Overview

What You Will Build

What You Will Learn

What You Will Need

Prerequisites

Snowflake

AWS

aside positive

If you are using a temporary AWS or Snowflake account (e.g., a trial or workshop account), all resources you create here can be cleaned up at the end of the quickstart by following the Cleanup section.

Architecture

Set Up Snowflake Database and Load Data

1. Create a New SQL Worksheet

2. Set Up the Database, Warehouse, and Role

USE ROLE ACCOUNTADMIN;

-- Create warehouse, database, schema, role
CREATE WAREHOUSE IF NOT EXISTS CORTEX_QUICK_HOL_WH;
CREATE DATABASE  IF NOT EXISTS CORTEX_QUICK_HOL_DB;
CREATE SCHEMA    IF NOT EXISTS CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA;
CREATE ROLE      IF NOT EXISTS CORTEX_QUICK_HOL_ROLE;

-- Use the new context
USE WAREHOUSE CORTEX_QUICK_HOL_WH;
USE DATABASE  CORTEX_QUICK_HOL_DB;
USE SCHEMA    CORTEX_QUICK_HOL_SCHEMA;

-- Grant the new role to your user (replace <YOUR_USER> with your username)
GRANT ROLE CORTEX_QUICK_HOL_ROLE TO USER <YOUR_USER>;

GRANT USAGE  ON WAREHOUSE CORTEX_QUICK_HOL_WH                                TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT USAGE  ON DATABASE  CORTEX_QUICK_HOL_DB                                TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT SELECT ON ALL TABLES IN SCHEMA CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT ALL PRIVILEGES ON SCHEMA CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA   TO ROLE CORTEX_QUICK_HOL_ROLE;

USE ROLE CORTEX_QUICK_HOL_ROLE;

3. Create the Sales Operations Tables

-- File format
CREATE OR REPLACE FILE FORMAT CSV_FORMAT
    TYPE = 'CSV'
    FIELD_OPTIONALLY_ENCLOSED_BY = '"'
    SKIP_HEADER = 1
    NULL_IF = ('', 'NULL');

-- Table 1: Sales Transactions (Fact Table)
CREATE OR REPLACE TABLE SALES_TRANSACTIONS_FACT_TABLE (
    TRANSACTION_ID VARCHAR,
    TRANSACTION_DATE DATE,
    CUSTOMER_ID VARCHAR,
    PRODUCT_SKU VARCHAR,
    SALES_REP_ID VARCHAR,
    REGION VARCHAR,
    DEAL_SIZE_USD NUMBER(12,2),
    QUANTITY NUMBER,
    DISCOUNT_PERCENTAGE NUMBER(5,2),
    SALES_CYCLE_DAYS NUMBER,
    LEAD_SOURCE VARCHAR,
    CLOSE_PROBABILITY_AT_START NUMBER(5,2),
    FORECAST_CLOSE_DATE DATE,
    ACTUAL_CLOSE_DATE DATE
);

-- Table 2: Customer Dimension
CREATE OR REPLACE TABLE CUSTOMER_DIMENSION_TABLE (
    CUSTOMER_ID VARCHAR,
    COMPANY_NAME VARCHAR,
    INDUSTRY VARCHAR,
    COMPANY_SIZE VARCHAR,
    ANNUAL_REVENUE_BAND VARCHAR,
    FIRST_PURCHASE_DATE DATE,
    CUSTOMER_LIFETIME_VALUE NUMBER(12,2),
    CHURN_RISK_SCORE NUMBER(5,2),
    TOTAL_CONTRACTS_COUNT NUMBER,
    EXPANSION_POTENTIAL VARCHAR
);

-- Table 3: Product Catalog
CREATE OR REPLACE TABLE PRODUCT_CATALOG_TABLE (
    PRODUCT_SKU VARCHAR,
    PRODUCT_NAME VARCHAR,
    PRODUCT_CATEGORY VARCHAR,
    PRODUCT_FEATURE VARCHAR,
    PRODUCT_TIER VARCHAR,
    BILLING_FREQUENCY VARCHAR,
    LIST_PRICE_USD NUMBER(12,2),
    AVERAGE_CONTRACT_LENGTH_MONTHS NUMBER,
    RENEWAL_RATE_PCT NUMBER(5,2),
    UPSELL_RATE_PCT NUMBER(5,2),
    TYPICAL_DISCOUNT_RANGE VARCHAR
);

