Data analysis Q4 2026

Study: How manufacturers get recommended in AI search.Measured, not guessed.

Every month, we ask ChatGPT and Claude real procurement questions from manufacturing and industry, then analyze every search and every source. Here are the key findings. The full study, with all charts and the methodology, is available as a free PDF.

~25,000search results analyzed per month
~4,000AI search queries per month
~600AI answers per month
8industrial sectors
Cover of the study on AI visibility in manufacturing and industry, Q4 2026
At a glance

AI search in manufacturing follows clear rules

  • Consistent across manufacturing

    From gearbox manufacturing to industrial refrigeration, the same patterns show up in every sector we studied.

  • Different from other industries

    Compared with 17 industries, AI relies more heavily on company websites in manufacturing than anywhere else. Advice from retail or IT doesn't apply here.

  • Not random

    Ask the same question several times, and the AI searches for the same terms again and again. What it does can be measured and deliberately influenced.

How we measure

Methodology

  • Questions from real data

    Developed together with our clients: from real search queries in their Google Search Console and, in some cases, from real inquiry emails sent by their prospects.

  • Checked against the business context

    Every question fits the company's portfolio, competitors and target groups, in line with its communication strategy.

  • Queried directly

    Through the OpenAI and Anthropic APIs, with no third-party tool and no user history.

  • Every source counted

    Every answer, every AI search query and every search result, around 25,000 a month, across eight industrial sectors.

How it works

Two ways into the AI answer

When a buyer asks AI for suppliers, it almost always searches the web, often several times. Whether a supplier then shows up in the answer is decided in one of two ways.

Prompt → AI searchesIn our October measurements: ChatGPT in 100% of answers, Claude in 95%

A

AI finds the supplier

Its website appears in the results of one of the AI’s open searches.

That takes a top position in search.

B

AI knows the supplier

ChatGPT searches the supplier’s website directly, using the command site:supplier.com.
Claude names it straight from memory.

That takes being known to the model.

The findings

What the data shows

91%
Your own website is what counts

of the sources ChatGPT links to for manufacturing queries are company websites.

That means manufacturer websites as a whole, including your competitors’. No other industry relies as heavily on company websites as manufacturing:

Our measurement, ChatGPT91%
Industrial & manufacturing81%
Real estate72%
Average, 17 industries56%
Consumer goods43%

Industry figures: OtterlyAI, The AI Search Index, July/August 2026, USA. Full ranking in the study.

66%
Being in the model is a competitive advantage

of ChatGPT’s search queries are targeted searches on websites the model already knows.

ChatGPT already has a shortlist in mind and uses the site: operator to search those companies’ websites directly. Claude goes even further: 62% of the suppliers Claude names didn’t appear in its search results at all.

If the model knows you, you don’t even have to show up in the open search results.

98%
Strong search visibility makes up for it

of answers named our client when ChatGPT found the client’s website in search.

When the website didn’t appear in any search, ChatGPT didn’t name our client a single time. In the search results, our clients ranked at a median position of 2 out of about 25.

Very strong search visibility largely makes up for not being known to the model, but only at the very top.

0.05%
LinkedIn plays no role in AI search

of the sources AI links to for manufacturing queries come from LinkedIn.

2 of 4,126 source links. Not one of the roughly 4,000 AI search queries contained the word LinkedIn.

LinkedIn reaches people. What you want AI to find belongs on your own website.

Get the full study for free

15 pages with every finding, how we arrived at it, and all the charts. The next issue comes to you automatically in January.

Get the study

The study is free. In return, we will send you the next issues of the study and occasional updates on AI search in manufacturing. You can unsubscribe at any time with one click.

All fields are required. You'll receive an email with a confirmation link. After that, you can download the study right away.

Cover of the study on AI visibility in manufacturing and industry, Q4 2026
  • What paying ChatGPT users see differently
  • What AI is really searching for in the background
  • How much of your website AI actually reads
  • What applies to everyone, and what you need to measure for your own company
For journalists

Quote the figures, use the charts

You’re welcome to quote the figures on this page with attribution: “eingenetzt, AI Search Study: Manufacturing, Q4 2026.” We’ll gladly send you charts to embed. The study is published every quarter; the next issue comes out in January 2027.

Dennis Redder
Mechanical engineer (Dipl.-Ing.), founder of eingenetzt
+49 231 13707678
info@eingenetzt.net