AI sends 0.48% of sessions to US manufacturing sites. That is not the whole story

What Semrush measured

Semrush, which sells SEO and AI visibility software, studied US sessions to sites in ten manufacturing and industrial categories from January to July 2026. Across those sites, 0.48% of sessions came from AI search tools or assistants.¹ Direct and organic search together made up nearly 80% of sessions.¹

The same study tracked 458 manufacturing and industrial keywords drawn from the top 20 manufacturing domains. AI Overviews appeared on 57% of that tracked search volume in July 2026, up from 38% in January.¹ So AI answers sit on top of most of the searches, while very few visits arrive from AI tools directly.

US manufacturing, January–July 2026

Few AI visits, many AI answers

0.48%of sessions to US manufacturing sites came from AI toolsSemrush, Jan–Jul 2026 ¹

57%of tracked manufacturing search volume showed an AI Overview, up from 38%Semrush, Jan–Jul 2026 ¹

2 of 15domains on both the most-mentioned and most-cited listsSemrush, Jan–Jul 2026 ¹

Why low sessions understate AI's role

Sessions count only the visits that come from an AI tool's link. An AI Overview that answers an engineer's question on Google, and names three suppliers, counts as organic search if the engineer clicks, and as nothing if they do not. A ChatGPT answer that names a brand without linking leaves no trace in analytics at all.

Links are scarce in ChatGPT answers. Similarweb's US data shows ChatGPT prompts carrying a citation rising from about 1.6% in June 2025 to about 6.8% in May 2026.² Its professional services figure was under 4%, against about 23% for travel.² Industrial buying questions are more likely to resemble professional services than travel, so most mentions will not link.

Our view: for a manufacturer, the referral share is the wrong headline. The question is whether your company is named, and described correctly, when a buyer asks which suppliers make a part to a given specification.

Mentions and citations are different lists

Semrush compared the 15 brands AI mentions most with the 15 domains it cites most. Only two names appeared on both lists, Vevor and Grainger.¹ Being talked about and being used as a source are, in this data, mostly different things.

Many of the cited sources were intermediaries. Semrush's list includes industry directories such as Thomasnet and GlobalSpec, a B2B marketplace, distributors, an engineering reference site and a professional body, alongside a few manufacturers' own sites.¹ Those are many of the pages AI engines read when they answer manufacturing questions.

For many manufacturers this means the directory profile and the distributor's product page may do more for AI answers than the company's own home page. Our synthesis of what AI engines cite by industry shows the same pattern of intermediaries in other B2B sectors.

How this compares with cross-industry benchmarks

The low referral share is not unique to manufacturing. Conductor's benchmark, published in November 2025, put AI referrals at 1.08% of visits across 1,215 of its US enterprise customer domains in ten industries, with IT at 2.80%. That data covers May to September 2025, so it is nearly a year old.³

The AI Overview rate looks high by comparison. In Conductor's sample of 21.9 million US searches (mid-September to mid-October 2025), AI Overviews appeared on 25.11% of searches overall, with healthcare at 48.75%.⁴ Different keyword sets and periods mean the figures cannot be compared directly. Still, a 57% rate on tracked manufacturing volume suggests industrial queries, often technical and informational, attract AI Overviews often. Our note on reading benchmark dates explains why the period matters as much as the percentage.

What should manufacturers check first?

Start with the sources AI already reads, then your own pages. Our view: this order fixes the largest share of answers for the least effort.

1. Directory and marketplace profiles. Check your listings on the directories in your category for capabilities, certifications, materials and locations served. Correct anything outdated.
2. Distributor product data. Make sure distributors carry your current part numbers, specifications and datasheets.
3. Specification pages on your own site. Publish tolerances, materials, standards met and lead-time ranges as text, not only inside PDFs.
4. Buying-question tests. Ask the engines the questions your buyers ask, such as "who manufactures stainless fasteners to a given standard in the United States", and record who is named and cited.

Limits of the study

An average is not your site. Small and mid-sized manufacturers have different traffic mixes, and the study's session figure is one number across ten categories. The keyword set is Semrush's choice of 458 terms, so the AI Overview rate describes that set, not all industrial searches.¹ And the mention and citation lists come from Semrush's own tracking, which does not cover every engine equally.

None of that changes the direction. AI answers already sit on top of much of the search demand that reaches manufacturers, and the sources they cite are mostly not manufacturers' own sites. Our guide to why referral studies disagree covers how to read AI traffic shares alongside that.