Semrush (2026) — AI search and manufacturing SEO

Summary:

This report from Semrush, an SEO and marketing software company that sells search, traffic and AI visibility tools, looks at how AI search is changing the way manufacturing brands get found online: how often Google shows AI Overviews, which brands and sites AI answers mention or cite, and whether that turns into traffic.

The study combines three Semrush datasets covering January–July 2026: 458 manufacturing and industrial keywords, tracked monthly to see how often Google showed an AI Overview (weighted by search volume), U.S. clickstream data on traffic channels to sites in ten manufacturing categories, and Semrush AI Visibility data on U.S. manufacturing domains across ChatGPT, Gemini, AI Mode and AI Overviews: the top 25 domains by mentions and by citations each month, compared with each domain's AI traffic.

Key findings:
• AI Overviews are spreading fast. They appeared on 57% of tracked search volume in July, up from 38% in January, including product and brand searches.
• AI still sends very little traffic. AI assistants and AI Mode brought 0.48% of all sessions. Direct and organic search together made up nearly 80%.
• Mentions and citations are different lists. Only 2 of the top 15 domains appear on both. Legacy brands like 3M, John Deere and Boeing get mentioned most; the most-cited sites are reference sites, marketplaces, distributors and directories.
• AI platforms cite different sources. Engineerfix.com is ChatGPT's top-cited domain but doesn't appear in the top-cited sources on Google's AI platforms.
• Citations don't reliably bring clicks. Grainger, made-in-china.com and rsc.org get more AI traffic than heavily cited reference sites. Semrush suggests people click when the page offers a next step, like buying a product or reading original research.

Margin notes:
This is a vendor report built on Semrush's own tools, so the data can't be checked independently. The keyword set starts from top domains in Semrush's rankings, and the three datasets measure different things, so the results give a directional view rather than one comparable sample. The explanations of why people click are Semrush's interpretation, not something the data measures.