AI-powered evaluation using the Model Context Optimization BS Detection Framework, based solely on publicly available website content.
Based on 587 businesses audited.
Medical Devices, Pharma & Biotech BS: Genentech (PERJETA) (perjeta.com)
This is a high-substance, low-fluff pharmaceutical site that prioritizes regulatory compliance and technical accuracy over marketing hyperbole. The BS score is driven almost entirely by a single emotional H1 and a technical schema flag regarding review counts. It represents the gold standard for substantive product-led medical communication.
Change the homepage H1 from ‘I’m all in’ to a descriptive technical heading such as ‘PERJETA Treatment for HER2+ Early and Metastatic Breast Cancer.’ Ensure the review_count in the schema_json is removed if it does not link to a verifiable third-party review platform. Include direct links to peer-reviewed ClinicalTrials.gov study results in the ‘Clinical Data’ or footer sections to maximize proof paths. Replace generic female patient icons with specific data visualizations of the PERJETA mechanism of action.
The information density is extremely high for the body text, though the primary H1 ‘I’m all in’ is a pure emotional power-word construct without a noun. The body text contains precise technical protocols, such as ‘initial dose of 840 mg infused over 60 minutes’ and specific household income thresholds for financial aid ($150,000). Specificity is maintained through technical nomenclature like ‘HER2-positive early breast cancer’ and ‘neoadjuvant treatment’ across all pages. The ratio of fluff to substance is negligible once past the hero section.
When edges drift or clusters collapse, your content becomes a set of disconnected islands. Inspect your internal link topology to identify where authority flow breaks or never forms.
Semantic drift is nearly non-existent; the homepage hero section promises a treatment path for HER2+ breast cancer which is then rigorously defined on the sub-pages. The sub-page for early breast cancer delivers exactly what the homepage navigation suggests, including a clear distinction between neoadjuvant and adjuvant treatment. There are no contradictions between the ‘Financial resources’ claim on the homepage and the detailed co-pay/assistance eligibility tables on the financial sub-page. The site transitions seamlessly from patient-centric emotional hooks to clinical data for HCPs.
Transition from a collection of strings to a machine verifiable identity. Generate your Clinical SEO Strategy to establish a robust Knowledge Graph Topology and eliminate semantic black holes.
The trust signals are mostly substantiated by regulatory requirements rather than marketing gimmicks. While the data shows a review_count of 1 with 0 external proof links, this is typical for pharma sites where ‘reviews’ are not allowed and ‘proof’ is provided by a link to the Full Prescribing Information. The presence of the FDA MedWatch reporting number (1-800-FDA-1088) acts as a high-authority proof path. The trust_theatre_flag is triggered by the structured data review count, which appears to be a technical placeholder rather than a substantive testimonial claim.
Proof density is high, with a ratio of approximately one clinical or logistical fact for every three sentences of marketing copy. Verifiable evidence includes specific clinical trial registration references implied by the Prescribing Information and the Genentech Patient Foundation income brackets. Vague assertions like ‘I’m all in’ are confined to the decorative H1s, while the actionable text remains data-driven. The site provides specific timeframes for enrollment (five business days) and claim submission (365 days) for financial aid.
For a high volume editorial domain example, open the Search Engine Journal Semantic HTML audit. View the SEJ Semantic HTML Audit to see how template drift and structural noise impact AI chunking.
The value proposition is product-specific rather than industry-generic because it centers on a specific FDA-approved molecule (pertuzumab). While template elements like ‘Starting the conversation’ are standard for patient-facing pharma sites, the content within these blocks is highly specific to HER2+ therapy. The financial assistance section is uniquely detailed with specific program terms that could not be easily copy-pasted by a generic competitor. Clichés are present (e.g., ‘Discover another treatment’) but are subordinate to the technical mechanism of action.
There are no authority gaps as the site is backed by Genentech, a verifiable biotechnology leader with a clear digital footprint in the schema_json. The organization schema includes a physical address in South San Francisco, a customer service telephone number, and official logo links. There are no expert claims from unidentifiable ‘gurus’; instead, the site deferentially points to ‘your doctor’ or ‘health care professional’ as the authority. The technical implementation of the heading hierarchy and metadata is clean and professional.
The site avoids bold marketing performance claims like ‘unrivaled success’ and instead uses clinically approved language such as ‘designed to treat’ or ‘high likelihood of coming back.’ These assertions are anchored in the prescribing information and safety data rather than vague performance promises. The site clearly lists ‘serious side effects’ and ‘boxed warnings’ alongside any mention of efficacy, which is the antithesis of marketing bullshit. The disconnect between what is marketed and what is proven is zero due to strict regulatory oversight.
Medical Devices, Pharma & Biotech BS: Genentech (PERJETA) (perjeta.com)
The site perfectly matches the Medical Pharma & Biotech category, focusing on a specific prescription medicine, pertuzumab. The content is heavily regulated, featuring mandatory safety information, indications for use, and clinical dosage protocols characteristic of a major pharmaceutical entity.
AI does not interpret your layout visually — it interprets your structure mathematically. Explore the Semantic HTML Technical Framework to understand how heading logic, boundaries, and DOM depth determine what an LLM can retrieve.
“The score of 13 is exceptionally low. The primary drivers were minor template artifacts (a review count of 1 without proof links) and the high fluff-saturation of the homepage H1. The site achieved 0/15 in Authority Gaps and 1/20 in Semantic Coherence due to the high quality of Genentech's structured data and logical messaging flow.”