Introducing Deep Audit: an AI-Powered, 9-Dimension Site Audit

For two years, SEO Score API has done one thing exceptionally well: tell you how healthy a URL is in seconds — 82 deterministic checks, JSON back before your coffee's poured. Great for a quick pulse. But a real audit — the kind an agency charges four figures for — needs judgment, not just rules. It needs to read the page the way Google and the new AI answer engines do.

That's Deep Audit: a thorough, AI-assisted site audit that scores any URL across nine dimensions, layering ~150 AI analysis passes on top of thousands of deterministic checks. It's live today in private beta for Pro and Ultra plans.

What it scores — the nine dimensions

Every Deep Audit returns a headline LAI score (0–5) and a grade, plus a score for each of nine sections:

1. Technical Foundation & Rendering — does the page render, server vs client, core markup health
2. Crawlability & Site Architecture — robots, sitemaps, internal linking, canonicalization
3. Content Quality & Entity Semantics — depth, E-E-A-T signals, entity coverage, answerability
4. Core Web Vitals & Performance — real CWV via PageSpeed Insights, speed, layout stability
5. Technical SEO & Index Control — meta directives, hreflang, structured indexing signals
6. Schema, Local & Trust — structured data coverage and correctness, trust signals
7. Security & Privacy — HTTPS, headers (CSP, HSTS, etc.), privacy posture
8. Multimodal Content & Media — images, alt coverage, video, media readiness
9. Off-Site Signals & Reputation — external footprint and reputation signals

You also get three sub-scores that tell different stories: AI-readability (how well an LLM can parse and answer from the page), classic SEO, and a composite beta score.

The full feature list

Scoring & analysis

• 9-dimension scoring with a 0–5 LAI score + letter grade, and per-section scores
• ~6,000 deterministic checks per audit + up to 150 AI (Claude) analysis passes
• Sub-scores for AI-readability, classic SEO, and overall
• Ranked findings with plain-English evidence, severity, and weight
• Coverage report — exactly how many checks ran, were scored, or were marked not-applicable, so you always know what was and wasn't evaluated

Accuracy & fairness

• Real rendering — every audit renders the page in headless Chromium, so it scores the DOM your visitors (and Googlebot) actually see, not just raw HTML
• Real Core Web Vitals via Google PageSpeed Insights
• Business-type applicability — pass business_type (saas, local_service, ecommerce, storefront, blog, publisher) and the engine only scores checks that apply. A SaaS homepage isn't docked for missing Event or Product schema.

Built for real workflows

• Asynchronous by design — submit a URL, get a job ID, poll for the result. While queued you get your queue position and an ETA; the SDKs can wait for you with a single call.
• Webhooks — get a POST when an audit finishes instead of polling
• CI/CD gate — a synchronous gate-scan endpoint returns pass/fail against a score threshold, so you can fail a build when SEO regresses
• SERP tools — live SERP results and head-to-head domain comparisons
• Brand radar — measure how AI answer engines describe your brand
• Public check catalog — browse every check the engine runs, no key required

SDKs everywhere

• Node and Python (deepAudit / deep_audit — one call, waits for the result)
• An n8n node (no-code "Deep Audit" operation)
• An MCP server — so Claude, Cursor, and other AI agents can run a Deep Audit as a tool

Why pay for this instead of Moz, Ahrefs, or Semrush?

Let's be straight: Moz, Ahrefs, and Semrush are great, and if you need a massive backlink index or a keyword-research database, keep them. Deep Audit isn't trying to replace those. It's the audit done right for how search actually works in 2026 — and it wins on five things that matter:

1. It scores you for the AI-search era, not just blue links. Every other tool grades you for classic Google rankings. Deep Audit gives you a dedicated AI-readability score — how well ChatGPT, Claude, Perplexity, and Google's AI Overviews can actually parse, understand, and cite your page. That's where discovery is moving, and it's the one thing the incumbents' rule-based crawlers don't measure.

2. It's an API, not a dashboard. The big suites make you log in and click around. Deep Audit runs where your team already works: fail a build in CI/CD when SEO regresses, trigger it from n8n, or let your AI agent run it as a tool (MCP). Auditing becomes automatic infrastructure instead of a monthly chore someone remembers to do.

3. It's a real analysis, not a checklist. Rule-based crawlers can only check rules. Deep Audit renders your page in a real browser and puts ~150 AI judgment calls on top — content depth, E-E-A-T, entity coverage, answerability — the subjective calls that actually separate a good page from a great one. Then it hands you findings in plain English, not a wall of red cells.

4. The score is honest — and you can trust it. It's applicability-aware (a SaaS homepage is never docked for missing Recipe or Event schema), it will never push you to fabricate reviews or ratings to game a number, and every audit ships a coverage report so you know exactly what was and wasn't evaluated. No dark-pattern "your score is 42, upgrade to see why."

5. It costs a fraction of the price. $39–$99/mo, versus the $100–$200+/mo tier the major SEO suites charge — and you get an automatable API and AI-agent integration, not one more dashboard seat.

How it works

1. You submit a URL (POST /site-audit) with your API key. Optionally tell it your business_type.
2. The engine renders and inspects the page — headless Chromium for the real DOM, PageSpeed Insights for Core Web Vitals, and thousands of deterministic checks across the nine sections.
3. AI does the judgment calls — the checks that need reading comprehension rather than a rule (content depth, E-E-A-T, entity coverage, answerability) are batched to Claude, running on AWS Bedrock, with the page's real structural facts supplied so it scores against ground truth.
4. Everything is scored and returned — the LAI score, all nine section scores, sub-scores, ranked findings with evidence, and a coverage report.

Because a full render + AI pass takes a minute or two, it runs as a background job: you poll (or use a webhook, or let the SDK wait). Under load, the queue absorbs bursts and hands you an honest ETA rather than falling over.

Plans

Deep Audit is a Pro / Ultra feature (it's the heavy, AI-backed analysis):

Both plans also include SERP queries — 250/month on Pro, 1,000/month on Ultra.

Already on Pro or Ultra? Your existing key already works — start now:

curl -X POST https://engine.seoscoreapi.com/site-audit \
  -H "X-API-Key: $YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"url":"https://yoursite.com","business_type":"saas"}'
# → {"job_id":"..."}   then poll GET /site-audit/{job_id}

Or one line with an SDK:

import seoscoreapi as seo
result = seo.deep_audit("https://yoursite.com", API_KEY, business_type="saas")
print(result["scores"]["lai_score"])

What "private beta" means

• It's live and stable, gated to Pro/Ultra while we scale.
• Scores are faithful to our methodology but we're still calibrating on real customer sites — if a score feels off, tell us and we'll tune it.
• Single-page/homepage today; multi-page crawls are coming.

We built Deep Audit because we wanted it ourselves — and the first thing we did was point it at our own site (and at Moz, Ahrefs, and Semrush). That story is in the next post. Run it on your worst page and tell us what you find.

Read the how-to guide → · Full API reference →