Read referring domains without guessing missing traffic sources

The referring-domain report describes source information that reached Stats4U. It is not a complete list of every route to the website, and a row is not automatically a distinct person. Useful analysis separates referring domains, the page being counted and deliberately marked campaign links.

Check what the report actually received

1. Open the referring-domain report for the intended counter and dates. Read the metric label before interpreting its figures. It groups reported source hosts; it does not provide the complete original article URL, search query or a verified number of unique people.
2. Check how the counter is installed. The JavaScript integration sends document.referrer as sndref. That value describes the referring document when the browser makes it available. The counted page address is sent separately and answers a different question.
3. With an image-only integration, the HTTP Referer of the image request can identify the embedding page. It does not identify where the reader was before arriving there. Treating that header as the traffic source would incorrectly describe the website itself as the source.

Missing referrers do not prove a direct visit

Browsers, link policies and applications can omit or shorten referrer information. A copied link, an email or a messaging application can arrive without a usable source. The absence of a referring-domain entry cannot reliably distinguish a bookmark from every other unreported route.

Do not weaken the website Referrer-Policy merely to fill a report. The policy controls what a browser may disclose. Cross-origin requests commonly expose only an origin, and a restrictive policy can expose nothing. Lost information cannot be reconstructed later from the counter total.

Investigate a change with matching evidence

Compare the same-length periods. Current collection excludes recognised own-domain referrers; older records can still contain them, particularly after changed domains or integrations. Set recognisable own hosts aside when investigating external arrivals. The presence of a large platform alone does not establish whether it is hosting infrastructure, a recommendation or another kind of link.

A new source host is a lead to investigate. Compare the dates with a known publication or campaign and check the available destination-page data. Do not identify a forum thread, a person or a precise marketing cause from a domain-only row. Missing source data and small samples remain limitations.

Use campaign links for planned distribution

For a newsletter, partner placement or another planned channel, enable optional campaign measurement and use the UTM link generator. Use short, predefined campaign labels without names, email addresses or personal message text. The final landing page needs to retain the campaign parameters.

Campaign attribution and referring domains are separate reports. A marked link can supply campaign labels even when the browser omits the referring document, provided the integration receives the landing address and measurement is enabled. That does not turn campaign totals into cross-device identities or prove that a later purchase was caused by the link.

Measure a planned campaign

The campaign guide explains setup, permitted labels, landing-page checks and interpretation limits.

Read the UTM campaign guide

MDN: Referrer-Policy