AIB · Artificial Intelligence Bureau
AI visibility, measured — not guessed
AI visibility is a distribution, not a single prompt. AIB scores unbranded discovery, live presence, and how consistently engines agree. Branded “tell me about {name}” belongs in accuracy — not in the discovery score.
AI search visibility · LLM visibility · AI search monitoring
What AIB measures
- Mention rate on unbranded buyer questions
- AI Live Presence Score (grounded + retrieval) — the UX Presence number
- AI Memory Score (parametric / open-model probes) — not mixed into Presence
- Knowledge gap: live presence minus memory
- Cross-AI consistency
How the Bureau works
- A planner selects prompt families, markets, and languages. We do not run a Cartesian explosion of every cell.
- Each observation stores model, mode, grounding, citations, and cost telemetry.
- Weights live in scoring profiles and vary by company type.
What this is not
- A 0–100 Presence Score is UX. The matrix underneath is the asset.
- One prompt × one run is not a measurement.
Questions
How is the Presence Score calculated?
Default weights: discovery 20%, recommendation 15%, share of voice 15%, citation authority 15%, accuracy 15%, consistency 10%, own-domain 5%, sentiment 5%. Profiles can override weights. See Methodology.
What is not measured?
Exact consumer ChatGPT / Gemini / Copilot UIs, Google AI Overviews as a separate Search product, and any claim that we control model outputs.