673,629 AI search impressions for a flame-retardant fabric manufacturer

Client background

This manufacturer makes inherently flame-retardant knitted and woven fabrics, plus FR curtains and bedding built to NFPA 701 and similar standards. About 90% of output is exported to Europe and North America, so most buyers search in English and other European languages before they ever send an inquiry.

The company sells into specification-driven markets: upholstery, hospitality, transport and protective textile buyers who compare standards, fiber behavior and test data. That made generative AI search a natural place for its content to be read, because buyers ask assistants direct questions about flammability and fabric choice.

Why generative AI search mattered

Buyers in this category do not only type a product name. They ask AI assistants whether polyester burns, which fabrics pass fire codes, how limiting oxygen index works, and what to specify for fire-resistant sleepwear. The pack shows that demand clearly: the client's first-page queries include "is polyester flammable" at average position 4.1, "fire resistant fabrics list" at 4.3, and German and Portuguese phrases such as "feuerfeste bettwäsche" and "tecido jersey". Those are comparison and how-to questions, exactly the kind that AI answers summarize before a buyer contacts a supplier.

What RAGSEO did

RAGSEO ran a GEO program with the SEO content base underneath it, using the standard method:

• Defined the query scope around the questions buyers actually ask: flammability of common fibers, fire-resistant fabric lists, bedding and sleepwear standards, and test terminology.
• Built a GEO knowledge base from the client's own product materials, standards references and test data, so drafts stayed accurate.
• Created EEAT-compliant articles for those target questions, drafted with the RAG system and reviewed by professional editors, covering themes that later became the most-cited pages: non-flammable fabric, whether polyester is flammable, and a deep comparison of flame-retardant fabrics.
• Applied content and technical optimization: Schema markup, loading speed and mobile adaptability, so AI systems could parse the pages cleanly.
• Distributed content across 20+ authoritative global platforms and the client's own site, then tracked citations with regular monitoring.

Results: AI search visibility

In Google AI features over the last 3 months, the client's content was cited repeatedly. The three most-cited pages were:

Across that same 3-month window, the account recorded 673,629 AI search impressions. The pattern is consistent with the query scope: one page answers the core material question, one handles the most common fiber question, and one works as a comparison list. Together they cover the three shapes of AI answer a buyer sees before shortlisting suppliers.

Results: classic search alongside

Classic search moved too. Over the 52 weeks ending 2026-08-17, the site recorded 81,737 clicks and 9,188,301 impressions, a 4.0x growth multiple versus the starting period. Weekly averages went from 782 clicks and 66.7K impressions in the first quarter to 2,492 clicks and 297.2K impressions in the last quarter. Semrush's US organic database shows the same direction: 408 organic keywords in Jul 2024, 740 in Aug 2025 and 2331 in Aug 2026, with estimated monthly organic traffic rising from 29 to 318 to 2686.

Honest reading and next step

The AI citation numbers are strong, but they sit on a narrow base: three pages carry most of the citations. Positions are also mixed. Some queries rank in the top 3, while others sit at 8.1 or 8.3, and the weekly peak of 2,825 clicks at week 46 shows the trend is not a straight line. One boundary matters here: ChatGPT answers either from live web search, which GEO can influence, or from knowledge stored in the model without web access, which cannot currently be optimized.

The next step is to widen the citation base beyond the three lead pages, pushing mid-page queries such as "fabric pilling test" and "feuerfeste bettwäsche" toward the top 3, and to keep monitoring citations after targets are reached.