Methodology / Traceable by design
Every score has a chain of evidence.
Citarra measures AI discovery from the buyer question through the engine response to the brands and sources inside it. Nothing begins with a mystery score.
Measurement pipeline
From question to result.
A tracked question represents something a buyer, shopper, or researcher may ask. Repeated runs turn changing responses into a trend that can be inspected.
Question
The exact prompt is stored with every result.
Engine
ChatGPT, Perplexity, and Google AI are reported separately.
Answer
The original response remains available for review.
Entities
Selected brands and their positions are identified.
Sources
Cited domains are recorded when citations appear.
Sample
Counts remain visible behind each metric.
Core metrics
Four measures, four questions.
Visibility score
The percentage of measured answers in which the tracked brand appears.
Share of voice
The tracked brand’s share of mentions among the selected competitor set.
Recommendation position
Where a brand or product appears when an answer gives an ordered recommendation.
Citation frequency
How often a domain appears across the measured answers and questions.
Quality controls
Missing data stays missing.
Provider calls can fail or return incomplete data. An engine below the reliability threshold is excluded from the affected result and the successful call count is shown.
Citarra uses provider APIs, not consumer app interfaces. Individual sessions may show different answers, and we disclose that limitation.
Interpretation
Change is not causation.
Citarra can show that a result changed after a recorded action. It does not claim that the action caused the change.
AI systems, sources, competitors, and market conditions may all move during the same period. The product supports informed judgment. It does not replace it.