Website Traffic Sources: A Practical GA4 Guide

A website traffic source report answers where a visit was attributed, not why a person came or whether the visit created value. GA4 derives source, medium, campaign, and channel dimensions from collected campaign data, referrals, advertising integrations, and fixed classification rules. The correct interpretation depends on scope: first user, session, or event.

Start with the scope printed in the dimension name.

Source, medium, campaign, and channel

Google defines source as the origin of traffic, such as Google, Bing, or a newsletter. Medium describes how the visit arrived, such as organic, cpc, email, or referral. Campaign identifies a specific marketing effort. A channel group applies rule-based categories to the collected source and medium values.

Field	Question it answers	Example	Common mistake
Source	Where was the visit attributed?	`newsletter`	Using a channel name as the source
Medium	What kind of route brought it?	`email`	Inventing a new spelling for every team
Campaign	Which marketing effort?	`summer_launch`	Reusing one value for unrelated promotions
Default channel group	Which Google rule matched?	Email	Treating it as raw collected data
Source platform	Where was the traffic activity managed?	Google Ads	Assuming it always equals source

Google's traffic-source tagging reference maps manual parameters and advertising identifiers to Analytics dimensions. Keep the raw source/medium pair beside the channel group when investigating a classification. The pair explains what was collected; the channel explains which rule matched it.

Which acquisition scope should you use?

GA4 organizes traffic-source dimensions into user, session, and event scope. Google's scope documentation explains that prefixes such as First user and Session identify what the dimension describes. Omitting that prefix in a discussion makes otherwise correct numbers easy to misread.

First user scope

First user source / medium describes how a person was first acquired. It remains tied to that first acquisition as the user returns. Use the User acquisition report when the question is: which sources brought new users into this property's measured audience?

Session scope

Session source / medium describes what originated a new session. It can change when a returning user comes back through another campaign. Google's Traffic acquisition report is session-scoped and covers new and returning users.

Event scope

Event-scoped acquisition dimensions support attribution for key events. They are not interchangeable with session dimensions and may use a property-selected attribution model. Use them when allocating credit for an outcome, then document the model and lookback settings with the report.

What are the main GA4 default channels?

GA4's default channel group includes more than the six channels often repeated in older tutorials. Current categories include Direct, Organic Search, Paid Search, Organic Social, Paid Social, Email, Referral, Organic Video, Paid Video, Display, Affiliates, SMS, Mobile Push Notifications, Cross-network, and others. Google maintains the exact default channel definitions.

Channel	Plain-language meaning	Evidence to inspect
Organic Search	Non-ad search-result links	Session source, landing page, Search Console where applicable
Paid Search	Ads on search-engine sites	Ad integration, click ID, campaign dimensions
Organic Social	Non-ad links on recognized social sites	Source, medium, landing page, post URL outside GA4
Paid Social	Ads on recognized social sites	UTMs or integration plus platform campaign data
Email	Links classified as email	Consistent source, email medium, campaign
Referral	Non-ad links from other sites or apps	Referring domain and landing page
Direct	Directly attributed session under GA4 rules	Tagging gaps, redirects, consent, prior session context
Unassigned	No default channel rule matched	Raw source and medium values

Channel labels are classifications, not grades.

Why Direct traffic is frequently misunderstood

Google describes Direct as visits through a saved link or entered URL, but a Direct row is not evidence that every visitor typed the address. Missing or unusable campaign information can also leave attribution without the expected source. Google's tagging troubleshooting guide links unexpectedly high Direct and Unassigned rows to configuration problems, identity handling, and incomplete tagging. Investigate before celebrating an increase: review landing pages, redirects, email and document links, mobile-app handoffs, consent changes, cross-domain journeys, tag coverage, and campaign naming. A sudden spike concentrated on one URL or browser is a diagnostic clue, not proof of brand demand.

Direct is best treated as a residual attribution bucket with several possible stories. Its commercial meaning comes from context: cohort behavior, landing page, geography, customer status, and downstream outcomes. Changing the size of the bucket without changing those outcomes does not improve the business.

Build a durable UTM naming plan

Manual tagging adds campaign parameters to destination URLs. Google's campaign URL guide recommends the core parameters utm_source, utm_medium, and utm_campaign, with optional identifiers and content fields. If one parameter is set, Google recommends setting all relevant parameters to avoid incomplete reporting.

Use controlled vocabularies

Define approved lowercase values in a shared registry. For example, choose email rather than alternating between email, newsletter, and e-mail as the medium. Reserve the source for the publisher, partner, platform, or owned list. Give each campaign a stable name that survives creative changes.

https://example.com/guide?utm_source=customer_newsletter&utm_medium=email&utm_campaign=product_launch_2026_07&utm_content=hero_link

Never attach UTMs to internal navigation. In GA4, a new campaign or traffic source encountered mid-session does not start a new session or change that session's source. Google's campaign and traffic-source documentation says new values can attach to subsequent events and event-based attribution. Internal UTMs can therefore distort event-scoped campaign analysis. Use event parameters or content identifiers for internal promotions.

