The decision now runs through AI
The place where belief about a category forms has moved. Now it can be studied directly.
In the last twenty-four months a new layer has appeared in the high-involvement information environment: the AI-mediated layer. When a customer asks an AI model about kitchen knives, mattress brands, or B2B software, what it returns is now part of the decision context. Increasingly it is the first part, before the search results, before the reviews, before the social signal.
This is a measurable behavioural shift. In a controlled study of 1,503 people by the Nuremberg Institute for Market Decisions (2026), those using ChatGPT for a product decision were faster and more accurate than Google users, yet only 12% clicked through to an external website, against 73% of the Google users. The trust is already there at scale: in its 2026 survey of more than 9,000 consumers across nine markets, the Boston Consulting Group (2026) found a majority now trust AI to help guide their purchases, with generative-AI shopping use rising sharply through 2025.
And the spend is starting to follow. Kaiser and Schulze (2026), in Marketing Science, found AI-referred shopping traffic still small and converting below traditional channels, but Adobe's first-party analytics (Adobe, 2026) put AI-referred traffic to US retailers up roughly 393% year on year in early 2026, and converting better than other traffic. The behaviour moved first, and the money is now moving with it.
That click-through gap is the one that matters. When the model is the decision context, what it says about a category is the information environment for that decision. The customer never visits the site, never clicks the search result, never leaves a trace in the behavioural data the old methods watch. This is the shift: the layer that forms belief about a category has moved, and unlike a customer's private reasoning, it is sitting right there to be read.
This is the opportunity. The previous two points say the old instruments are losing their grip and that preference is made in the information environment. This point says where that environment now lives. The next section is what I want to do about it.
References
- Adobe. (2026). Adobe Analytics: AI-driven traffic surges across retail in early 2026. Adobe.
- Boston Consulting Group. (2026). Consumers trust AI to buy better: Brands must adapt. BCG.
- Kaiser, M., & Schulze, C. (2026). Frontiers: ChatGPT referrals to e-commerce websites: How do LLMs compare against traditional channels? Marketing Science.
- Nuremberg Institute for Market Decisions. (2026). ChatGPT vs. Google: How AI and traditional search compare in consumer decision-making.