Development of an AI system for analyzing the reasons for customer refusal (Lost Deal Analysis)
Lost deals are a valuable source of learning. In practice, they're lost, marked as "lost," and forgotten. The AI system systematizes the causes of losses, identifies patterns, and translates insights into actionable recommendations for the sales team.
Data Collection
Sources:
- CRM lost reason (often filled out formally - "the client chose another")
- Transcriptions of the last conversations before the loss
- Email correspondence with the client
- Post-mortem manager notes
LLM-enrichment: Based on all sources, the model extracts real reasons—more detailed than formal CRM reasons. "The client chose a competitor" → "The competitor offered SAP integration that we don't have; the price was comparable."
Pattern analysis
Clustering: NLP clustering of loss causes (K-Means on sentence embeddings). Identifying the top N real causes.
Correlation Analysis: What deal attributes correlate with a loss to competitor X? Company size, industry, and stage of the loss.
Competitive Intelligence: Automatic competitive analysis: who are we losing to, in which segments, and for what reasons.
Actionable Output
Weekly report: top reasons for losses this week, trends vs. previous period, and competitors and who lost. Quarterly deep dive: patterns for the quarter with recommendations for product, price, and process.