Myth: llms.txt improves Google AI Overviews
Google stance: no stated ranking or eligibility impact.
Do instead: keep Googlebot crawl/index strong; consider llms.txt only for Lens B.
SEO & GEO Audit
Polymorph — audit deliverable for Google AI, ChatGPT, Perplexity, Claude, and WordPress / Yoast implementation
This audit covers 100 URLs from Polymorph’s Yoast sitemap (pages, case studies, and recent resources) on polymorph.co.za (canonical apex; www correctly 301s). Technical foundations are strong: Organisation + WebSite JSON-LD, sitemap, viewport, and all nine checked crawlers (Googlebot plus GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, Bingbot, and peers) are allowed. For Google AI Overviews / AI Mode, the highest-impact gaps are missing H1s on homepage / about / careers / contact, missing meta descriptions on about / careers / request-consultation, and ensuring commercial pages stay people-first with measurable outcomes — not commodity “custom software” copy. For ChatGPT, Perplexity, and Claude, answer-first intros are effectively absent site-wide (0% of audited pages), concrete statistics appear on only ~20% of pages, FAQPage schema is not paired with visible FAQ blocks on key commercial URLs, and optional llms.txt is missing (Lens B only). Priorities: fix H1 + meta on money pages, lead services and case studies with extractable answers and numbers, deepen the 7 Fundamentals / franchise content as citation assets, then add FAQ content + schema and optional llms.txt.
Scope: 100 URLs (cap) from sitemap_index.xml → page, case-studies, and post sitemaps (July 2026).
Site-level: robots.txt allow-all with Sitemap directive; Yoast schema graph present (Organisation, WebSite, WebPage); no llms.txt; FAQPage recommended where visible FAQs exist.
Critical: missing H1s on core pages; missing meta on about / careers / consultation; zero answer-first intros; thin statistics coverage outside hero case studies.
Different AI surfaces discover and cite content in different ways. Recommendations are tagged where the best action differs by platform.
Homepage and site-wide checks (July 2026).
Full-site extractability: answer-first intros 0%; statistics ~20%; citations strong; freshness signals present via Yoast dateModified on audited URLs.
WordPress / Yoast — fastest impact for search and AI citation.
Includes concrete “show up more” plays for Google AI and multi-platform LLM citation.
H1s on money pages — Homepage, About, Careers, Contact, and Request consultation currently lack a proper H1. One descriptive H1 improves snippets, accessibility, and entity clarity.
Meta descriptions on About / Careers / Consultation — Fill Yoast with the task-list copy so Search and Google AI have a useful snippet, not an auto-excerpt.
People-first service pages — Keep Concept Design / Advisory / Mobile focused on buyer problems and outcomes. Avoid generic “custom software agency” listicles that look commodity across AI answers.
Internal links toward Contact / Concept Design — From Resources and case studies, link to the diagnostic pathway so organic and AI-referred traffic can convert.
Answer-first intros everywhere that matters — 0% of audited pages open with a direct answer. Add one-sentence capsules on services, case studies, and cornerstone resources.
Statistics with context — Only ~20% of pages carry concrete numbers. Push measurable outcomes (installs, signups, ROI, timeline) into case-study and service intros — with surrounding explanation.
FAQ content + FAQPage schema — Visible FAQs on Concept Design / Contact; schema matched to on-page Q&A for rich results + extractability (not a Google AI requirement).
Double down on citation-worthy franchises — 7 Fundamentals hub, Jargon Decoder, Bullshit Translator, and 5 Red Flags are original POV assets. Keep them fresh with visible last-updated dates and internal links to services.
Optional llms.txt — Brief map of key pages and brand entity. Neutral for Google AI Overviews; useful for Perplexity and emerging LLM tools.
Do not block AI crawlers — Current allow-all robots.txt is correct for ChatGPT browse, Perplexity, Claude, and Bing/Copilot paths. Re-check after any security plugin changes.
Enrich Organisation sameAs — Ensure LinkedIn, Business Day features, and authentic profiles appear in schema and footer consistently.
Freshness on cornerstone posts — Yoast dateModified exists; add a visible “Last updated” on 7 Fundamentals and other hub articles when you revise them.
Skip Google GEO myths — Do not chase llms.txt, AI-tone rewrites, or fan-out spam for AI Overviews. Invest in expertise, proof, and crawl/index hygiene instead.
Google-scoped myths. Some tactics remain useful for broader AI citation when framed correctly.
Google stance: no stated ranking or eligibility impact.
Do instead: keep Googlebot crawl/index strong; consider llms.txt only for Lens B.
Google stance: no special chunking requirement.
Do instead: structure for readers with clear headings, summaries, and direct answers.
Google stance: helpful, reliable, people-first content wins.
Do instead: keep Polymorph’s diagnostic-first voice, case proof, and original franchises.
Google stance: scaled thin content is risky.
Do instead: deepen Concept Design, Advisory, and case-study pages you already have.
Google stance: inauthentic reputation signals are risky.
Do instead: authentic Business Day / client corroboration and consistent entity profiles.
Traditional SEO and Google generative search (AI Overviews, AI Mode) share the same foundation: indexed, crawlable pages and helpful, reliable, people-first content. Google retrieves supporting links from its Search index using ranking systems — for Google, strong SEO is generative search optimisation. ChatGPT, Perplexity, Claude, and Copilot use different retrieval paths (training data, live browse, proprietary indexes), so this audit also covers multi-platform citation: AI crawler access, extractable content (citations, statistics, expert quotes, clear answer blocks), schema for entity clarity, freshness, and authentic authority signals. Where Google and other platforms disagree — for example llms.txt — recommendations are tagged so you know what helps where.
Yoast already outputs Organisation / WebSite / WebPage. Add FAQPage only where visible FAQ content exists — for rich results and multi-platform entity resolution, not as a Google AI Overviews requirement.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What does Polymorph do?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Polymorph builds people-centred custom software: advisory, concept design, prototyping, and software development for businesses that want smart software aligned to their vision and users."
}
},
{
"@type": "Question",
"name": "What is concept design at Polymorph?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Concept design is Polymorph’s diagnostic-first phase that decreases software risk and maximises ROI by aligning customer needs, UX, technical feasibility, and cost before full-scale build."
}
},
{
"@type": "Question",
"name": "Where is Polymorph based?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Polymorph Systems (Pty) Ltd is based in Techno Park, Stellenbosch, South Africa, and partners with clients across South Africa and beyond."
}
}
]
}
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Polymorph",
"legalName": "Polymorph Systems (Pty) Ltd",
"url": "https://polymorph.co.za/",
"logo": "https://polymorph.co.za/wp-content/uploads/polymorph-logo.png",
"description": "People-centred custom software development, advisory, and concept design since 2008.",
"email": "[email protected]",
"telephone": "+27-21-880-1034",
"address": {
"@type": "PostalAddress",
"streetAddress": "Unit 204, Workshack, 16 Elektron Avenue, Techno Park",
"addressLocality": "Stellenbosch",
"postalCode": "7600",
"addressCountry": "ZA"
},
"sameAs": [
"https://www.linkedin.com/company/polymorph/"
]
}
After adding FAQ or enriching Organisation schema, validate markup and confirm Google can see key URLs.