Layer 1 — Access (do this first)
- robots.txt allows the search crawlers that current-answer and Search eligibility depend on: OAI-SearchBot, Claude-SearchBot and PerplexityBot (the assistants' own search crawlers), plus Googlebot — the ordinary Google Search crawler whose eligibility AI Overviews and AI Mode run on.
- Training and AI-use controls (GPTBot, ClaudeBot, CCBot, Google-Extended) are a separate decision — whatever you choose there does not affect the row above.
- CDN bot protection reviewed: Cloudflare and similar services can block AI crawlers by default, invisibly to you — check the AI-bots setting explicitly.
- AI crawler hits actually appear in your server logs (the only proof access works end to end).
- Key pages render their content server-side — text that only exists after JavaScript runs is invisible to most crawlers.
- sitemap.xml exists, is referenced from robots.txt, and lists your real pages.
Everything below is wasted effort while this layer is red. It's also the fastest layer to fix — usually one config change.
Layer 2 — Content (the quotability audit)
- Every page targeting a question opens with a direct 2–3 sentence answer under a matching heading — not a warm-up paragraph.
- Claims carry numbers, named sources, and dates; "customers love us" is not liftable, "4.8/5 across 214 reviews (G2, 2026)" is.
- Pages show a visible published/updated date, and the date is honest.
- One page per question cluster — engines pick the best passage, and five thin pages competing for one question all lose.
- Your pricing, category, and key differentiators are stated in plain text somewhere crawlable (not only in an image or a JS widget).
Layer 3 — Identity (machine-readable trust)
- Organization JSON-LD on the homepage: name, URL, sameAs links to your real profiles.
- FAQPage JSON-LD wherever you genuinely answer questions.
- Brand facts consistent everywhere — same name, same category description, same claims across your site, LinkedIn, directories (engines cross-check).
- llms.txt if you want the cheap option value — honestly weighted, it's a minor signal (see our llms.txt guide).
Layer 4 — Presence and proof
- You know which domains engines cite for your category's questions (ask the engines, note the citations).
- You're honestly listed in the recurring ones: directories, review sites, comparison posts.
- Someone participates genuinely where your category is discussed (Reddit, niche forums, communities).
- You track a fixed question set across engines on a schedule, and record mentions, citations, and who won instead.
- You measure AI-driven visits and revenue — including the share hiding in "Direct" — so the work connects to money.
- After every fix batch: re-check the same questions and compare. No before/after, no proof.
Promvia automates layers 3's audit and all of layer 4 (multi-engine tracking, source lists, revenue attribution, re-check verification). Layers 1–2 are yours either way — no tool writes your answers for you honestly.
Frequently asked questions
How often should I re-run this checklist?
Layer 1 quarterly or after any infrastructure change (CDN swaps silently break access). Layer 2 whenever you publish. Layer 4 is continuous by design.
What's the single highest-leverage item?
For most sites: checking CDN/robots access, because failure there zeroes everything else and takes minutes to fix. After that, answer-first rewrites of your top five pages.
Can I skip straight to off-site work?
If your access and content layers are broken, mentions elsewhere help competitors' comparisons more than you — engines will read ABOUT you from third parties but can't quote YOU. Fix the base first; it's days, not months.