How to Build an AI-Powered Win/Loss Analysis System

You can not improve what you do not measure. An ai win loss analysis of your sales deals turns closed deals into a playbook. I run this quarterly for clients and it consistently reveals why deals die. The patterns are always there.

What You Need Before Starting

• Python 3.8+
• Claude or GPT API key
• CRM with deal history and notes
• pandas installed

Step 1: Export Win/Loss Data

Pull all closed deals from the last 12 months with notes and timeline data.

import pandas as pd

def export_closed_deals(crm_client, months=12):
    deals = crm_client.get_deals(
        status=["won", "lost"],
        date_range=f"last_{months}_months",
        include=["notes", "activities", "timeline"]
    )
    df = pd.DataFrame(deals)
    df["outcome"] = df["status"].map({"won": 1, "lost": 0})
    return df

Step 2: Analyze with AI

Feed deal summaries to Claude. The AI finds themes humans miss across hundreds of deals.

import anthropic

def analyze_deals(lost_deals):
    client = anthropic.Anthropic()
    summaries = ""
    for _, deal in lost_deals.head(50).iterrows():
        summaries += f"Deal: {deal['name']}, Value: ${deal['amount']:,.0f}, "
        summaries += f"Stage Lost: {deal['lost_stage']}, Notes: {deal['notes']}\n\n"

    message = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=2000,
        messages=[{
            "role": "user",
            "content": f"Analyze these lost deals. Identify the top 5 patterns.\n\n{summaries}"
        }]
    )
    return message.content[0].text

Step 3: Build the Comparison Matrix

Compare won vs lost deals across key dimensions.

def build_comparison(df):
    comparison = {}
    for metric in ["amount", "cycle_days", "meetings_held", "stakeholders"]:
        comparison[metric] = {
            "won_avg": df[df["outcome"] == 1][metric].mean(),
            "lost_avg": df[df["outcome"] == 0][metric].mean(),
        }
    return comparison

Step 4: Generate the Report

Combine quantitative analysis with AI insights into one actionable document.

def generate_report(comparison, ai_analysis, output_path):
    report = "# Win/Loss Analysis Report\n\n"
    report += "| Metric | Won (Avg) | Lost (Avg) |\n"
    report += "|--------|-----------|------------|\n"
    for metric, data in comparison.items():
        report += f"| {metric} | {data['won_avg']:.1f} | {data['lost_avg']:.1f} |\n"
    report += f"\n## Pattern Analysis\n\n{ai_analysis}\n"
    with open(output_path, "w") as f:
        f.write(report)

Step 5: Track Quarterly Improvements

Run the analysis quarterly. Compare loss patterns over time.

def track_trends(current_quarter, previous_quarters):
    loss_rates = []
    for q in previous_quarters + [current_quarter]:
        total = len(q)
        lost = len(q[q["outcome"] == 0])
        loss_rates.append({
            "quarter": q["quarter"].iloc[0],
            "loss_rate": lost / max(total, 1),
        })
    return loss_rates

What to Build Next

Feed your win/loss patterns back into your battle cards and sales training. The analysis is useless if it stays in a report nobody reads.

Related Reading

• Creating Automated Win-Loss Analysis - automated win loss analysis guide
• AI for Customer Retention - ai customer retention analysis
• Using Claude for Business Analysis - claude ai business analysis