Lead Discovery
Two-phase scoring explained
A deep dive into how the keyword pre-filter and AI scoring pipeline work together.
Phase 1: Keyword pre-scoring (0 to 0.5)
The first phase runs instantly on every new post. It checks for signals that suggest the post might be relevant to your product:
- Keyword match — each matching keyword from your product adds +0.1 to the score
- Intent phrases — 20 built-in phrases like "looking for", "anyone recommend", "need help with" each add +0.08
- Question indicators — posts containing question marks get +0.05
- Reddit engagement — posts with 10+ upvotes get +0.03, 50+ upvotes get +0.06
- Discussion activity — posts with 5+ comments get +0.02, 20+ comments get +0.05
Note: Phase 1 scores are capped at 0.5. Posts scoring below 0.2 are discarded — they don't contain enough signals to justify AI analysis.
Phase 2: AI analysis (0 to 1.0)
Posts that score 0.2+ in Phase 1 are sent to our AI for deep analysis. The AI reads the full post content alongside your product description and evaluates four dimensions:
- Solution-seeking — is the person actively looking for a tool or solution?
- Pain points — are they expressing frustration with their current approach?
- Buying signals — do they mention budget, timeline, or readiness to purchase?
- Commercial intent — does the context suggest a business need rather than casual browsing?
AI output
For each post, the AI returns:
- Relevance score (0 to 1) — the final score used for ranking and filtering
- Relevance reason — a short explanation of why this post was scored this way
- Intent signals — specific buying signals detected (e.g., "comparing tools", "budget allocated")
- Sentiment — the post author's tone: Positive, Negative, Neutral, or Question