Test the final URL

Open the exact campaign URL through every redirect. Confirm that parameters reach the final page, the correct property receives the event, and the values appear in Realtime or DebugView before launch. After processing, Google's Manual report can show manually tagged campaign dimensions.

Auto-tagging and advertising integrations

Advertising integrations can add click identifiers and platform-specific dimensions. Auto-tagging generally provides richer advertising context than manual tags alone. Do not overwrite click attribution casually by mixing incomplete UTMs with an auto-tagged URL; Google notes that manual parameters can take precedence when a click identifier cannot be used as intended. Compare GA4 with the ad platform using compatible definitions and time zones. Clicks, users, sessions, events, and key events are different units. A discrepancy is not automatically a tracking failure, and identical totals are not the goal.

Cross-domain and referral hygiene

Journeys that move between an owned site, checkout, booking system, or application domain need consistent measurement. Without correct setup, the second domain can create a new user or session, or an owned domain can appear as a referral. Google's cross-domain measurement guide explains how the linker parameter carries identifiers between configured domains. Verify that redirects preserve the _gl parameter and that each participating page uses the same web stream tag. Add an unwanted referral only after understanding the journey; hiding a domain from referral attribution does not repair missing cross-domain identity.

How should controlled QA traffic be labelled?

Controlled visits can help test an owned page, tag, redirect, source label, or passive event. They are not customer acquisition and should not be disguised as Organic Search, social engagement, or brand Direct. If the approved test supports a tagged landing URL, use a clear label such as utm_source=traffic_creator, utm_medium=qa, and a unique campaign. Keep the test away from pages serving live ads or impression-based monetization. Do not simulate form submissions, account creation, cart additions, checkout, purchases, or ad clicks unless the action is explicitly authorized and contained in a test environment. Exclude QA traffic from leads, sales, revenue, engagement benchmarks, marketing efficiency, and organic-growth claims.

The specific-page QA guide provides a safe pilot template. Use the GTM and GA4 testing guide for browser-level validation and the delivery reconciliation checklist when counters disagree.

Diagnosing Direct, Unassigned, and (not set)

Direct is larger than expected

Segment by landing page, hostname, device, browser, country, and date. Check new redirects, untagged email links, documents, mobile apps, tag changes, consent behavior, and cross-domain configuration. Compare the change date with releases and campaigns.

Unassigned appears in the channel report

Inspect Session source / medium. Unassigned means the values did not match a default channel rule. Standardize the medium, complete the relevant UTM set, and validate against Google's current channel definitions. Do not rename data merely to make the chart look balanced.

(not set) appears in a dimension

The required value was unavailable or incompatible for that row. Check scope compatibility, event timing, collection method, report dimensions, and tag configuration. A custom exploration with fewer dimensions can help identify which combination creates the missing value.

There is no universal healthy channel mix

A local emergency plumber, enterprise SaaS company, subscription newsletter, mobile game, and established marketplace should not share one channel benchmark. Market, brand maturity, repeat behavior, sales cycle, campaign timing, privacy choices, and business model all change the mix. Percentages copied from another site can turn legitimate concentration into a false warning. Judge concentration by resilience and economics, not by visual symmetry. Ask what happens if the largest source falls by half, whether another channel can replace qualified demand, and whether the source remains profitable after full costs. That produces a decision; a decorative 30/30/20/20 split does not.

A monthly traffic-source review

Confirm property, stream, time zone, attribution settings, and date range.
Review User acquisition for first-user sources and Traffic acquisition for sessions.
Export Session source / medium, campaign, channel, landing page, sessions, engaged sessions, key events, and revenue where appropriate.
Separate employees, developers, QA, partners, and controlled traffic from customer cohorts.
Investigate new Direct, Unassigned, self-referral, and (not set) patterns.
Reconcile major campaigns with ad, email, affiliate, or partner platforms.
Record tagging fixes and annotate the effective date.
Evaluate qualified outcomes and source economics before shifting budget.

In our reporting workflow, every anomaly is logged with the first affected date, landing pages, source/medium values, and the release or campaign active that day. This small incident record is more useful than a screenshot because the next reviewer can reproduce the diagnosis. The UTM tracking guide expands the naming workflow. For source quality rather than classification, use the traffic-quality framework; for commercial outcomes, move to the conversion-rate guide.

Frequently asked questions

What is a website traffic source in GA4?

Source identifies where traffic was attributed, while medium describes how it arrived. The exact meaning also depends on whether the dimension is first-user, session, or event scoped.

Is Direct traffic always someone typing the URL?

No. Saved and typed URLs are Direct examples, but missing campaign information, redirects, identity configuration, and other measurement gaps can also contribute. Diagnose the landing pages and collection path.

What is the difference between source and channel?

Source is an origin value such as a platform, partner, or newsletter. A channel is a rule-based category such as Organic Search, Email, Referral, or Paid Social.

Should every external link use UTM parameters?

Tag campaign links you control when manual campaign attribution is needed. Do not put UTMs on internal navigation, and do not overwrite a correctly configured advertising integration without a documented reason.

What is a healthy Direct or Organic percentage?

There is no universal percentage. Compare your own qualified cohorts over time, investigate attribution changes, and judge source concentration by resilience, customer outcomes, and economics.