Methodology
Testing Methodology
Our AI Search Visibility Audit uses a 17-section structured assessment across five signal layers. It tests how AI platforms currently interpret your business, not how they should.
It does not manipulate AI systems. It observes what AI platforms can currently find, read and understand about a business – and identifies where interpretation is incomplete, inconsistent or absent.
What We Test
Six platforms: ChatGPT, Claude, Gemini, Perplexity, Copilot and Google AI. Five signal layers: Identity Clarity, Subject Authority, Meaning Architecture, Ecosystem Validation, Signal Consistency.
Two Scores
AI Visibility Score (weighted)
A weighted score reflecting the relative importance of each signal layer for your specific business category.
Structural Reference Score (unweighted)
An unweighted score showing raw signal completeness across all 17 sections.
Limitations
Inclusion in AI answers cannot be guaranteed. AI systems evolve. Weightings shift. Retrieval varies by phrasing, context, and factors that are not always visible externally. We engineer stability. We do not control platforms.
The audit never assumes something does not exist simply because it was not detected. AI interpretation is probabilistic, not deterministic.
Frequently asked questions
How do you test AI visibility across platforms?
Each brand is tested against 50+ queries spanning brand, category, comparison, and buyer-intent. We run each query on ChatGPT, Claude, Gemini, Perplexity, Copilot, and Google AI Overviews. We record whether the brand is cited, which competitors are cited, and what sources the AI references.
How often is your data refreshed?
Citation tests run weekly. Ecosystem audits are quarterly. Site crawls are daily. Our UK AI Visibility Signal Study (745 businesses across 80+ sectors) is updated monthly.
Can I trust the methodology?
Our methodology is public, replicable, and peer-reviewed by independent SEO and AI researchers. We publish every data source we use, every test query, and every scoring rule. 100% of our published stats link back to the underlying data.