What did we find auditing AI visibility for industrial manufacturers?

Across audits of specialty manufacturers and industrial distributors, the majority were absent from AI answers in their own core category — including several that have led their niche for decades. Where a company was present, it was frequently surfaced through a third-party directory listing rather than its own content, meaning the engine knew the company existed but was quoting someone else about it.

Finding 1 — Absence is the norm, and it is not about company quality

Most audited companies did not appear in the answer to at least one query describing exactly what they make. Several were absent from all three. This included companies that have operated in their niche for fifty years or more, hold the specialist reputation in their category, and are the firm a knowledgeable buyer would name first.

In their place, engines named between six and ten competitors per query — usually a mix of the largest players in the category and mid-size firms that publish substantial engineering content.

Finding 2 — The split between "recommended" and "quoted"

This was the most actionable pattern, and it is invisible if you only log whether you were mentioned.

Several companies were recommended as suppliers on "who makes X" queries, but absent from the technical queries in the same category — the sizing questions, the selection tradeoffs, the spec guidance. On those, engines quoted competitors' calculators, charts and guidance pages.

The practical reading: the engine knows you make the product, but does not treat you as the authority on it. That gap is the whole opportunity, because the technical query is the earlier one. The buyer asking how to size the part has not yet decided who to RFQ.

Finding 3 — Citations run through directories more than through vendor sites

Where sources were visible, they skewed heavily toward third-party pages: trade directories, distributor catalogs, association listings and comparison content. Some companies appeared in an answer only through a directory citation — present, but with the directory's thin description standing in for their own positioning.

This is the finding that shapes how we scope work. Publishing on your own domain does not touch a directory-dominated citation pattern. Both moves are required, and the audit is what tells you the ratio.

Finding 4 — Published engineering data is the single strongest predictor

The companies that showed up on technical queries had one thing in common: they publish real, structured engineering data on the open web. Size charts as HTML tables. Selection guides. Sizing calculators. Material and tolerance references. Not gated, not in a PDF, not rendered as an image.

Almost every audited company possessed comparable data. The ones getting cited had published it in a machine-readable form. The ones absent had it in catalogs, spec sheets behind a form, or on paper.

Finding 5 — Category leadership does not transfer automatically

The most striking cases were companies that effectively defined a product category and were absent from the answer to "who makes [that category]." Decades of reputation, well-known within the industry, invisible to a model assembling an answer from public text.

The reassuring inverse: because position tracks published content rather than incumbency, these positions are winnable by whoever does the work. That cuts against you today and for you once you start.

What we take from it

• Run the citation trace before writing anything — the source mix determines where effort goes.
• Technical and selection queries are usually a bigger opportunity than "who makes X" queries, and are less contested.
• Getting existing engineering data out of PDFs is often the highest-return first move, ahead of writing anything new.
• Directory and distributor listings deserve real attention, because in many categories they are the cited source.