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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Lead Scoring Agent

Lead Scoring Agent

Author: Venkata Sudhakar

In a high-volume sales environment, not every lead deserves immediate attention. ShopMax India receives hundreds of inbound enquiries daily across its B2B electronics division. A Lead Scoring Agent automatically evaluates each lead on behavioural signals, firmographic data, and buying intent - so the sales team focuses on prospects most likely to convert.

This tutorial builds a Lead Scoring Agent using Google ADK and Gemini 2.0 Flash. The agent accepts lead details, scores them on a 100-point scale, and routes high-priority leads to senior account managers while low-priority leads enter a nurture sequence.

The below example shows the Lead Scoring Agent evaluating a B2B lead from a manufacturing company in Pune.

import os
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

# Lead scoring weights (out of 100)
SCORE_WEIGHTS = {
    "company_size": 20,    # employees
    "annual_revenue": 25,  # Rs crores
    "industry_fit": 20,    # match with target verticals
    "buying_intent": 20,   # website visits, demo requests
    "budget_confirmed": 15 # has confirmed budget
}

def score_lead(company_name: str, employees: int, annual_revenue_cr: float,
               industry: str, demo_requested: bool, budget_confirmed: bool) -> dict:
    """Score a B2B lead and return priority routing decision."""
    score = 0

# Company size scoring
    if employees >= 500:
        score += 20
    elif employees >= 100:
        score += 12
    elif employees >= 50:
        score += 6

# Revenue scoring
    if annual_revenue_cr >= 100:
        score += 25
    elif annual_revenue_cr >= 25:
        score += 15
    elif annual_revenue_cr >= 5:
        score += 8

# Industry fit (target: manufacturing, retail, logistics)
    target_industries = ["manufacturing", "retail", "logistics", "healthcare"]
    if industry.lower() in target_industries:
        score += 20
    else:
        score += 8

# Buying intent
    if demo_requested:
        score += 20
    else:
        score += 5

# Budget
    if budget_confirmed:
        score += 15

# Routing decision
    if score >= 75:
        route = "Senior Account Manager - immediate follow-up within 2 hours"
        priority = "HOT"
    elif score >= 50:
        route = "Inside Sales Team - follow-up within 24 hours"
        priority = "WARM"
    else:
        route = "Marketing Nurture Sequence - automated email drip"
        priority = "COLD"

return {
        "company": company_name,
        "score": score,
        "priority": priority,
        "routing": route,
        "breakdown": {
            "employees": employees,
            "revenue_cr": annual_revenue_cr,
            "industry": industry,
            "demo_requested": demo_requested,
            "budget_confirmed": budget_confirmed
        }
    }

def get_lead_history(company_name: str) -> dict:
    """Check CRM for previous interactions with this company."""
    # Simulated CRM lookup
    history = {
        "Pune Auto Components Ltd": {
            "previous_enquiries": 2,
            "last_contact": "2026-01-15",
            "outcome": "proposal sent, no response"
        }
    }
    return history.get(company_name, {"previous_enquiries": 0, "last_contact": None, "outcome": "new lead"})

# Create tools
score_tool = FunctionTool(func=score_lead)
history_tool = FunctionTool(func=get_lead_history)

# Build the agent
lead_agent = Agent(
    model="gemini-2.0-flash",
    name="lead_scoring_agent",
    instruction="""
        You are a Lead Scoring Agent for ShopMax India B2B Sales.
        When given lead details:
        1. Check the CRM for prior interaction history
        2. Score the lead using the scoring tool
        3. Provide a summary with score, priority, routing, and recommended next action
        4. Mention any prior history that affects the approach
        Keep the summary concise and actionable for the sales team.
    """,
    tools=[score_tool, history_tool]
)

session_service = InMemorySessionService()
runner = Runner(agent=lead_agent, app_name="lead_scorer", session_service=session_service)

# Evaluate a lead
session = session_service.create_session(app_name="lead_scorer", user_id="sales_ops")
prompt = """
    Evaluate this lead:
    Company: Pune Auto Components Ltd
    Industry: Manufacturing
    Employees: 350
    Annual Revenue: Rs 45 crores
    Demo Requested: Yes
    Budget Confirmed: No
"""
message = Content(role="user", parts=[Part(text=prompt)])
for event in runner.run(user_id="sales_ops", session_id=session.id, new_message=message):
    if event.is_final_response():
        print(event.content.parts[0].text)

It gives the following output,

Lead Evaluation: Pune Auto Components Ltd

Score: 67 / 100 - WARM LEAD

Breakdown:
  Company Size (350 employees): 12 pts
  Annual Revenue (Rs 45 Cr):    15 pts
  Industry Fit (Manufacturing): 20 pts
  Demo Requested (Yes):         20 pts
  Budget Confirmed (No):         0 pts
  Total: 67 pts

Priority: WARM
Routing: Inside Sales Team - follow-up within 24 hours

CRM History: 2 prior enquiries (last contact Jan 2026).
A proposal was sent previously with no response.
Recommended: Reference the prior proposal in outreach - ask
what changed and whether budget has been approved.

The Lead Scoring Agent can be extended to pull live data from CRM systems like Salesforce or Zoho, read website visit logs, and use Gemini to analyse free-text notes from previous calls. In production, integrate it with your inbound lead pipeline so every new enquiry is automatically scored and routed before a human ever reads it.

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