How SignalFox measures what AI says about your brand.

Every SignalFox audit covers four dimensions:

• Presence — does the model surface your brand for category-defining prompts?
• Accuracy — does it describe your product, pricing, and positioning correctly?
• Sentiment — is the tone neutral, favorable, or critical?
• Competitive context — which alternatives appear alongside (or instead of) you?

We combine first-party data — your live site crawl, structured data, llms.txt, and sitemap — with multi-provider AI probes (OpenAI, Google Gemini, Anthropic Claude, Perplexity, Microsoft Copilot) and public attribution sources the models commonly cite.

We never train models on your data, and we never surface another customer's data inside your workspace.

Each audit runs a standardized set of buyer-intent prompts derived from your category, plus a custom set generated from your own positioning. Prompts are versioned so re-runs over time are directly comparable.

The composite Visibility Score is a weighted average of presence, accuracy, sentiment, and share-of-voice across providers. Each provider is scored independently — we never copy a score from one model to another, and we never fabricate scores when a probe has not completed.

This methodology page is reviewed quarterly. Provider weights, prompt sets, and scoring formulas evolve as the generative search landscape changes; we publish the changelog inside the app under Settings → Methodology.

• If we don't have evidence, we say so — no placeholder scores.
• Every claim in the report links to the underlying probe or page.
• Recommendations are prioritized by effort vs. impact, not vendor preference.