The advertising industry now has a framework for measuring how AI influences a purchase. It still does not have the data needed to make that framework work.
The Interactive Advertising Bureau’s new “AI-Era Attribution Blueprint” tries to define who created intent, who made the decision, and who executed the transaction when an AI assistant sits somewhere in the journey. But the more important point remains clear: recommendations, decisions, and even transactions can happen inside AI platforms without advertisers or measurement companies seeing the evidence.
The measurement problem starts inside the answer
The blueprint separates AI’s claims from the evidence needed to support them. That matters because a consumer can ask an AI assistant for a product recommendation, read an answer shaped by a publisher review, and then visit a retailer through search or direct traffic. The sale may be influenced by AI, while the credit goes somewhere else.
The numbers show how badly the current reporting can miss the behavior. Similarweb data cited by the IAB found that consumers receiving a brand recommendation from an AI assistant were 2.5 times more likely to visit that brand’s website within seven days. Yet only 8.8% of those visits came directly from AI referrals. Search received attribution for 55.9% of them, while direct traffic accounted for 19.9%.
That is not a small reporting wrinkle. It changes which channels appear to be creating demand, and therefore where marketing budgets go.
AI visibility is not the same as business impact
Publishers are already trying to measure parts of the new journey. Time is working with Mobian on ads served to AI agents, while Bustle Digital Group is working with Profound on measuring its generative engine optimization campaigns. Those efforts can show visibility, citations, or activity. They cannot yet reliably connect an AI answer to a completed purchase.
The same tension is emerging for brands. OpenAI is testing an analytics dashboard that could show how often a business appears in ChatGPT answers and how many clicks those appearances generate. Google is also expected to provide U.K. publishers with clickthrough data for AI search features under pressure from the Competition and Markets Authority. Both developments could improve visibility, but neither automatically proves that an AI interaction caused a sale.
That is why the IAB blueprint is best understood as a measurement vocabulary and a call for access, not a finished attribution solution. Publishers may be feeding the answers while lacking the receipts. Until AI platforms expose more of the path from recommendation to conversion, brands will keep measuring the traffic that can be captured rather than the influence that actually shaped the decision.