Most win/loss programs rely on post-mortems — a sales rep fills out a form after the deal closes, weeks after the key conversations happened. The insights are sparse, biased toward recency, and almost never actionable at scale.
I built something different.
The Problem
At Bucketlist Rewards, we had thousands of recorded sales calls sitting in Gong and our CRM but no systematic way to extract patterns from them. The sales team had intuitions about why deals were won or lost, but leadership needed data — not anecdotes — to make repositioning decisions.
The System
The AI win/loss analysis system I designed processes sales call transcripts through a multi-stage pipeline. First, calls are segmented by deal outcome, stage, and buyer persona. Then, a series of Claude-powered extraction agents identify specific pain points mentioned by prospects, competitive alternatives being evaluated, objections raised during the sales process, and the language buyers use to describe their problems.
The outputs feed into a structured database that surfaces patterns across hundreds of conversations simultaneously — something no human analyst could do manually.
The Impact
The system analyzed over 10,000 sales calls and surfaced insights that directly drove our repositioning toward frontline and healthcare verticals. We discovered that buyers in these segments consistently described problems around fairness and transparency in recognition programs — language that wasn't reflected in our existing messaging.
This insight became the foundation of a repositioning strategy that doubled win rates in target segments and generated over $1M in new pipeline.