AI Visibility Analysis: Chemicals & Petrochemicals Industry

The Molecules Are World-Class. The Machine Readability Isn’t.

BASF. Dow. DuPont. ExxonMobil. LyondellBasell. Ineos. Covestro. SABIC. LG Chem. Formosa Plastics Group. Between them, these ten companies manufacture a meaningful share of the polymers, specialty chemicals, and petrochemical feedstocks that sit inside nearly every physical product a business buys or sells. Procurement teams, engineers, distributors, and increasingly AI research agents all rely on these companies’ websites to answer a basic question: what does this company actually make, and can I trust what I’m reading.

That’s the question I set out to answer, not from a brand perception angle, but from a structural one. When an AI system, ChatGPT, Perplexity, Gemini, or an enterprise procurement copilot, tries to identify, parse, and cite a chemical manufacturer’s corporate website, what does it actually find. I ran the ten largest chemicals and petrochemicals companies by global relevance through the AI Visibility Inspector using the Ivica Srncevic Framework, and the results are the tenth entry in this independent research series.

This is the eleventh installment in an ongoing series examining how major global industries perform under AI-driven discovery, following analyses of the legal industry, global pharmaceutical, SaaS CRM, global banking, industrial tools manufacturing, life and health insurance, automobile industry, commercial vehicle sector, hospitality and tourism, and global light vehicle manufacturers.

Key Takeaways

• Sector average: 57.6, Grade C – Fair, calculated across the nine companies the AI Visibility Inspector was able to fully evaluate. Not one company in this dataset reached Grade B or A.
• LG Chem leads at 69, the closest any chemicals major came to good-band visibility, driven by the highest Schema score in the dataset at 70.
• SABIC could not be evaluated. The company’s global site took 17 seconds to load and the AI Visibility Inspector was unable to extract data at all, a distinct and more severe failure mode than a low score.
• Freshness is the sector’s collapse point. Seven of nine measurable companies scored 0, meaning AI systems have no way to confirm whether the content they retrieve reflects 2026 operations or something years out of date.
• DuPont’s page carries 10 H1 tags. Formosa Plastics Group’s carries 2. Both trigger fragmented intent, the point where AI parsers can no longer confidently identify a page’s primary topic.
• Structure and Depth are consistently strong (sector averages of 91.7 and 78.3), confirming a pattern this series has now documented across nine industries: chemical majors write for engineers and procurement officers well, and for machines poorly.

What “AI Visibility” Means for a Chemical Manufacturer

AI visibility is the measurable degree to which an AI system, an LLM-based search engine, a chatbot, or an autonomous procurement agent, can correctly parse, verify, and cite a company’s website when answering a query about that company. It has nothing to do with domain authority or backlink count in the traditional SEO sense. It’s a structural property: does the page have one clear topical anchor, does it carry the schema markup that lets a machine assert facts with confidence, and can the system verify when that content was last true.

For a chemicals company, that translates into a specific commercial risk. When a procurement engineer asks an AI copilot to compare polycarbonate suppliers, or a sustainability officer asks which petrochemical majors have published verified emissions reduction targets, the AI system answers from whatever it can retrieve with structural confidence. A company with strong content but weak structural signals doesn’t get excluded gently. It gets replaced in the answer by a competitor, or by a third-party distributor site, that the machine trusts more.

Methodology

Each company’s primary global corporate domain was evaluated with the AI Visibility Inspector across four dimensions defined by the Ivica Srncevic Frameworks:

• Structure – whether the page has a clear, singular topical anchor (H1 clarity) and coherent navigational architecture
• Depth – the substantive quality and machine-extractability of the on-page content
• Schema – the presence of structured data (JSON-LD) that lets AI systems assert facts about the company with confidence rather than inference
• Freshness – whether dateModified JSON-LD or equivalent date signals exist, letting AI systems verify content currency

The composite AI Retrieval Index runs 0 to 100. Below 50 is Grade D, structurally invisible. 50 to 74 is Grade C, fair but materially gapped. 75 and above is Grade B or A, genuine AI readiness. A Structural Decay warning fires when a page is missing an H1 entirely, carries multiple competing H1 tags, or has no verifiable date signal.

The Scores

Sector average: 57.6 – Grade C, AI Retrieval Index (calculated across the nine companies the Inspector could fully evaluate)

Seven companies landed in Grade C. Two landed in Grade D. Zero reached Grade B or A. And one, SABIC, didn’t produce a score at all. This is the chemicals and petrochemicals industry’s AI visibility profile in July 2026.

Five Findings the Sector Cannot Ignore

Finding 1: A Sector Ceiling, Not a Sector Range

No company in this dataset broke 70. LG Chem’s 69 is the closest anything came to Grade B, and it got there almost entirely on the strength of a Schema score of 70, more than double the sector’s average. Compare that to the light vehicle manufacturing sector this series covered previously, where General Motors reached 75 on a near-identical structural profile. In chemicals, there is no equivalent leader. The best performer in this dataset would rank in the middle of the pack in several other industries this series has analyzed.

Finding 2: SABIC’s Failure Is a Different Category of Problem

Every other company in this dataset has a visibility problem. SABIC has a rendering problem, and it’s worth separating the two clearly. The AI Visibility Inspector recorded a 17-second load time on SABIC’s global site and was unable to extract structural data at all. A slow, unscored page isn’t a Grade D result. It’s a page that AI crawlers and retrieval agents may simply time out on and skip. A company that cannot be measured cannot be cited, and a page an AI system abandons before it finishes loading might as well not exist to that system at all. That is a materially worse position than any numeric score in this dataset, and it’s the single most urgent structural finding of this research.

