How the scan works
Four layers, one honest read of the page.
The scan is a four-layer pipeline — technical hygiene, schema, AI visibility, and accessibility. Each layer says what it measures, what a finding means, and where its blind spots are. Goblin Mode (default, every tier) keeps human review on every delivered fix. Autopilot Mode — available on Mention Engine and Goblin Forge — posts pre-classified low-risk content on a schedule you approve. You choose the mode.
Recon
Before the layers run
- Recon infers category, topic, summary, early stack fingerprints, and up to two competitors when the operator did not provide competitors.
- Auto-discovered competitors are flagged research inferences the client confirms, never asserted facts.
- The first homepage fetch is cached so later audit nodes do not repeatedly probe the same site.
The pipeline
The four layers
Technical hygiene
What it measures
Crawl path, robots and llms.txt presence, indexability, head tags, canonical tags, and static Core Web Vitals proxies.
What a finding means
A hygiene issue means a crawler or answer engine may have trouble reading, summarizing, or trusting the page shell.
Static fetches have a blind spot: an unreadable, JS-rendered, SPA, or WAF-blocked site is flagged, never scored 0.
Schema
What it measures
JSON-LD types present, parseable, and appropriate for the business model.
What a finding means
A schema issue means the page is harder to parse as an entity or service.
Schema is hygiene, not a citation lever. It helps engines parse you; it does not earn a citation.
AI visibility
What it measures
Answer-engine citation gaps, competitor citation graph differences, freshness, topical depth, and third-party platform presence.
What a finding means
A visibility issue means the answer surface currently points elsewhere or lacks a verifiable source pattern we can work on.
Citation movement is measured over re-runs. No citation number is guaranteed.
Accessibility
What it measures
Static WCAG 2.1 AA and Section 508 pre-screen signals such as labels, image alt text, language, and landmarks.
What a finding means
An accessibility issue means the page may block assistive technology or fail compliance review.
Automated checks catch only part of WCAG; full conformance needs rendered axe checks and human review.
Severity
What each finding level means
| HIGH | A live blocker, noindex, unreadable surface, or missing critical entity path that needs human review first. |
| MED | A fixable issue that can materially improve crawlability, clarity, accessibility, or measurement quality. |
| LOW | A cleanup or supporting signal that improves the surface without being a standalone citation lever. |
What we won't claim
The promises we hold ourselves to
- We never fabricate metrics, clients, testimonials, or citations.
- Schema and llms.txt are hygiene, never a promised citation lever.
- Unreadable, JS-rendered, SPA, or WAF-blocked pages are flagged, never scored 0.
- Service and government sites correctly use Service, Offer, and OfferCatalog patterns; we do not prescribe commerce schema for them.
- Every recommended change is human-reviewed and human-gated by default. Nothing auto-deploys or auto-sends — unless you opt into Autopilot Mode's scheduled low-risk posts on Mention Engine and Goblin Forge.
- Mock, sample, and demo paths are illustrative and are never reported as real passes.
- The refund covers the work, never a citation number.