---
name: seo-ready-website
description: Build or audit websites against the 80-point SEO-Ready Website Checklist (Google rankings + AI-search citations). Use when building any new page or site, auditing an existing site's SEO, or when the user asks "is this site SEO-ready", "why isn't my site ranking", or "make this page rank / get cited by ChatGPT".
---

# SEO-Ready Website (80-Point Standard)

You hold the site you are building or auditing to the 80 pass/fail checks in `seo-checklist.md` (keep both files together). The checklist covers classic Google SEO **and** AI-search readiness (AEO/GEO), so the site ranks in blue links and gets cited by ChatGPT, Perplexity, and Google AI Overviews.

By Calibre Consulting · calibreconsulting.ca

## Two modes

**Build mode** — the user is creating or editing pages. Apply the checklist as you write code, don't bolt it on after:

1. Before writing a page, read sections C (on-page), F (schema), and G (AI readiness). Every page you emit ships with: one H1, unique 50–60-char title, unique 150–160-char meta description, logical heading nesting, descriptive alt text, OG + Twitter Card tags, and a connected `@graph` of JSON-LD (Organization/LocalBusiness + WebSite + BreadcrumbList + the page-type node).
2. Open key pages with a self-contained 40–60 word answer to the question the page targets (check 62). Phrase at least some H2s as real questions (check 63). State key facts as quotable standalone sentences with a number or date (check 64).
3. Site-level work on every project: robots.txt that allows GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot (check 7); XML sitemap of canonical URLs only (check 2); `llms.txt` answer map at the root (check 65); one canonical host with https enforced (check 12).
4. Performance is a build-time decision (section E): compress images to WebP/AVIF with width/height set, lazy-load below the fold, prioritize the hero, `font-display: swap`, minify. Do not ship and "optimize later".

**Audit mode** — the user points you at an existing site. Produce a scored report:

1. Fetch and inspect what you can verify directly: `/robots.txt`, `/sitemap.xml`, `/llms.txt`, page source of the key pages (titles, metas, H1s, canonicals, noindex, OG tags, JSON-LD, alt text), redirect behavior (http→https, www/non-www, trailing slash), 404 status codes.
2. Mark each of the 80 checks **PASS / FAIL / NEEDS-HUMAN**. Use NEEDS-HUMAN only for items you genuinely cannot verify from outside: Search Console data (9, 36, 78), CrUX field data (52), backlink profile (75–76), Google Business Profile state (72), analytics internals (79).
3. Output, in this order: score as **N/80** (count NEEDS-HUMAN as not-passed and say so), a one-paragraph verdict in plain language, a per-section table (section, passed/total, the failing check numbers), then the **top 10 fixes ranked by impact ÷ effort** with the concrete change for each (file, tag, or config to edit, not advice-shaped advice).
4. Quick-win bias for ranking fixes: anything in sections A–B that blocks crawling or indexing outranks everything else; then E (Core Web Vitals) on the money pages; then F–G (schema + AI readiness), which most competitors still skip.

## Rules

- A check passes only as written. "Has a title tag" is not check 21; a *unique, 50–60-char* title with the keyword near the front is.
- Never fake a pass you didn't verify, and never emit schema for content that isn't visibly on the page (check 61 — penalty risk).
- Sections gate each other: don't polish schema (F) on pages that fail crawlability (A) or indexation (B). Fix in A → H order.
- When the user's stack limits a fix (e.g. shared hosting, no CDN), say what's achievable there instead of marking it hopeless.
- Re-audit quarterly (check 80). When asked to "make it rank", run audit mode first; never start optimizing unmeasured.