You open Google Analytics, look at the traffic graph, and see three main columns: Direct, Referral, Organic. The numbers are there. But there's no understanding of where visitors actually come from and which channel works better.

Website traffic is like water flows: each comes from its own source. If you don't know which channel brings conversions and which just generates noise with visits but no results, your promotion budget is being poured into the unknown. This is especially acute in GA4, where attribution logic has changed compared to Universal Analytics.

We break down each traffic source in Google Analytics, without fluff, with technical nuances that are rarely explained in basic guides, and practical conclusions for website owners.

💡 Quick overview:

• What actually ends up in Direct besides manual URL entry: bookmarks, messengers, HTTPS→HTTP and desktop email clients
• How Referral differs from Organic Social in GA4 and where to find referral reports
• Why Organic doesn't show keywords and how to restore them through Google Search Console

A short video on the topic showing the difference between channels using real GA4 data:

How Google Analytics determines traffic source

In Google Analytics, traffic source is the answer to the question "where did the visitor come from". The system captures the referrer, the HTTP Referer header, and matches it against channel grouping rules. The logic is simple from there:

• If the referrer is empty, GA doesn't know the source and marks the visit as Direct.
• If the referrer contains a search engine domain (google.com, yandex.ru, bing.com), it's Organic.
• If it's any other site, it's Referral.

But the devil is in the details. GA4 uses three levels of attribution: User scope (the very first user source, locked in forever), Session scope (the source of each new visit) and Event scope (the source of a conversion event with data-driven model). Understanding this three-level logic is key to reading reports without illusions.

Direct, what's hidden behind "direct" visits

The classic definition of Direct: a visitor typed the site address into the browser bar. In practice, this category includes far more scenarios, and many of them have nothing to do with brand recognition.

What actually ends up in Direct:

• Manual URL entry and transitions from browser bookmarks, the classics.
• Transitions from messengers: WhatsApp, Telegram, Signal, Viber don't pass referrer. A link you sent to a client in a chat will become a Direct visit in GA4.
• HTTPS → HTTP: when transitioning from a secure site to an unsecure one, the browser doesn't pass the referrer. In 2026, there are almost no HTTP sites left, but the problem is relevant for legacy projects.
• Desktop email clients without referrer: Outlook, Thunderbird and some corporate clients. Visits from your own newsletter settle into Direct.
• Non-tracking link shorteners and mobile applications, links from native apps often come without referrer.

This is exactly why Direct is often overestimated. You see a spike in "direct" visits and think brand recognition is excellent. But in reality, half of them are transitions from email newsletters without UTM tags or HTTPS→HTTP referrer loss.

Practical conclusion: if Direct is growing but conversion is falling, you're probably not tracking part of your channels. Put UTM tags on links in messengers, check email clients and HTTPS redirects.

Referral, when traffic comes through links from other sites

Referral is a transition through a link from any other site that isn't a search engine. GA4 automatically assigns to Referral domains that aren't in the list of search engines.

Referral traffic is divided into two fundamentally different segments:

• Editorial links, placed voluntarily: an article on a niche blog referenced your material, a news outlet mentioned a study. Such visitors arrive "warmed up": they've already been told about you, they click consciously. Conversion of editorial referral is many times higher than random.
• Partner and commercial links: paid placements, guest posts, aggregators, directories. Traffic quality directly depends on platform relevance. A link from a niche forum in your area can bring targeted leads; a link from a garbage directory, visits and zero inquiries.

The key parameter for evaluating referral traffic is not absolute volume, but Engagement rate. In GA4, look at this metric for each referral source in the Traffic Acquisition report. Sessions with Engagement rate below 30 percent are most likely random or bot traffic. High Engagement rate with small volume is a sign of a quality channel.

A separate setting that's often skipped: in GA4 you can exclude domains from Referral so they don't create new sessions. For example, if you have a payment gateway on a separate subdomain, add this domain to the exclusion list: Admin → Data Streams → Configure Tag Settings → List unwanted referrals.

