Where does direct traffic really come from?

Direct traffic usually comes from two places: people who already know your address, and clicks that arrived without a readable source. Think links pasted into chats, untagged emails, PDFs and stripped redirects. If GA4 recognises a returning buyer, it often files the visit under their earlier campaign instead, so Direct keeps what it could not trace.

What one store's data shows

The usual answer is a picture of loyal fans typing your address. It is half right, and an export can show which half. One store's file is a good place to look, because Direct brings in more than half of its revenue.

One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.

On the Channels sheet, Direct takes 57.7% of revenue whichever view you pick: last click, first click or touched. On the Journeys sheet, every journey with 2 or more touches adds up to 20.6% of revenue.

Set those side by side. In this one store, Direct's share is far bigger than all the multi-touch revenue put together. So most of Direct's money must come from journeys of a single touch, where GA4 recorded one visit and nothing before it.

On the Journeys sheet, single-touch journeys hold 79.5% of revenue, with 0.5 days to buy. People who pay within hours of their first recorded visit had usually made up their minds before it. The deciding happened somewhere GA4 does not look.

Where the export does see more, journeys run longer. On the Journeys sheet, journeys with 2 to 3 touches hold 12.2% of revenue and took 12.5 days to buy. Those are the journeys in which you could read what came before a Direct visit, and they carry the smaller share.

So for most of this store's Direct money, the real source sits outside the file: a person, a message, a memory. The export can say how much revenue arrived that way. It cannot say who sent it.

Why does the typed-address answer mislead?

Because typing is only one door, and GA4 moves many typed visits out of Direct anyway. It gives a direct entrance the UTM values it already holds for that user. Google says the lookback window also applies to that session attribution. So a known buyer who once clicked a campaign, and types your address inside the window, usually wears that campaign's label.

What lands in Direct is the rest. Google's help lists a link someone sends that the reader copies and pastes into the browser. It adds saved bookmarks, traffic from social media, offline documents and HTTPS to HTTP hops. Its page on (direct) / (none) adds untagged links, plus redirects and URL shorteners that strip tags. Ad blockers count too, since they interfere with the cookies that identify a visit's origin.

Sort those by what really happened before the visit and you get a short map:

Most rows are a person or a message, not a typed address. A friend pasting your product link into a group chat is word of mouth with the label torn off. Email clicks and dark social each get their own answer.

The last row is not a visit at all. In GA4's attribution paths report, Direct is also the value used when there is no path data to credit, for example after a data import. Key events that arrive by import can sit under Direct with nobody behind them.

What can the Direct row not tell you?

It cannot tell a typed visit from a pasted link, or a regular from a stranger on a new phone. All of them arrive without a source and wear the same label.

It cannot tell you what started the journey. The real first touch, a recommendation or an ad seen without a click, happened before GA4's first record.

And it cannot tell you what a channel is worth. If an ad works by making people remember you, its results arrive later as Direct, with nothing in GA4 linking the two. A holdout test is the plainest way to see that link.

What to do this week

1. Size the revenue that arrives with no trail. In GA4, click Advertising, then Attribution paths under Attribution, pick purchase and note the total purchase revenue. Then set Path length to equal to 1 touchpoint and find the row that is Direct alone. Pass: that row holds less than half the total, so most of your money carries a named source somewhere on its path. Fail: it holds more than half, so most of your revenue starts and ends off the record.
2. Read ten orders in Shopify. In your Shopify admin, go to Orders, click an order number and click View conversion details under Conversion summary. Note the origin of each customer's first visit for ten recent orders. Pass: most first visits name a source, so the trail exists and tags can recover it. Fail: most show no source, so ask buyers how they found you.
3. Start one holdout this week. Pick the channel you suspect feeds Direct, such as paid social or a creator push. Pause it in a few regions for four weeks and leave it running everywhere else. Pass: Direct and total sales dip where it went quiet, so it was a real source. Fail: nothing moves, so look elsewhere before you pay it more.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: User acquisition report (Google); Understand (direct) / (none) traffic (Google); Scopes of traffic-source dimensions (Google); Select attribution settings (Google); Key events attribution paths report (Google); Get started with attribution (Google); Viewing order conversion summary (Shopify); URL builders: Collect campaign data with custom URLs (Google)

Related answers

• All Direct traffic questions
• How to trace where direct traffic comes from, step by step
• Direct traffic, defined
• Incrementality testing, explained