ACCEPTANCE CRITERIA (12/12)

☑ LeadScoringService created with `calculateScore`, `classifyLead`, `getScoreBreakdown` methods

☑ Scoring criteria implemented: customer type (CNPJ/CPF), volume interest, urgency, data provided, engagement, history

☑ Classification thresholds: HOT (70+), WARM (30-69), COLD (0-29)

☑ SegmentationService created with `detectSegment` and `getBANTStatus` methods

☑ BANT adapted for Propek: Budget, Authority, Need, Timeline

☑ Customer segment detection: consumidor vs revendedor vs indefinido

☑ Integration in whatsapp.handler.ts — non-blocking background scoring after every message

☑ HOT leads flagged for commercial handoff in logs

☑ Database migration 016 for lead_scores table with proper indexes

☑ All operations non-blocking (do not affect main message flow)

☑ TypeScript compiles without errors (`npx tsc --noEmit`)

☑ Follows existing code patterns (query, logger, service class pattern)

ARQUIVOS (5)

src/services/lead-scoring.service.ts Created

src/services/segmentation.service.ts Created

src/handlers/whatsapp.handler.ts Modified (added imports + background scoring)

src/migrations/016_create_lead_scores.sql Created

docs/stories/story-3.2-bot-phase2-lead-scoring-sdr.md Created