Transparent intelligence, not a black-box score.
RecommendationIndex structures provider responses into comparable rankings, evidence, consensus, and historical movement.
1. Prompt design
We organize prompts by category, buyer intent, use case, audience, and decision stage so results reflect meaningful commercial questions rather than isolated tests.
2. Multi-model collection
Each prompt is evaluated across supported AI providers. Responses are stored with provider, model, timestamp, source material, and run context.
3. Recommendation extraction
The platform identifies companies, rank positions, supporting language, confidence signals, and cited evidence within each response.
4. Consensus scoring
Scores consider recommendation frequency, position, provider agreement, stability, evidence quality, and prompt coverage. No single provider determines the result.
5. Historical tracking
Repeated scans reveal movement over time—showing whether a company’s visibility is durable, improving, or losing ground.
6. Human-readable reporting
Results are translated into executive summaries, competitive observations, risks, and suggested actions. Scores support decisions; they do not replace judgment.