The Maze: AI search is not hitting every industry with the same force. In NP Digital's June 2026 visual, healthcare leads the shift at 51.6%, while education and media/publishing both sit at 39.0%. Ecommerce is last at 14.0%. That is the useful tension. Commerce is not immune to AI search. It is just harder to compress a shopping decision into one answer when price, inventory, reviews, shipping, returns, and trust all have to line up.
The fastest-moving categories are the ones where authority already matters. Healthcare at 51.6% is more than 3.6x the ecommerce reading. Education and media/publishing both sit at 39.0%, with finance close behind at 32.1%. These are information-heavy categories where a user often wants a credible answer before they want a transaction. That makes them natural early targets for answer engine optimization and generative engine optimization. The finding also fits broader research showing AI search has become a real distribution layer, not just a UI tweak. One 2026 study ran 24,000 queries across 243 countries and found Google AI Overview exposure expanded sharply from 2024 to 2025.
Ecommerce is the laggard because shopping has more moving parts than answering. The 14.0% ecommerce reading should not be read as safety. It is a reminder that commerce depends on execution data that AI systems still need to trust: product attributes, availability, delivery dates, returns, seller reputation, price movement, and post-purchase risk. A health or education query can often end in a summary. A purchase still needs a basket. That gives retailers a temporary moat, but a narrow one. Once product feeds, reviews, merchant policies, and inventory become easier for AI agents to read, the bottleneck moves from ranking to being selected.
Classic SEO strength may not transfer cleanly into AI answers. A separate empirical study of Google Search, Gemini, and AI Overviews found AI Overview source sets can differ materially from traditional search results, with average source-overlap below 0.2. Another 2026 study found nearly 30% of AI Overview cited domains did not appear in the co-displayed first-page results. That matters for commerce because many teams still treat generative visibility as "SEO plus schema." It is more awkward than that. AI systems reward answerable evidence, entity clarity, citations, and third-party trust signals. A product page built only to rank may be a poor source for a model trying to recommend.
The real commerce impact may arrive before the click. Neil Patel's post text says AI visitors from 60 B2B and B2C campaigns converted faster, converted at a higher rate, and spent more. The visible comments asked the right question: are those visitors better because AI sends better prospects, or because users finish discovery and comparison before they arrive? Either way, the implication is uncomfortable. Retailers may see fewer casual sessions, but more compressed intent. The top of funnel moves into AI answers, comparison prompts, and recommendation surfaces. The site becomes the closing room, not the discovery floor.
Why it matters: Ecommerce teams should not overreact to the low 14.0% reading. The smarter conclusion is sequencing. Healthcare, education, and media are already deep into the authority fight. Commerce is next, but it will be judged on structured product truth, credible third-party proof, and operational reliability. The old search game asked: can you rank? The AI commerce game asks: should the model trust you enough to recommend you?
