Jawdah Industrial: a Saudi manufacturer's platform rebuilt for SEO, GEO and AI discovery

GEO architecture

B2B buyers ask AI assistants which manufacturer makes a product, to which standard, and who supplies it locally. The platform answers each of those questions on a dedicated page.

• Five product families, their product groups and individual products each have their own Arabic and English page, so an answer can cite the exact product rather than a generic catalogue.
• Applications pages (roofs, building envelopes, ceilings, partitions, wet areas and more) connect buyer problems to the products that solve them.
• Branch pages give each location its own entity with address and contact details.

SEO architecture

• A sitemap index with six segments — pages (54), products (132), product groups (64), industrial services (30), projects (32) and branches (16) — 328 URLs in total at verification.
• Arabic and English alternates on every page through hreflang, with self-referencing canonicals.
• 289 legacy WordPress URLs were each mapped to a decision (keep, redirect or retire) and checked live on launch day.

AEO implementation

• Product and service pages lead with what the product is and where it is used before the specifications.
• Questions buyers ask are answered in visible FAQ sections on the pages that own them.

Entity work

• One organisation entity — Jawdah Industrial Co. Ltd. — referenced by every product, project and branch.
• Consistent product naming across Arabic, English, schema and llms.txt.

Schema

• Organization and WebSite for the company; LocalBusiness with PostalAddress for each branch.
• Product and BreadcrumbList on catalogue pages; FAQPage only where the questions are visible.

Content architecture

• Products → product groups → products, plus applications, industrial services, projects and branches as parallel hubs.
• Arabic and English versions kept at parity, with internal links between related products, applications and projects.

Technical implementation

• Next.js with incremental static regeneration so pages are served from cache and refreshed from the CMS.
• Catalogue, projects and branches managed in FeelStack; images served through ImageKit with versioned URLs so replaced photos refresh reliably.
• Blue/green releases with a retained rollback slot.

AI crawler readiness

• At launch, eight AI and search crawler user agents (including GPTBot, ClaudeBot and PerplexityBot) returned 200 on all seven tested paths after an overly broad edge bot rule was corrected.
• /llms.txt lists the company facts and the same URLs as the sitemap.

Search and AI measurement method

• Tag Manager limited to the production hostname so QA traffic is not counted; lead events for forms, phone, email and WhatsApp carry no personal data.
• Search Console tracks organic search; AI visibility will be measured as a separate, dated prompt dataset.

What we built

• The complete bilingual website that replaced the WordPress site, including the catalogue, projects, branches and industrial-services sections.
• The migration map, sitemaps, schema, llms.txt and analytics described above.
• The release pipeline, with a “Powered by DFeelings” credit in the site footer.