics & improvement

Review visit sources and device coverage

Inspect recorded acquisition evidence with the right scope and browser limits.

CCorthex product guidesUpdated October 15, 20263 min read
Where to find itAnalytics → Sources / Herkomst, at workspace or assistant level

Before you begin

Open Analytics for activity and usage, or the relevant agent’s Learning tab for reviewable conversation-quality findings. Confirm the workspace, agent, period and scan coverage before drawing conclusions.

Example workspace · Demo data. These are illustrative demo values; compare the same scope and reporting period. Select the image to enlarge it.
Example analytics overview cards with usage and conversation counts

Example workspace · Demo data. These are illustrative demo values; compare the same scope and reporting period. Select the image to enlarge it.

Step by step

  1. 1

    Choose the report scope

    Select the dates, assistant and channel. Sources uses recorded visits in that selection. The first recorded date describes available recording for the assistant/channel; it does not reconstruct traffic before recording began.

  2. 2

    Read the source summary

    Check recorded visits and the most-used observed external referring domain. Source coverage separates external referrers, campaign-tagged visits, Direct / not provided, internal arrivals with an earlier source unknown, and unavailable evidence. A tagged visit may also have an external referrer, so those coverage counts overlap.

  3. 3

    Inspect a referring domain

    Open an external domain to see the referring origins and paths the browser actually supplied. An origin-only referrer reveals no exact article, search phrase or social post. Missing source details cannot be recovered from the report.

  4. 4

    Compare useful dimensions

    Review campaign source, medium and name, landing pages and operating-system families. Campaign tags are reported labels, not verified advertising attribution. Unknown device evidence stays Unknown; exact hidden OS versions are not guessed.

  5. 5

    Export the displayed lists

    Download CSV for the displayed top-20 aggregate lists. Shares use all recorded non-automated visits in the selection, not only the visible rows. Distinct visitor counts use assistant/browser identities; adding counts across rows does not produce distinct people across devices.

Good to know

The captured arrival is retained through ordinary full-page/SPA navigation and same-origin tabs when persistent browser identity is available. Previous-page context is separate. A new visit starts after 30 minutes of observed inactivity; passive heartbeats do not themselves renew that activity window. Older clients without a token use an observed-update fallback with less precise inactivity evidence.

Direct / not provided is ambiguous: email, apps, bookmarks, typed URLs, redirects and privacy settings can all hide the source. Referrer policies are not weakened. This is first observed arrival per visit, not lifetime first-ever attribution, multi-touch conversion attribution or proof of an ad impression.

Browser identities are scoped to each assistant. Clearing/denying storage, multiple domains and changed identities can split visits. No new high-entropy device fingerprint is used to identify unknown people. Acquisition URLs retain HTTP(S) origin/path rather than arbitrary queries; campaign labels are stored separately.

Known automated agents are excluded from report totals using explicit user-agent markers and counted separately. This is not comprehensive bot detection, and geography alone does not identify automation.

Check the result

Compare the same scope and period before drawing conclusions. Investigate real conversations and sources alongside aggregate metrics.

Questions & troubleshooting

The report has no earlier visits

Recording does not backfill mutable historic presence snapshots. Check the first recorded date and selected scope. An empty result is different from a visible report-loading or service error.

An exact referring page is missing

The browser may have provided only a domain or no referrer at all. A campaign tag does not reveal the missing page or establish a visitor’s identity. Compare captured evidence without guessing.

Was this article helpful?

Rate this guide on this device. Ratings are not sent to Corthex.