Running nine separate audits on one page is a reasonable way to audit a page. It is also a terrible way to spend a Tuesday morning. You open the Site Analyzer, then the E-E-A-T Audit, then Voice & Tone, then Mobile Parity. You paste the same URL four times. You collect four prompts. You feed them to Claude one at a time. By the end you have a pile of partially overlapping fix suggestions and no clear priority.
The Mega Analyzer is what I built when I got tired of doing that. One URL, one run, one report, one prompt.
What It Measures
Six buckets, scored out of 100, rolled into one overall grade. Each bucket maps to checks the other tools already run — I just wired them into a single pass so you stop paying the context-switching tax.
SEO covers the fundamentals: title length, meta description, H1/H2 structure, canonical, lang attribute, viewport (including WCAG 1.4.4 zoom allowed), HTTPS, charset, robots meta, and whether robots.txt and sitemap.xml exist.
Schema parses every JSON-LD block, counts types, flags parse errors, checks for Organization, Person, BreadcrumbList, FAQPage, and Speakable, and runs the Article sub-validation — text or image or video required, datePublished with a timezone, author.url present.
E-E-A-T looks at Person and Organization schema, sameAs depth, Wikidata reference, ORCID reference, rel="me" links, content depth, and whether an llms.txt exists at the site root.
Voice runs the Flesch Reading Ease, Flesch-Kincaid grade level, passive-voice rate, hedge density, sentence-length variance, and CTA density against the page body. The targets are published rubrics: Flesch ≥ 60, grade 8 to 12, passive under 10%, sentence-length standard deviation between 6 and 12.
Mobile fetches the page a second time with an Android Chrome User-Agent and diffs everything that matters against the desktop fetch — word count, H1, H2 set, schema types, HTML size. This is the one the other tools can't do in a single pass because it requires the dual fetch; the Mega Analyzer handles it inline.
Perf + AI covers HTML weight, DOM node count, image dimensions and modern formats, duplicate IDs, accessibility landmarks, AI crawler rules in robots.txt for GPTBot / ClaudeBot / PerplexityBot / Google-Extended / CCBot, and the Open Graph + Twitter card completeness that decides whether social previews render cleanly.
One Grade, Six Bars
The Overview tab shows an overall letter grade — A through F — and a per-bucket bar chart. This is the part I wish had existed three years ago. If your Schema bar is at 55 but your SEO bar is at 90, you know exactly where to spend your time. Not "improve SEO." Specifically: add the three schema types that are missing.
Each bucket tab drills into individual checks with pass / fail icons and explanations. Clicking through takes a minute and tells you exactly which lines in your HTML are producing the drag.
The Mega AI Fix Prompt
The last tab produces a single long prompt, ready to paste into Claude or ChatGPT. It includes:
- Every score and the overall grade
- The full current state of the page — title, meta, H1, H2 list, canonical, lang, viewport, word count, schema types detected, JSON-LD parse errors, sameAs URLs, Wikidata and ORCID flags, image stats, HTML weight, DOM nodes
- The voice metrics with targets
- The mobile parity diff
- AI crawler allow/disallow status
- Instructions to produce a prioritized fix list with exact code, file locations, which bucket each fix moves, and verification steps
It also carries the deployment-layer caveat: security headers, 301 redirects, and schema injected at build live in a CDN / SSG layer the audit can't see, so before writing warning-level fixes the LLM should confirm which deployment stack is in use.
If you have the skill level set to Advanced anywhere in the jwatte.com tool suite, the prompt skips the explanations and asks for deployable code. Beginner adds reasoning. The level is shared across all tools, so set it once.
Velocity Snapshot Baked In
Every Mega Analyzer run also writes a snapshot to localStorage under the same key the Content Velocity Monitor uses. Run the tool today, come back in a month, open Content Velocity, and you can diff exactly what changed — word count, H2s, schema, title, meta — without any extra step. It's labeled as auto-saved so it won't clutter your named baselines.
When to Run It
First audit on a new site. The Mega Analyzer is now the fastest way to get a complete picture before you start making changes.
Before and after a significant rewrite. Run once on the old version, once on the new. The bar chart tells you which bucket moved and by how much. Mobile parity is the bucket that most often regresses silently during rewrites — catch it before anyone notices.
Quarterly on the pages that matter most. Your top traffic-driving pages, the ones whose rankings affect revenue. Fifteen seconds per page. Export the JSON via the Content Velocity tool if you want a quarterly record.
When a client asks "where do we stand." One grade, six bars, one mega prompt. That's a meeting deliverable in 30 seconds.
How This Fits the Methodology
The Mega Analyzer is the automated version of the comprehensive per-page audit the $97 Launch Chapter 35 (Competitive Site Auditing) walks through manually. The per-bucket scoring mirrors the chapter structure of $20 Agency — chapters 5 through 11 each correspond to one of the buckets the tool scores. And the AI-crawler check, llms.txt detection, and velocity snapshot form the network-scale monitoring stack described in the $100 Network chapters 16, 17, and 26. Pick the book that matches where you are and the Mega Analyzer is the tool you pair with it.
Honest Limits
It reads static HTML. Pages whose content is injected after first paint will look thin. This is consistent with what Google and AI crawlers see, so it's useful information either way.
It scores. It does not fix. The AI prompt is the bridge between "here is the gap" and "here is the code" — and the LLM's output still needs a human pass.
It is a single-URL tool. For multi-site comparison, use the new Mega Batch — same six buckets, up to 10 URLs at once, one combined prompt.