How last non-direct click works
A direct visit is one that arrives with no source attached: nothing in the link says where the person came from. Under a plain final-click rule, a conversion on such a visit is credited to direct, which is to say credited to nothing you can act on.
Last non-direct click steps over that visit. It looks back through the journey inside the lookback window, finds the most recent touch that did carry a source, and gives the conversion to that instead. Only when no identifiable touch exists anywhere in the window does the conversion fall back to direct.
The reasoning is sound. Typing an address or opening a bookmark is not a marketing channel; it is what somebody does after a channel has already done its work. Crediting the search or the ad that put you in their head is closer to the truth than crediting the act of remembering you. This has long been the default behaviour in analytics reporting, which is one reason analytics and ad platform numbers rarely agree.
Why last non-direct click matters
Without it, your best-performing marketing quietly leaks into a bucket labelled direct, and the channels that earned those sales look weaker than they are. That distortion grows with the length of the buying cycle, because the longer people take to decide, the more likely they are to return by typing your name rather than by clicking anything.
It also matters in Nepal and other messaging-heavy markets more than the model’s designers intended. Links shared in Viber, WhatsApp and Messenger often arrive stripped of any source, and a rule that skips those visits recovers credit that would otherwise vanish.
Common mistakes with last non-direct click
The serious one is reading a small direct figure as proof that tracking is healthy. This rule reduces direct by reassigning it, so measurement failures — missing tags, broken redirects, blocked identifiers — get hidden inside other channels rather than showing up as an obvious problem.
The second is comparing the result with an ad platform report and calling the difference an error. Google Ads and Meta credit their own clicks by their own rules and do not apply this one, so the two views will always differ even when both are working correctly.
How to act on it
Know which of your reports uses it and which does not, and stop treating the gap between them as a fault to be fixed. Then attack the underlying cause rather than the symptom: tag every campaign, email and shared link so fewer visits arrive anonymous in the first place. Building links with a consistent scheme — the UTM builder is there for exactly that — does more for accuracy than any model setting.
Watch the direct share over time as a health check. A sudden rise usually means a tag was dropped or a redirect started stripping parameters, and that is a job for a tracking audit rather than a reporting decision.
