[THE 2026 LAYER · 06 CHECKS]
Each layer is independently auditable, and each one fixes a different way AI search will quietly route around your site.
LAYER-01 / 0601 / 06
If your content only exists after JavaScript runs, AI crawlers — most of which don’t execute JS reliably — see nothing. SSR or SSG is now table stakes.
- HTML fully rendered server-side or statically generated
- Content exists in raw HTML without JS execution
- Spot-check priority templates via ‘view source’, not rendered DOM
- If using React/Vue: SSR (Next.js, Nuxt), SSG, or selective hydration
- GSC URL Inspection ‘rendered HTML’ matches the served HTML for content
LAYER-02 / 0602 / 06
Plain-text index of high-value URLs for LLM agents. A modern analogue to sitemap.xml — emerging convention, low cost, real upside.
- llms.txt published at /llms.txt
- Includes high-value URLs (services, pillars, about, contact)
- Hierarchical structure with headings if the site is large
- Linked from robots.txt where appropriate
- Kept in sync with the canonical sitemap
LAYER-03 / 0603 / 06
If your WAF or robots.txt silently blocks AI bots, your brand will never surface in answer engines — and you won’t see the failure in any tool.
- robots.txt explicitly allows GPTBot
- robots.txt explicitly allows ClaudeBot (Anthropic)
- robots.txt explicitly allows PerplexityBot
- robots.txt explicitly allows Google-Extended (Gemini training)
- robots.txt explicitly allows CCBot (Common Crawl)
- WAF / bot-management rules don’t shadowban the above
- Strategic block? Only if you have a documented reason — and you understand the discoverability cost
LAYER-04 / 0604 / 06
Schema isn’t a plugin output — it’s how AI systems parse you into facts they can cite. Validate it, and make sure the fields agree with what’s on the page.
- Structured data validates cleanly in Google Rich Results Test
- Structured data validates cleanly in Schema.org validator
- Entity consistency: org name, founding date, founders, product names appear identically across schema and visible copy
- Schema generated from the same source of truth as the rendered content
- Validation step wired into CI so regressions don’t ship
LAYER-05 / 0605 / 06
LLMs favor short, definitional paragraphs near the top of the page and H2s phrased as questions. Long preamble buries you in the citation race.
- Definitions appear near the top of the page, not buried under context
- Short, declarative paragraphs (2–4 sentences)
- Clear H2 questions for the queries you want to be cited on
- Tables and lists for comparison-style queries
- Author bylines + credentials for E-E-A-T
LAYER-06 / 0606 / 06
The only test that matters: are you cited when buyers ask the questions your service answers? Run it manually, monthly, across the four major answer engines.
- 5–10 brand-relevant queries spot-checked in Perplexity
- Same queries spot-checked in ChatGPT (with browsing)
- Same queries spot-checked in Claude (with web tool)
- Same queries spot-checked in Gemini
- Citations tracked over time — are you gaining or losing surface?
- Where competitors are cited and you aren’t, the gap becomes a content target
