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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Sales Coaching Agent

Sales Coaching Agent

Author: Venkata Sudhakar

Sales managers at ShopMax India oversee teams of 8 to 12 reps across multiple cities. Providing personalised coaching to each rep is time-consuming. A Sales Coaching Agent analyses each rep's activity data, conversion rates, and deal outcomes to identify specific skill gaps and deliver targeted coaching recommendations - allowing managers to run focused 1:1s backed by data.

This tutorial builds a Sales Coaching Agent using Google ADK and Gemini 2.0 Flash. The agent evaluates rep performance across key sales metrics, compares against team benchmarks, and generates a personalised coaching plan.

The below example shows the Sales Coaching Agent analysing the performance of a rep in the Hyderabad territory.

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

# Team benchmark metrics
BENCHMARKS = {
    "calls_per_week": 30,
    "demo_conversion_pct": 35,
    "proposal_to_close_pct": 40,
    "avg_deal_size_lakh": 25,
    "avg_cycle_days": 28
}

# Rep performance data
REP_DATA = {
    "Suresh": {
        "calls_per_week": 22,
        "demo_conversion_pct": 28,
        "proposal_to_close_pct": 55,
        "avg_deal_size_lakh": 32,
        "avg_cycle_days": 35,
        "deals_won_qtd": 6,
        "deals_lost_qtd": 4,
        "city": "Hyderabad"
    }
}

def get_rep_performance(rep_name: str) -> dict:
    """Fetch performance metrics for a sales rep."""
    if rep_name not in REP_DATA:
        return {"error": "Rep not found: " + rep_name}
    rep = REP_DATA[rep_name]
    gaps = []
    strengths = []
    for metric, benchmark in BENCHMARKS.items():
        rep_val = rep[metric]
        pct_of_benchmark = round(rep_val / benchmark * 100)
        if pct_of_benchmark < 80:
            gaps.append({"metric": metric, "rep_value": rep_val,
                          "benchmark": benchmark, "pct_of_benchmark": pct_of_benchmark})
        elif pct_of_benchmark >= 120:
            strengths.append({"metric": metric, "rep_value": rep_val,
                               "benchmark": benchmark, "pct_of_benchmark": pct_of_benchmark})
    win_rate = round(rep["deals_won_qtd"] / (rep["deals_won_qtd"] + rep["deals_lost_qtd"]) * 100)
    return {
        "rep": rep_name,
        "city": rep["city"],
        "win_rate_pct": win_rate,
        "metrics": rep,
        "skill_gaps": gaps,
        "strengths": strengths
    }

def get_coaching_resources(skill_gap: str) -> dict:
    """Return coaching resources for a specific skill gap."""
    resources = {
        "calls_per_week": {
            "focus": "Prospecting discipline",
            "actions": ["Block 2 hours daily for cold outreach",
                        "Use LinkedIn Sales Navigator for warm leads",
                        "Role-play cold call opening with manager"]
        },
        "demo_conversion_pct": {
            "focus": "Discovery and qualification",
            "actions": ["Complete SPIN selling refresher module",
                        "Record next 3 discovery calls for review",
                        "Ask 3 pain-point questions before demo"]
        },
        "avg_cycle_days": {
            "focus": "Deal velocity and urgency creation",
            "actions": ["Set mutual action plans with prospects",
                        "Create time-limited offer for stalled deals",
                        "Review stalled deals in weekly 1:1"]
        }
    }
    return resources.get(skill_gap, {"focus": "General improvement", "actions": ["Discuss with manager"]})

perf_tool = FunctionTool(func=get_rep_performance)
coaching_tool = FunctionTool(func=get_coaching_resources)

coaching_agent = Agent(
    model="gemini-2.0-flash",
    name="sales_coaching_agent",
    instruction="""
        You are a Sales Coaching Agent for ShopMax India.
        When given a rep name:
        1. Fetch their performance metrics and identify gaps
        2. For each skill gap, fetch coaching resources
        3. Produce a personalised coaching plan:
           - Performance snapshot (strengths and gaps vs benchmark)
           - Top 2 priority coaching areas with specific actions
           - 30-day improvement goals with measurable targets
           - Manager talking points for the 1:1
        Be encouraging but specific. Focus on behaviours, not outcomes.
    """,
    tools=[perf_tool, coaching_tool]
)

session_service = InMemorySessionService()
runner = Runner(agent=coaching_agent, app_name="coaching", session_service=session_service)

session = session_service.create_session(app_name="coaching", user_id="sales_mgr")
message = Content(role="user", parts=[Part(text="Generate a coaching plan for Suresh ahead of our monthly 1:1.")])
for event in runner.run(user_id="sales_mgr", session_id=session.id, new_message=message):
    if event.is_final_response():
        print(event.content.parts[0].text)

It gives the following output,

COACHING PLAN - Suresh (Hyderabad)
====================================

Performance Snapshot (Q2 2026)
  Win Rate      : 60%  (team avg: ~55%)
  Deals Won     : 6  |  Deals Lost: 4

Strengths
  + Proposal-to-close rate: 55% (benchmark 40%) - excellent closing
  + Avg deal size: Rs 32 L (benchmark Rs 25 L) - targets right accounts

Skill Gaps
  - Calls per week: 22 (benchmark 30) - 73% of target
  - Demo conversion: 28% (benchmark 35%) - 80% of target

Priority Coaching Areas

1. Prospecting Discipline (Calls: 22 vs 30 target)
   Actions:
   - Block 2 hours daily for cold outreach (9-11am)
   - Use LinkedIn Sales Navigator for warm leads
   - Role-play cold call opening with manager this week
   30-day Goal: Reach 28 calls/week by end of April

2. Discovery and Qualification (Demo conversion: 28% vs 35%)
   Actions:
   - Complete SPIN selling refresher module
   - Record next 3 discovery calls for joint review
   - Ask 3 pain-point questions before scheduling demo
   30-day Goal: Demo conversion rate to 32% by end of April

Manager Talking Points
  - Acknowledge strong closing and deal size - these are real skills
  - Challenge: "If we get more leads into your funnel, your close rate
    means quota is very achievable. What is stopping 30 calls a week?"

The Sales Coaching Agent scales personalised coaching across a large team without requiring the manager to manually review spreadsheets. Extend it to analyse call recordings using Gemini audio understanding, or feed in CRM activity logs automatically each week to produce a coaching brief before every 1:1.

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