Dark Traffic

Dark traffic is the visits you cannot trace. Someone arrives, reads, perhaps buys — and your analytics records no source at all, filing them under "direct" as though they had simply known your address by heart. Direct was never really a channel. It is the drawer that everything unlabelled gets put into.

It is growing, and AI answers are much of the reason. Some assistants pass no note of origin when a person follows a link out of an answer. Some are applications rather than browsers, so there is nothing to pass. Privacy defaults strip the note in other cases. And often nothing is stripped at all: the person read about you inside an answer on Tuesday, thought about it, and typed your name into the address bar on Thursday. No mechanism could have joined those two events.

The danger is not the missing data. It is the conclusion drawn from it. An owner looks at a report showing nothing arriving from AI, decides the whole business of being readable by machines is hype, and stops doing the work that was quietly feeding the unlabelled column. The consequence appears months later as a decline with no visible cause, which is the hardest kind to diagnose and the hardest to reverse.

The honest response is to stop demanding one clean line of attribution and start reading several weak signals together: whether direct visits are trending upward, whether the number of people searching your name specifically is rising, what assistants actually say when asked about your field, and what customers tell you when you simply ask how they heard of you. No one of these is proof. Together they are enough to steer by.

The plain version: word of mouth was never measurable either, and no sensible owner ever concluded from that it was not happening. This is word of mouth conducted at machine scale — the same unprovable, decisive thing, now taking place in conversations you were never part of.