Organic, search traffic as a long-term asset

Organic traffic is visits from search engine results. GA4 automatically recognizes Google, Bing, Yahoo, Yandex, DuckDuckGo and other search engines. This is typically the largest traffic channel for most sites.

The main advantage of organic is the compound interest effect. An article written and optimized once continues to bring traffic after 6, 12, 24 months without additional costs. Paid advertising stops working the minute you turn off the budget. Organic lives as long as the page ranks.

What you need to know about organic in GA4:

• "Not provided" hasn't gone anywhere. Google encrypts search queries for logged-in users. In GA4 you still see the lion's share of keywords as "not provided". The alternative is Google Search Console: the Performance report shows real queries your site was shown and clicked on.
• Yandex and Bing, separately. GA4 recognizes most search engines automatically. If traffic from Yandex ended up in Referral, check if the search engine list in data stream settings is misconfigured.
• Organic Shopping and Organic Video. GA4 separates them into distinct subchannels. If you have product snippets in Google Shopping (free) or videos on YouTube, they'll end up here, not in regular Organic Search.

For small sites, the organic traffic strategy is not "get into top 3 for a high-frequency query", but to collect dozens of low and medium-frequency queries where competition is lower. Twenty pages, each bringing a few hundred visits per month, provide stable targeted traffic without heroic efforts on one mega-query.

⁉️🤔 Frequently asked questions

Why did direct traffic in GA4 spike sharply while organic dropped, even though positions didn't change?

Most likely the problem is in consent mode or cookie banner. If a user declines tracking cookies, GA4 loses referrer data and records the session as Direct. Check consent mode settings in Google Tag Manager and make sure data collection starts before cookies are declined. Another possible reason is site protocol change from HTTP to HTTPS without redirect setup: some external links lose referrer.

How does Referral differ from Social in GA4?

Social is traffic from platforms that GA4 classifies as social networks by default: Facebook, Instagram, LinkedIn, Twitter, Pinterest, TikTok and several others. Referral is transitions from sites not in the list of search engines and social networks. If a transition from Telegram ended up in Referral instead of Social, Telegram isn't in GA4's default list, and that's normal. The difference matters for analytics: Social more often works for awareness, Referral for direct conversions.

Can you separate branded organic traffic from non-branded in GA4?

There's no standard filter for this in GA4. But Google Search Console in the Performance report allows you to filter queries containing the brand name and queries without it. This gives an approximate ratio of branded and non-branded search traffic. For a more accurate picture, you can create a custom report in Looker Studio using Search Console data.

What is "source (not set)" in the GA4 traffic report?

(not set) means that Analytics received a session but couldn't determine the source. Often this is direct traffic with outdated tracking code, transitions from apps without markup, or technical failures in passing campaign parameters. If the (not set) share exceeds a couple percent of total traffic, check that the GA4 tag is correctly installed on all site pages.

How do you make sure UTM tags don't break channel statistics?

Use a single case in all tags (lowercase only), standardize source names: for example, always utm_source=telegram, not a mix of Telegram, telegram and tg. Check the Acquisition → Traffic acquisition report: GA4 will show how tagged transitions were distributed among default channel groups. If you see some UTM traffic ended up in the wrong channel, review how utm_medium values match GA4's standard channel groups.

What understanding traffic sources gives you in practice

Knowing the difference between Direct, Referral and Organic isn't a matter of terminology. It's a matter of money. When you see the real picture by channels, decisions become obvious:

• If organic is growing but there are no conversions, the problem is on the landing page, not in SEO.
• If Direct is off the charts after launching a newsletter, it's time to add UTM tags, otherwise analytics goes blind.
• If Referral suddenly brought transitions from one domain, someone linked to you, and this can be worked with: amplify, thank, negotiate permanent placement.

Open Reports → Acquisition → Traffic acquisition in your GA4 right now and run through the three main channels. Saw an unexpected source? Then you've already read this breakdown for a reason.