How rAsh Score is Calculated
Traditional SEO evaluated website crawlers. rAsh Score measures foundation AI perception, recommendation share-of-voice, and model steerability in real-time.
Recommendation Authority
Measures how frequently, explicitly, and authoritatively foundation models (OpenAI, Gemini, Claude, Groq, NVIDIA) recommend your brand when queried for top choices in your sector.
Sentiment Tone
NLP sentiment polarity analysis across AI outputs. Evaluates whether the generated descriptions of your products carry positive, neutral, or negative sentiment qualifiers.
Prominence & Rank Position
Calculates where your brand appears in numbered rankings and lists. In conversational AI, brands in positions #1-#3 capture over 85% of downstream click-throughs.
Factual Accuracy
Detects model hallucinations vs verified grounding data regarding your brand pricing, features, and executive leadership.
How the Evaluation Pipeline Works
Prompt Standardization
Standardized buyer-intent prompts are dispatched across 6 frontier LLM architectures.
Entity Extraction
Structured parser computes co-occurrence, mention order, and sentiment polarity.
Longitudinal Indexing
Results are persisted daily to Turso LibSQL database to track algorithm drift over time.
🔥 The Story Behind the Name "rAsh"
The name carries the signature of creator Prashanth Kumar (prashanth → rAsh). It also carries a blunt commercial warning: in the age of ChatGPT and answer engines, if AI doesn't recommend your brand, your visibility turns to ash.
Frequently Asked Questions
Why might my score vary between runs?
Large language models are inherently non-deterministic. We run multiple standardized evaluation queries to establish stable 7-day moving averages.
How can I improve my brand's rAsh Score?
Large Language Model Optimization (LLMO) requires structured JSON-LD schema, authoritative co-occurrence in industry whitepapers, and entity grounding. Check out our Pro Plan for customized LLMO action plans.