Last non-direct click is an attribution rule that gives all the credit for a conversion to the last marketing channel a person arrived from, skipping any later visits where they came to the site directly. If someone clicks a Google ad on Tuesday, then types your web address in on Friday and buys, the sale is credited to the Google ad rather than to direct.
How last non-direct click works
Direct traffic is the label analytics gives a visit when it has no referring site and no campaign information: a typed address, a bookmark, a link in a PDF or a messaging app, or a visit where the referrer was stripped along the way. Many direct visits are people coming back after finding you some other way. Last non-direct click assumes that earlier source deserves the credit, so it looks back past the direct visit to the last one that had a known channel.
A conversion is only credited to direct when there is no earlier source on record for that visitor. That makes the rule different from a plain last-click model, which would hand the Friday sale above to direct.
You meet the rule in several places:
- Universal Analytics used last non-direct click as the default in its standard reports, which is why it shaped how a generation of marketers read their channel figures.
- GA4’s traffic acquisition report applies the same idea to sessions: a session that starts direct is given the visitor’s previous non-direct source, where GA4 knows one.
- GA4’s last-click model in Attribution settings also ignores direct visits unless the journey contained nothing else, so in practice it is a last non-direct click model.
Ad platforms work differently. Google Ads only sees Google Ads clicks, so its own last-click model is really “last Google Ads click”.
Why it matters
The rule stops direct traffic from absorbing credit that belongs to marketing. Without it, a customer who found you through an email and came back via a bookmark would make the email look worthless.
The flip side is that it can over-credit whichever digital channel happened to come last. A plumbing firm whose vans and leaflets are seen all over south London will have plenty of customers who already know the name. If those people search the company name and click a brand ad before calling, last non-direct click gives the brand ad the job, even though the van did the persuading.
Consent adds another gap for UK sites. When a visitor declines analytics cookies, GA4 cannot link their later visits to the earlier one, so a return visit can show up as direct even though they first arrived from an advert.
Common mistakes
- Reading direct traffic as “people who typed in our address”. It is the leftover: visits with no source GA4 can identify.
- Comparing old Universal Analytics channel figures with GA4 and treating the difference as a change in performance.
- Letting brand search and remarketing take credit for demand that social, video or offline marketing created.
- Sending untagged links in emails and on QR codes, so those visits arrive as direct and the credit passes to whatever came before.
- Assuming every GA4 report uses the same rule. The user acquisition report credits the first source instead.
How to act on it
First, know which rule each report uses: traffic acquisition for sessions, user acquisition for first source, and the Advertising section for whichever model your property is set to. Then reduce the amount of traffic that arrives as direct by adding UTM parameters to every email, social post, partner link and QR code you control.
Compare the picture with data-driven attribution to see which channels tend to start journeys, and add a “How did you hear about us?” question to your enquiry form to capture word of mouth and offline marketing that no analytics tool can see.
Reading attribution rules correctly is basic hygiene in performance marketing, because the channel that looks cheapest under one rule can look expensive under another.