Attribution When The Journey Ends In A Chatbot
Attribution has always been a convenient fiction, but it was a shared one. Everyone agreed to pretend the last click caused the sale, and budgets were allocated on that agreement. Assistants break the agreement rather than improving it.
The mechanics are simple and unforgiving. A conversation that ends in a recommendation passes no referrer, carries no campaign parameter, and reports nothing back about what was considered or why.
What Attribution Loses From The Dashboard
Not the sale. The sale still shows up, filed under direct or organic, which are the two buckets where unexplained revenue goes to be ignored. What disappears is the reason, and the reason is what the next budget is argued from.
The uncomfortable version: a brand can be the most recommended option in its category inside the assistants and see nothing in its reporting that says so.
Direct traffic used to mean somebody typed your name. Increasingly it means somebody asked a machine and it said yours.
What Actually Helps
Three things, none of them a dashboard. Track share of recommendation by asking the assistants the questions your customers ask, on a schedule, and recording what comes back. Watch branded search volume, which still moves when an assistant names you. And treat unexplained direct revenue as a signal to investigate rather than a rounding error.
The Part Nobody Will Like
This rewards brands that were already strong, because assistants lean on what is well documented and widely cited. The measurement gap is real, but the deeper shift is that visibility inside an answer is earned over years rather than bought in a quarter. Google documents its assistant surfaces on its own blog, and none of them return a referrer.
More from The AI Shift.