	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Win/Loss Analysis Agent

Win/Loss Analysis Agent

Author: Venkata Sudhakar

Understanding why ShopMax India wins or loses deals is critical to improving the sales playbook. A Win/Loss Analysis Agent processes closed deal data - including loss reasons, competitor involvement, deal size, and sales cycle length - and uses Gemini to surface patterns and strategic recommendations that a manager would otherwise spend hours compiling manually.

This tutorial builds a Win/Loss Analysis Agent using Google ADK and Gemini 2.0 Flash. The agent aggregates closed deal data by quarter, identifies the top win and loss drivers, and generates a strategic insight report.

The below example shows the Win/Loss Agent analysing Q1 2026 closed deals for the South India region.

from google.adk.agents import Agent
from google.adk.tools import FunctionTool
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai.types import Content, Part
from collections import Counter

# Closed deals data - Q1 2026
CLOSED_DEALS = [
    {"id": "CD-01", "outcome": "won", "value_lakh": 85, "industry": "manufacturing",
     "loss_reason": None, "competitor": None, "cycle_days": 32, "discount_given": 8},
    {"id": "CD-02", "outcome": "lost", "value_lakh": 120, "industry": "retail",
     "loss_reason": "price too high", "competitor": "CompetitorX", "cycle_days": 45, "discount_given": 5},
    {"id": "CD-03", "outcome": "won", "value_lakh": 45, "industry": "logistics",
     "loss_reason": None, "competitor": "CompetitorY", "cycle_days": 18, "discount_given": 12},
    {"id": "CD-04", "outcome": "lost", "value_lakh": 200, "industry": "retail",
     "loss_reason": "slow response time", "competitor": "CompetitorX", "cycle_days": 60, "discount_given": 3},
    {"id": "CD-05", "outcome": "won", "value_lakh": 60, "industry": "healthcare",
     "loss_reason": None, "competitor": None, "cycle_days": 22, "discount_given": 7},
    {"id": "CD-06", "outcome": "lost", "value_lakh": 35, "industry": "manufacturing",
     "loss_reason": "price too high", "competitor": "CompetitorZ", "cycle_days": 28, "discount_given": 6},
    {"id": "CD-07", "outcome": "won", "value_lakh": 95, "industry": "logistics",
     "loss_reason": None, "competitor": "CompetitorY", "cycle_days": 25, "discount_given": 10},
    {"id": "CD-08", "outcome": "lost", "value_lakh": 70, "industry": "healthcare",
     "loss_reason": "missing feature", "competitor": None, "cycle_days": 38, "discount_given": 0},
]

def get_win_loss_summary(quarter: str) -> dict:
    """Summarise win/loss metrics for the quarter."""
    won = [d for d in CLOSED_DEALS if d["outcome"] == "won"]
    lost = [d for d in CLOSED_DEALS if d["outcome"] == "lost"]
    win_rate = round(len(won) / len(CLOSED_DEALS) * 100, 1)
    won_value = sum(d["value_lakh"] for d in won)
    lost_value = sum(d["value_lakh"] for d in lost)
    avg_cycle_won = round(sum(d["cycle_days"] for d in won) / len(won))
    avg_cycle_lost = round(sum(d["cycle_days"] for d in lost) / len(lost))
    loss_reasons = Counter(d["loss_reason"] for d in lost if d["loss_reason"])
    competitors_lost_to = Counter(d["competitor"] for d in lost if d["competitor"])
    industry_win_rate = {}
    for ind in set(d["industry"] for d in CLOSED_DEALS):
        ind_deals = [d for d in CLOSED_DEALS if d["industry"] == ind]
        ind_won = [d for d in ind_deals if d["outcome"] == "won"]
        industry_win_rate[ind] = round(len(ind_won) / len(ind_deals) * 100)
    return {
        "quarter": quarter,
        "total_deals": len(CLOSED_DEALS),
        "win_rate_pct": win_rate,
        "won_value_lakh": won_value,
        "lost_value_lakh": lost_value,
        "avg_cycle_days_won": avg_cycle_won,
        "avg_cycle_days_lost": avg_cycle_lost,
        "top_loss_reasons": dict(loss_reasons.most_common(3)),
        "competitors_lost_to": dict(competitors_lost_to.most_common(3)),
        "win_rate_by_industry": industry_win_rate
    }

analysis_tool = FunctionTool(func=get_win_loss_summary)

winloss_agent = Agent(
    model="gemini-2.0-flash",
    name="winloss_analysis_agent",
    instruction="""
        You are a Win/Loss Analysis Agent for ShopMax India.
        When given a quarter to analyse:
        1. Fetch the win/loss summary data
        2. Produce a structured insight report:
           - Win rate and revenue summary
           - Key loss drivers with recommended fixes
           - Competitive analysis (who are we losing to and why)
           - Industry performance breakdown
           - Sales cycle comparison (won vs lost)
           - Top 3 strategic recommendations
        Use business language. Make recommendations specific and actionable.
    """,
    tools=[analysis_tool]
)

session_service = InMemorySessionService()
runner = Runner(agent=winloss_agent, app_name="winloss", session_service=session_service)

session = session_service.create_session(app_name="winloss", user_id="sales_director")
message = Content(role="user", parts=[Part(text="Analyse Q1 2026 win/loss data for South India region and give me strategic recommendations.")])
for event in runner.run(user_id="sales_director", session_id=session.id, new_message=message):
    if event.is_final_response():
        print(event.content.parts[0].text)

It gives the following output,

WIN/LOSS ANALYSIS - Q1 2026 (South India)
==========================================

Revenue Summary
  Total Deals   : 8
  Win Rate      : 50%
  Won Revenue   : Rs 285 lakhs
  Lost Revenue  : Rs 425 lakhs

Sales Cycle
  Avg cycle (won) : 24 days
  Avg cycle (lost): 43 days  <- Longer cycles correlate with losses

Top Loss Reasons
  1. Price too high    (2 deals, Rs 155 L lost)
  2. Slow response     (1 deal, Rs 200 L lost)
  3. Missing feature   (1 deal, Rs  70 L lost)

Competitive Losses
  CompetitorX : 2 losses (retail sector, price-driven)

Industry Win Rates
  Logistics     : 100% | Manufacturing: 50%
  Healthcare    : 50%  | Retail       : 0%

Top 3 Recommendations
  1. Introduce a competitive pricing floor for retail - we lost Rs 320 L
     in retail with 0% win rate. Address CompetitorX directly.
  2. Set a 24-hour response SLA for deals above Rs 50 L. Slow response
     cost us our biggest deal this quarter.
  3. Conduct a product gap review for healthcare. One loss due to missing
     feature - identify and roadmap the gap.

The Win/Loss Analysis Agent turns raw CRM data into strategic intelligence. Run it quarterly with different regional filters to compare performance across territories. Feed in free-text call notes using Gemini to extract qualitative insights that structured data alone cannot capture.

Send your comments, suggestions or queries regarding this site to [email protected].