How to Find Traffic Sources in GA4: Source, Medium, and Scope

A traffic source in GA4 can describe first acquisition, a session, or attribution for an event. Those are different questions, so opening a source/medium table is only the starting point. Choose the reporting scope and comparison conditions before evaluating a campaign. The product guidance below was checked against official documentation on October 11, 2026.

Choose the acquisition question before opening a report

Use First user dimensions to investigate initial acquisition, Session dimensions to investigate visits, and event-scoped dimensions for event attribution. Google’s scope documentation explains these distinctions. A dimension without a Session or First user prefix should not be treated as interchangeable with either.

Consider a fictional visitor who first arrives through organic search and returns through a tagged email the next day. Different sources across the first-user and session reports are expected. If two success events are sent during the return visit, that does not establish two registrations or two people.

Separate source, medium, and channel

Source identifies an origin, medium describes the method, and a channel groups traffic using classification rules. A source can be a manually assigned value such as newsletter; it is not always the domain immediately preceding the visit. See the Traffic acquisition report guide for the relevant dimensions.

Keep a record of the distributed link. A platform name alone does not distinguish paid advertising, an ordinary post, and a manually tagged campaign.

Read session traffic in the acquisition report

1. Select the intended GA4 property and reporting dates.
2. Open Traffic acquisition. If it is missing, use report search or check which report collection your administrator published.
3. Select Session source / medium and review sessions alongside engagement.
4. Save the comparison conditions before changing dimensions or filters.

Navigation varies with report customization. Finding a familiar menu does not prove that the property or scope is correct. With small counts, describe actual sessions before percentage changes; one additional visit is weak evidence for a broad performance claim.

Compare landing pages with consistent host and channel filters

For article acquisition, distinguish a visit that starts on the article from one that merely includes a later article view. Use the landing-page dimension and a session channel condition for Organic Search. Record the production hostname, language, time zone, dates, and exact paths. A path alone can include a matching page on a test host.

Keep all-channel entrances separate from organic-search entrances. Do not divide sessions by Search Console clicks from another period to create a conversion rate. Likewise, do not average rounded row-level engagement times to manufacture a cohort total. Consistent conditions matter more than a convenient combined number.

Diagnose UTM, direct, and missing source values

Use a campaign dictionary for shared links. Google’s URL-builder guidance describes the relevant UTM fields. Missing source or medium does not make every parameter in the URL invalid; incomplete fields can produce missing report values. Test the original link, redirects, received values, and processed acquisition report.

The official direct-traffic explanation also makes clear that direct does not mean only typed URLs. Missing campaign or referral information can contribute. Page referrer is a separate field: it may describe an internal page, contain only an origin, or be unavailable. It is not a complete inventory of external backlinks.

Check referral handoffs and validate key events

A payment return and a journey between owned domains require different checks. Cross-domain measurement addresses identifier continuity, while unwanted-referral configuration controls whether a domain supplies referral attribution. Excluding a referrer does not join identities or repair historical data.

Before reporting a business outcome, verify the success condition, production host, consent behavior, authentication or app handoff, duplicate sends, and counting method. An event count is not a verified headcount. Finish with a short record of the observation, unverified conditions, next test, and owner so that an unexpected source becomes an actionable investigation.

Continue with the relevant guide

• What Is Referral Traffic? GA4 Referral, Source, and Direct Explained
• URL Parameters Explained: Query Strings, UTM Tags, and GA4
• GA4 Dimensions vs. Metrics: Scope, Examples, and Reporting