The Maze: AI referrals are still small enough for most dashboards to ignore. NP Digital's June campaign sample argues that may be the wrong lens. Across 60 campaigns and 828 B2B/B2C cohorts, AI-referred visitors converted at 5.97% versus 0.72% for traditional visitors, reached conversion in 3 days instead of 8, and generated $18.04 per visitor versus $2.56. Tiny channel. Large signal. The buyer may already be half-sold before the click.
AI traffic looks less like discovery and more like pre-sales. The strongest number is not only the 8.3x conversion gap. It is the three-part pattern: higher conversion, faster decisioning, and higher revenue per visitor. Traditional visitors converted at 0.72%; AI-referred visitors converted at 5.97%. Traditional visitors took 8 days; AI-referred visitors took 3. Traditional visitors produced $2.56 per visitor; AI-referred visitors produced $18.04. That is not a normal top-of-funnel traffic story. It looks more like a recommendation layer doing qualification work before the session starts.
The intent caveat matters because it changes the strategy. The visible debate under the LinkedIn post is the right one: are AI referrals better because AI is better traffic, or because the people using AI are already asking sharper questions? A shopper typing "best running shoes under $120 for flat feet" into an assistant is not the same economic unit as a user scanning generic blue links. For operators, that means AEO is not just SEO with new labels. It is product proof, comparison logic, availability, reviews, pricing clarity, and answer-ready content.
Low volume can still be worth disproportionate attention. A channel that sends fewer visits but produces 7x revenue per visitor deserves a different dashboard. Session share will understate the channel if the next click is already loaded with intent. The more useful questions are: which AI answers mention the brand, what evidence gets cited, what product pages receive the visit, and how quickly those visitors move from comparison to purchase or lead capture.
The risk is over-reading a campaign sample. NP Digital's source line says the data covers 60 campaigns and 828 B2B/B2C cohorts, but it does not disclose vertical mix, traffic definitions, attribution windows, or matching rules. If AI-referred traffic is overrepresented in high-intent categories, the performance gap may partly reflect who asked the question, not just where the answer came from. That does not weaken the finding. It makes the real takeaway sharper: brands need to separate AI visibility from AI-qualified demand.
Why it matters: Search strategy is moving from ranking pages to qualifying decisions. In classic SEO, the job was to win the click. In AI-mediated discovery, the job starts earlier: become the answer that survives the assistant's filtering. That pushes ecommerce teams toward cleaner product data, sharper comparisons, proof-rich pages, and measurement that values buyer quality over raw traffic. The old funnel loved volume. The new one may reward being selected before the visit begins.
