SEO Audit & Action Plan Generator

What it does

It audits a website along SEO's dependency chain: first crawlability & indexation (can Google find and index it?), then technical foundations (speed and function), on-page optimization, content quality, and finally authority/backlinks. The order matters — fixing a title tag on a page that cannot be indexed has zero value; the gate must open first.

The audit runs on concrete thresholds. Core Web Vitals: LCP < 2.5s, INP < 200ms, CLS < 0.1 (Google's 75th-percentile measurement). International SEO: hreflang self-reference + reciprocal-link rule, valid ISO codes (en-GB, never en-UK), x-default, canonical–hreflang conflicts. On-page: single H1, 50-60 char titles, keyword cannibalization, internal-link/orphan detection. Content: E-E-A-T signals (Experience, Expertise, Authoritativeness, Trustworthiness) and thin content. Every finding is reported as Issue → Impact (High/Medium/Low) → Evidence → Fix → Priority (1-5).

The output is an audit document: an executive summary, technical/on-page/content findings, and a 4-tier prioritized action plan (critical blockers → high impact → quick wins → long-term).

When to use it

For "why am I not ranking," "my traffic dropped," "a Google update hit me," "I'm not showing up in Google," "indexing issues," "page speed," "Core Web Vitals" — even something as vague as "my SEO is bad" starts here. Ideal for post-migration traffic drops, pre-launch health checks, or untangling a multilingual site.

Method / frameworks

• Priority-order model — Crawlability & Indexation → Technical → On-Page → Content → Authority (downstream principle).
• Core Web Vitals (Google) — LCP / INP (replaced FID) / CLS, 75th-percentile good/needs-improvement/poor bands.
• E-E-A-T — Google Quality Rater framework for content-quality signals.
• hreflang & canonical doctrine — self-reference, reciprocity, x-default, no cross-locale canonical.
• Issue → Impact → Evidence → Fix → Priority — standard action card per finding.
• Site-type issue sets — tailored checklists for SaaS / e-commerce / blog / local / multilingual.

Schema detection via web_fetch is unreliable (JS-injected JSON-LD is invisible); validation uses Rich Results Test / Screaming Frog.