Background
This case study documents the AI visibility turnaround of a mid-market industrial manufacturer of specialty fluid-handling components for food and beverage, pharma, and biotech end markets. The company has approximately 400 employees, generates roughly $120 million in annual revenue, and sells through a combination of direct enterprise accounts, industrial distributors, and a small but growing direct e-commerce channel.
The company had strong traditional SEO and mature Google rankings for most of its product categories. It had won awards for its sales enablement content and was well-regarded among engineering buyers who already knew the brand. Despite all of this, an AI visibility audit in mid-2025 revealed that the company was effectively invisible in AI responses for its core category and specification queries.
The Starting Point
The baseline audit covered 30 prompts across ChatGPT, Claude, Perplexity, and Gemini, spanning specification queries ("stainless steel hygienic valve 2 inch"), supplier queries ("sanitary valve manufacturers for pharma"), certification queries ("FDA 21 CFR 177 compliant fluid handling"), and comparison queries ("[competitor] alternatives"). The company appeared in 2 of 30 prompts across all four platforms, or about 1.7 percent visibility. Its primary direct competitor appeared in 22 of 30.
The root-cause analysis identified four issues. First, the company's primary product catalog was a JavaScript-hydrated web app that AI crawlers could not parse. Second, its datasheets were PDF-only, with no HTML equivalents. Third, its robots.txt blocked several AI crawlers inherited from a 2023 policy decision that nobody had revisited. Fourth, its Wikipedia entry had been stubbed years earlier and never updated, leaving AI systems with outdated and incomplete entity data.
The Intervention
The remediation plan ran across three phases over nine months.
Phase 1 (weeks 1 to 6): access and entity foundation. The team updated robots.txt to explicitly allow GPTBot, ClaudeBot, PerplexityBot, MistralAI-User, and Google-Extended. They updated and expanded the Wikipedia entry with cited references to peer-reviewed validation studies, patent filings, and regulatory certifications. They completed the Wikidata entity with every supported property. They added consistent Organization schema markup to the homepage and every major product category page.
Phase 2 (weeks 7 to 20): structured product data. The team published HTML versions of 150 top-selling product datasheets, with complete Schema.org Product markup including manufacturer part number, regulatory certifications, material specifications, dimensions, and connected offer data. They rewrote the product catalog front-end to be server-rendered rather than JavaScript-only, making it fully parseable by AI crawlers. They added a dedicated certifications page listing every regulatory approval, standards compliance, and third-party validation, each with supporting documentation.
Phase 3 (weeks 21 to 36): content depth. The team published 20 technical whitepapers on application-specific engineering trade-offs, a comprehensive comparison page against each top-five competitor with feature-by-feature detail, and a buying guide for each target end market (food and beverage, pharma, biotech). They earned coverage in three major industrial trade publications through targeted editorial outreach.
The Results
After nine months, the same 30-prompt audit showed the company appearing in 19 of 30 prompts, or 63 percent visibility. On specification queries specifically, visibility reached 75 percent. Perplexity citation rate rose from essentially zero to consistent citation across about a third of relevant informational queries. ChatGPT presence grew more slowly (as expected given the training-data timeline) but was visible by month 9.
Downstream business metrics showed correlated movement. RFQ volume from the website rose 34 percent over the same period, with a disproportionate share tagged as originating from AI-platform referrals in customer discovery conversations. The sales team reported meeting with more buyers who arrived already educated about the company's specific differentiators, a shift the team attributed to AI-assistant-informed pre-meeting research.
What Made the Difference
Three interventions contributed disproportionately to the turnaround.
HTML product data: moving from JavaScript-only catalog to server-rendered HTML with Product schema was the single largest lift. The company went from being invisible to specification-query crawlers to being routinely parsed and cited.
Wikipedia completion: the updated Wikipedia entry propagated through AI training data over the course of six months, producing a visible step-change in ChatGPT and Claude presence around month 7 and 8 that coincided with those platforms' model update cycles.
Certifications page: compliance queries ("FDA approved", "ATEX certified", "NSF listed") are high-intent and disproportionately drove new buyer engagement. Publishing one dedicated, well-structured certifications page produced outsized citation lift for this query category.
Lessons for Manufacturers
PDF datasheets are invisible to AI. HTML equivalents are essential for industrial AI visibility. The company estimated that roughly 80 percent of the competitive advantage in specification queries came from this single change.
Robots.txt audits pay back immediately. Inherited policies rarely reflect current priorities. Any manufacturer that has not reviewed robots.txt with AI crawlers specifically in mind is almost certainly leaving visibility on the table.
Wikipedia is undervalued in industrial categories. B2C brands treat Wikipedia as routine. Industrial brands often neglect it. The compounding benefit across AI training cycles is too large to ignore.
Case-study-worthy results take 6 to 12 months. Training-data effects lag publication by one model release cycle. Retrieval effects can appear within weeks. Plan programs for a full model-release cycle, not a single-quarter campaign.
How Presenc AI Helped
Presenc AI ran the baseline audit, tracked monthly visibility progress across all four AI platforms during the intervention, and surfaced specific prompts where competitor citations were happening that the company was not. The platform's crawler-access audit identified the robots.txt misconfiguration in week 1. The schema-markup audit prioritized which product pages to convert to HTML first based on expected citation lift. For the company's marketing and engineering teams, Presenc provided the measurement infrastructure that made the 9-month program defensible to leadership throughout.