The three-layer audit is the diagnostic framework behind Kodiac Audit, the first step of the Kodiac journey. Most AI visibility tools scan a single layer - usually AI output - and tell you to write more content. The three-layer audit scans output, input, and ecosystem so you can see the actual root cause: not just what AI says about you, but why it understands you the way it does.
Each layer answers a different question. Each layer has its own metrics, sources, and interventions. None of them work in isolation. The three together form the only complete diagnostic for AI-mediated brand discovery.
How ChatGPT, Perplexity, Gemini, Claude, Grok and Meta AI describe you right now. Visibility scored 0–100 per system. Inaccuracies flagged. Competitive share of voice tracked on every scan.
Per-page AI-readiness scored across 10 dimensions. Crawlability, structured data, semantic clarity, factual density, brand entity clarity. Fix list ranked by impact, with the uplift measured after each fix.
Every third-party source AI cites, weighted by how much AI relies on it. Reddit, Wikipedia, G2, news. The 74–92% of brand representation that owned content can never reach (University of Toronto, 2025).
The output layer is the surface most teams already worry about - what a buyer actually sees when they ask an AI assistant about your category. Kodiac monitors all six major AI systems on the cadence you set, scores each one on a 0–100 visibility scale, flags accuracy issues, and benchmarks share of voice against named competitors.
ChatGPT (OpenAI), Perplexity, Google Gemini, Anthropic Claude, xAI Grok and Meta AI. Single-engine monitoring is no longer sufficient because AI search has fragmented permanently.
Composite score per system, plus an aggregate across all four. Trended over time. Benchmarked against your tracked competitors. Designed for board-ready reporting that holds up to scrutiny.
Verifies AI claims against your Source of Truth registry. Flags critical, moderate, and minor mismatches. Pricing inaccuracies and product claims surfaced for immediate intervention.
Drill into the actual AI response for any tracked prompt. See which sources AI cited, what claims it made, and what changed since the last audit cycle.
The input layer measures whether AI crawlers can read, understand, and cite your owned content. Kodiac scores every page across 10 dimensions of AI-readiness, identifies structured data gaps, and produces a prioritised fix list, with the score uplift measured after each fix, not generic recommendations.
Crawlability for AI, structured data, semantic clarity, authority signals, content freshness, factual density, AI-readability, brand entity clarity, speed and accessibility, competitor gap. Each dimension scored, weighted, and ranked for impact.
Maps every schema.org type present, missing, or invalid. Exports Product, Offer, Organization, FAQPage, Article and Service markup as JSON-LD. Ready-to-deploy structured data.
Verifies your robots.txt and meta robots tags are not accidentally blocking GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and other AI retrieval bots. The single most common cause of zero AI visibility.
Every low-scoring page maps directly to a fix in Kodiac Content. Fix the record in Kodiac, publish the corrected version to AI, and watch the score recover. Your CMS stays canonical for your website.
The ecosystem layer is what the rest of the market calls Source Intelligence. University of Toronto research found AI search engines cite earned third-party sources up to 92% of the time when describing brands - and most AI visibility tools either ignore this layer entirely or treat it as a footnote. For Kodiac Audit it is a core layer, not a footnote.
Reddit, Wikipedia, Wikidata, G2, Hacker News, ProductHunt, industry news, trade press, partner directories, forum discussions, review sites: every source AI cites when describing your brand, collected on each scan and tracked with sentiment alerts.
Every source weighted by how heavily each AI system relies on it when describing your brand, from the URLs Perplexity and Gemini cite to the sources inferred behind ChatGPT and Claude answers.
Every fix is measured against a baseline scan and the first scan after the change, with a confidence band on the result. Real numbers, not generic recommendations.
Every source you act on moves through not started, in progress, remediated or failed, with mention volume before and after and a 30 to 60 day verification window.
If you scan only the output layer, every recommendation collapses to “write more content.” If you scan only the input layer, you optimise pages that AI never reaches. If you scan only the ecosystem layer, you miss accuracy issues on your own site. The three layers are diagnostic siblings - none of them stands alone.
| If you only scan… | You see | You miss | The recommendation collapses to |
|---|---|---|---|
| Output (Layer 01) | What AI says today | Why AI says it | “Publish more content.” |
| Input (Layer 02) | Owned content gaps | That owned content can be as little as 8–26% of the equation | “Add more schema markup.” |
| Ecosystem (Layer 03) | Third-party signals | That AI is misreading your own site | “Get more press coverage.” |
| All three | The full picture | Nothing | A ranked, prioritised fix list per layer |
Most organisations begin with the audit. Content gets the information behind the answers approved and kept current. The agent is where it leads.
See how AI describes your organisation and identify what needs attention.
Explore AuditConnect and maintain trusted information for AI to use.
Explore ContentVisibility is the entry point. The destination is a Brand Agent of your own: one your customers and their AI can ask directly, answering only from information you have approved, remembering with permission, on the AI model you choose.
Explore Kodiac AgentThe diagnosis is where control begins. Scan your brand every month across six AI systems. Website AI-readiness score. The sources shaping your brand, weighted by influence. Prioritised fixes.