Workflow

1. Classify the page

• Page type, primary query, search intent
• Primary keyword + 3–8 variants
• Primary entity + supporting entities (map entity IDs/URIs where possible)
• Target platforms: Google Search (Web Guide / AI-overview), Google-Agent/AI Mode, ChatGPT/agent platforms, Gemini, Perplexity, vertical assistants (video, maps)
• Actionability: list the actionable flows an agent could take (e.g., booking, purchase, API call, contact)

2. Audit SEO coverage

• Keyword placement, title quality, meta description, clean slug
• Heading hierarchy, internal links, canonical URL
• Image alt text, video metadata, structured data presence and completeness
• Publication/provenance fields: datePublished, dateModified, author, publisher, licenses

3. Audit GEO / AI citability & agent-readiness

• Does the page answer the query directly and early? Provide a 1–3 sentence extractable summary at the top
• Are claims specific, sourced, and structured with anchors and date stamps?
• Is factual density sufficient for extraction without hallucination risk?
• Schema checks: entity types, sameAs links, identifiers, product/offer schema, localBusiness fields, videoObject fields, and citation/provenance fields
• Agentability checks: explicit CTAs, canonical endpoints or APIs, machine-readable pricing/inventory, GDPR/consent signals for commerce flows
• Local live signals: ensure Business Profile fields are up-to-date and feedable (offers, hours, availability)

4. Fix and implement

• Rewrite weak headings, thin intros, and missing entity coverage — add TL;DR and a clear answer early
• Add or fix JSON-LD schema; include provenance fields and sameAs links; ensure schema mirrors on-page entity mentions
• Add explicit citation anchors for claims (inline links + reference list) and mark update dates
• Add llms.txt and llms-full.txt at site root: include sitemap links, allowed/disallowed paths for AI crawlers, and preferred canonical
• For local: make GBP/Business Profile dynamic — update posts, inventory, service areas, and appointment links regularly
• For video: add robust videoObject schema, accurate descriptions, and review auto-generated summaries to protect brand voice

Entity gap analysis

• Compare your page's entities against the top 5 ranking pages. Add missing relevant entities — subtopics, tools, standards, brands, comparisons — without keyword stuffing.
• Map entities to identifiers (Wikidata, ISNI, product IDs) where available to improve grounding

Monitoring & rollout playbook

• Baseline organic and AI-driven visibility weekly; set alerts for big rank or traffic shifts during core or spam updates
• Track AI citation appearance (which platforms cite your content), branded extraction instances, and agent-driven conversions
• If volatility follows a platform change (core update, Web Guide, AI rollout), prioritize quality fixes: add provenance, richer entity coverage, and remove low-value duplicate pages

Notes

• Avoid claiming to "rank" in LLMs; prioritize being citable and actionable. LLM results are probabilistic and experiment frequently.
• Do not depend on a single distribution platform — optimize for classic SERP, AI citation, and agentic action flows in parallel.