Finding 3: Freshness Is Where the Sector Collapses

Sector-wide Freshness average: 11.6. Five of nine measurable companies, Dow, Ineos, LyondellBasell, DuPont, and effectively Formosa Plastics Group, scored 0. Only ExxonMobil (66) and Covestro (22) show meaningful date-verification signal. In an industry where regulatory compliance status, product formulations, and sustainability commitments change on a near-quarterly basis, an AI system that cannot verify when a claim was last true has to treat every claim on the page as potentially stale. That’s a specific and avoidable liability for a sector under constant environmental and regulatory scrutiny.

Finding 4: Fragmented Intent Shows Up at Both Extremes

DuPont’s corporate page carries 10 H1 tags. Formosa Plastics Group’s carries 2. Both trigger a Structural Decay warning for the same underlying reason: the AI parser cannot resolve which topic on the page is primary. Ten competing H1s scatter the machine’s attention across ten candidate topics with no clear winner. Two competing H1s do the same thing with less noise but the identical structural consequence, an entity representation the AI system can only build probabilistically, not declaratively.

Finding 5: Structure and Depth Are Strong. Schema and Freshness Are Not.

Sector-wide, Structure averages 91.7 and Depth averages 78.3, both genuinely strong numbers. Schema averages 38.3. Freshness averages 11.6. This is the same pattern this research series has now documented across nine prior industries, and the chemicals sector confirms it with unusual clarity: these companies write technically rich, well-organized content for human engineers and procurement teams. They have not extended that same investment to the structured data and date-verification layer that AI systems actually depend on to trust and cite that content.

This isn’t a judgment of product quality, R&D capability, sustainability performance, or commercial reputation. BASF’s chemistry, ExxonMobil’s refining capacity, and LG Chem’s battery materials business are not in question here, and nothing in this dataset should be read as a comment on them. This is strictly a structural, machine-readability assessment of one primary corporate domain per company, evaluated at a single point in time. A company can be an outstanding chemical manufacturer and still be poorly represented to the AI systems now mediating a growing share of B2B research and procurement discovery. Those are two separate facts, and conflating them is exactly the kind of imprecision this research is built to avoid.

What AI Actually Sees

The Inspector’s entity-interpretation output for ExxonMobil is the clearest illustration in this dataset of what “fair but gapped” looks like in practice. The system correctly parses the page as being about Exxon Mobil Corporation, a positive baseline signal most of this dataset’s D-grade companies don’t achieve. But its own assessment flags that key structural or schema signals are missing, which limits how confidently that page can be cited in an AI-generated answer. That’s a company doing the basics right and still losing citation eligibility on the details.

DuPont and Formosa Plastics Group present the opposite pattern. Their fragmented H1 structures mean an AI parser encounters multiple competing candidate topics on the same page and has no reliable way to choose between them. And SABIC presents no pattern at all, because there was nothing to parse within the tool’s processing window.

Cost of Inaction

Every quarter these gaps go unaddressed, more procurement research, competitive benchmarking, and supplier discovery shifts from search engines into AI copilots that answer directly rather than linking out. A chemicals major with zero Freshness signal and Schema in the 25 to 35 range isn’t losing visibility gradually. It’s already being routinely passed over in AI-synthesized supplier comparisons in favor of competitors, distributors, or trade publications whose structured data the AI system trusts more. For a company the size of the ones in this dataset, that’s not a marketing problem. It’s a pipeline and RFP-inclusion problem, and it compounds every month it’s left unaddressed.

An Uncomfortable Truth

The chemicals industry has spent decades building technically excellent, compliance-dense websites for a human reader who reads carefully and slowly. AI systems don’t read that way. They parse structurally, in milliseconds, and they discard what they can’t confidently anchor. A page can be factually complete and still be functionally invisible to the exact systems now doing a growing share of first-pass supplier research. Content quality was the old differentiator. Structural signal architecture is the new one, and right now, in this sector, almost nobody has it.

If your team is already thinking about what this means for your own domain, I work with enterprise organizations on exactly this gap through Enterprise Search Advisory, diagnosing where AI systems lose confidence in your content and fixing it at the structural level before it becomes a competitive disadvantage.

Frequently Asked Questions

Key Takeaways (Recap)

• Sector average of 57.6 places chemicals and petrochemicals in the lower-middle tier of every industry this series has analyzed to date.
• LG Chem’s 69 is the ceiling, not a floor others are clustering below, no company reached Grade B.
• SABIC’s non-render is the standout finding of this analysis and the clearest illustration yet in this series of what total AI invisibility looks like in practice.
• Freshness, at a sector average of 11.6, is the single most urgent and least expensive gap to close across this dataset.

Research Date: July 2026 | Methodology: Ivica Srncevic Framework + AI Visibility Inspector. This research is independent, not sponsored by any organization or legal entity. All company names and logos are used for identification and analysis purposes only.

Where This Goes From Here

If you’re evaluating your own organization against these findings, there are two ways I can help. For a hands-on structural diagnosis and remediation roadmap specific to your domain, my Enterprise Search Advisory engagement walks through the same framework applied here, at the level of your actual site architecture. For organizations that need ongoing, always-on monitoring of how AI systems are representing their brand, NovaX AI Visibility Intelligence tracks these exact signals continuously rather than as a point-in-time snapshot. Both start with the same question this article asked about SABIC, BASF, and the rest of this dataset: what does AI actually see when it looks at you.

This research is part of an ongoing independent series analyzing AI visibility across global industries. Previous installments cover the legal industry, global pharmaceutical, SaaS CRM, global banking, industrial tools manufacturing, life and health insurance, automobile industry, commercial vehicle sector, hospitality and tourism and global light vehicle manufacturers. All assessments use the AI Visibility Inspector and the Ivica Srncevic Framework.

Further discussion available in r/RetrievalOptimization.