Manufacturing marketers watching AI search have a new number to process: across January through July 2026, AI assistants and Google’s conversational search surfaces sent just 0.48% of sessions to manufacturing and industrial sites, according to a Semrush study published September 8.
The finding might sound like permission to ignore AI. The same study shows why that would be a mistake: AI Overviews reached 57% of tracked manufacturing search volume by July, up from 38% in January.
The traffic picture
Semrush clickstream data across ten industrial categories shows direct visits at 56.65% and organic search at 22.28%. Together they account for nearly 80% of sessions. AI assistants contributed 0.45% and Google AI Mode 0.03%.
The stable-looking mix hides a shift. Direct traffic is growing while organic search declines. Semrush says demand for industrial goods has not visibly changed; the route to a supplier has.
Mentions vs. citations
The more important split is between being named in an AI answer and being linked as a source. Among the top 15 domains on both lists, only Vevor and Grainger appear twice.
Boeing recorded 771,000 mentions across ChatGPT, Gemini, Google AI Mode and AI Overviews in figures shared with press, ahead of 3M and John Deere. Yet Boeing does not appear among the most-cited sources. Meanwhile, engineerfix.com led citations in ChatGPT but ranked far below on other platforms.
Semrush’s explanation is structural: legacy brand recognition earns mentions, but AI citations reward the page that best answers the query. That is often a directory, distributor, marketplace or reference publisher, not a manufacturer’s own domain.
The “dark SEO funnel”
Gaetano DiNardi describes the measurement problem as the “dark SEO funnel.” A buyer may ask ChatGPT for supplier options, see a brand named, then later type that brand directly into a browser. Analytics record direct traffic, not AI influence.
This matches Semrush’s survey of more than 600 B2B professionals, where buyers reported using AI to build vendor shortlists and inform purchases above $1,000, then following up through search or direct visits. Earlier PPC Land coverage of a related survey found 519 US respondents said AI shaped 92% of vendor shortlists, while brand recognition determined notice in an AI answer for only 7%.
What to do about it
Use a three-part AI visibility scorecard instead of one traffic metric:
- Mentions: Is your brand named in relevant AI answers? This is a recognition play.
- Citations: Do AI systems link to a page that actually answers the query? This is a content and distribution play.
- Sessions: Which cited or mentioned pages produce real visits? Citeable reference content may earn links but not clicks.
Engineerfix is the cautionary example: it dominates ChatGPT citations but barely registers on Gemini, AI Mode or Google AI Overviews, and it captures only a fraction of AI referrals because its guides can be fully summarized inside an answer. Made-in-china.com and Grainger pick up more AI traffic because citations land on catalogue or product pages where the next step is an order.
For industrial teams, the operational takeaway is to stop asking “how much AI traffic did we get?” and start tracking “are we mentionable, citable and clickable in the answers that matter?” Those are now three separate jobs.
Source: PPC Land
