Four decades of steel expertise that buyers — and now AI assistants — couldn’t find.

The problem: technical buyers search by grade, not by brand

Steel procurement search is unusually precise. A buyer does not search "steel supplier"; they search "C45 steel strip supplier", "EN42J hardened and tempered strips", "CRCA DC04 sheet", or "50CrV4 spring steel". Each grade is effectively its own market with its own buyers, and KND had a single catalogue page trying to serve all of them.

Three further gaps compounded it. The site carried no content explaining grade selection — the exact question a sourcing engineer asks before shortlisting a supplier. Industry-specific pages were missing, so automotive, defence, railway and tooling buyers all landed on generic copy. And despite being a manufacturer with a physical facility, the business had no meaningful map presence for buyers searching locally for suppliers.

Then there was the newer problem. Sourcing engineers increasingly ask an AI assistant before they open a browser tab: "what is EN8 steel used for", "C45 vs EN9 for springs", "which grade for friction plates". Those answers named other suppliers, or no supplier at all.

What we did

Grade-level architecture

• Individual, technically credible pages for the grades KND actually supplies — C10/SAE 1010, C20, C40/EN8, C45/S45C, C55/EN9, C60, C65, C75, C80/EN42J, C98/EN44D, SK4, SK5/SK85, 50CrV4/SUP10/EN47 and the CRCA family
• Each page carries specification, typical applications, available dimensions and the questions a sourcing engineer asks before enquiring
• Structured so that the long tail of grade + form + application searches has somewhere relevant to land

Industry pages for six buyer types

• Dedicated pages for automotive and OEM, electrical and defence, medical and railway, hand tools and friction plates, textile machinery, and industrial engineering
• Each speaks that sector’s requirements and tolerances rather than repeating generic capability claims
• Credibility signals placed where procurement looks: since 1981, 10,000+ MT supplied annually, 500+ customers, 95% repeat rate, audit ratings

The AI layer — getting cited when engineers ask an assistant

• Opened crawler access for GPTBot, PerplexityBot, ClaudeBot and Google-Extended, and published an llms.txt describing the grade catalogue and authority pages
• Rewrote the highest-value technical questions as citation blocks of 130–170 words — grade comparisons, hardness and tempering ranges, application suitability — the self-contained format AI systems lift
• Added Organization, Product and FAQPage schema so machines can resolve what KND makes, in which grades, for which applications
• Built a fixed panel of real buyer questions and re-ran it against Google AI Overviews and ChatGPT each quarter to measure actual appearances

Local and B2B visibility

• Google Business Profile built out properly for a manufacturer — categories, facility photos, product lines and service areas — producing map presence for buyers searching locally
• Consistent business details across B2B directories and industrial marketplaces where sourcing teams verify suppliers
• Technical SEO cleanup so a large catalogue could actually be crawled and indexed grade by grade

Results

The takeaway

B2B manufacturing SEO fails when it is written for a general audience. It works when the site answers precisely what a technical buyer types — a grade, a form, an application — and proves competence at that level of detail. That same specificity is what makes content quotable to an AI assistant: a clear, self-contained answer about EN8 versus C45 is exactly what an AI Overview wants to cite. KND had four decades of genuine expertise; the work was making it legible to search engines, AI models and sourcing engineers alike, one grade at a time.