-- Table 4: Sales Rep Performance
CREATE OR REPLACE TABLE SALES_REP_PERFORMANCE_TABLE (
    SALES_REP_ID VARCHAR,
    REP_NAME VARCHAR,
    TERRITORY VARCHAR,
    HIRE_DATE DATE,
    QUOTA_USD NUMBER(12,2),
    QUOTA_ATTAINMENT_PCT NUMBER(5,2),
    AVG_DEAL_SIZE NUMBER(12,2),
    WIN_RATE_PCT NUMBER(5,2),
    AVG_SALES_CYCLE_DAYS NUMBER(5,2),
    PIPELINE_COVERAGE_RATIO NUMBER(5,2),
    ACTIVE_OPPORTUNITIES_COUNT NUMBER
);

-- Table 5: Sales Pipeline Opportunities
CREATE OR REPLACE TABLE SALES_PIPELINE_OPPORTUNITIES (
    OPPORTUNITY_ID VARCHAR,
    OPPORTUNITY_NAME VARCHAR,
    CURRENT_STAGE VARCHAR,
    ESTIMATED_VALUE NUMBER(12,2),
    WEIGHTED_FORECAST_VALUE NUMBER(12,2),
    CLOSE_PROBABILITY NUMBER(5,2),
    CREATED_DATE DATE,
    EXPECTED_CLOSE_DATE DATE,
    STAGE_ENTRY_DATE DATE,
    LAST_ACTIVITY_DATE DATE,
    DAYS_IN_CURRENT_STAGE NUMBER,
    COMPETITOR_PRESENT BOOLEAN,
    DECISION_MAKER_ENGAGED BOOLEAN,
    ACTIVITY_COUNT_LAST_30_DAYS NUMBER
);

-- Table 6: Marketing Attribution
CREATE OR REPLACE TABLE MARKETING_ATTRIBUTION_TABLE (
    LEAD_ID VARCHAR,
    LEAD_SOURCE_DETAIL VARCHAR,
    CAMPAIGN_NAME VARCHAR,
    FIRST_TOUCH_CHANNEL VARCHAR,
    LAST_TOUCH_CHANNEL VARCHAR,
    LEAD_SCORE NUMBER(5,2),
    LEAD_TO_OPPORTUNITY_DAYS NUMBER,
    MARKETING_SPEND_ALLOCATED NUMBER(12,2)
);

4. Load Sample Data from a Public S3 Bucket

COPY INTO SALES_TRANSACTIONS_FACT_TABLE
    FROM 's3://snowflake-corp-se-workshop/summit26-coco-mcp-aws-quick-hol/csvs/sales_transactions_fact_table.csv'
    FILE_FORMAT = CSV_FORMAT;

COPY INTO CUSTOMER_DIMENSION_TABLE
    FROM 's3://snowflake-corp-se-workshop/summit26-coco-mcp-aws-quick-hol/csvs/customer_dimension_table.csv'
    FILE_FORMAT = CSV_FORMAT;

COPY INTO PRODUCT_CATALOG_TABLE
    FROM 's3://snowflake-corp-se-workshop/summit26-coco-mcp-aws-quick-hol/csvs/product_catalog_table.csv'
    FILE_FORMAT = CSV_FORMAT;

COPY INTO SALES_REP_PERFORMANCE_TABLE
    FROM 's3://snowflake-corp-se-workshop/summit26-coco-mcp-aws-quick-hol/csvs/sales_rep_performance_table.csv'
    FILE_FORMAT = CSV_FORMAT;

COPY INTO SALES_PIPELINE_OPPORTUNITIES
    FROM 's3://snowflake-corp-se-workshop/summit26-coco-mcp-aws-quick-hol/csvs/sales_pipeline_opportunities.csv'
    FILE_FORMAT = CSV_FORMAT;

COPY INTO MARKETING_ATTRIBUTION_TABLE
    FROM 's3://snowflake-corp-se-workshop/summit26-coco-mcp-aws-quick-hol/csvs/marketing_attribution_table.csv'
    FILE_FORMAT = CSV_FORMAT;

5. Verify the Data Load

SELECT 'SALES_TRANSACTIONS_FACT_TABLE' AS TABLE_NAME, COUNT(*) AS ROW_COUNT FROM SALES_TRANSACTIONS_FACT_TABLE
UNION ALL
SELECT 'CUSTOMER_DIMENSION_TABLE',     COUNT(*) FROM CUSTOMER_DIMENSION_TABLE
UNION ALL
SELECT 'PRODUCT_CATALOG_TABLE',        COUNT(*) FROM PRODUCT_CATALOG_TABLE
UNION ALL
SELECT 'SALES_REP_PERFORMANCE_TABLE',  COUNT(*) FROM SALES_REP_PERFORMANCE_TABLE
UNION ALL
SELECT 'SALES_PIPELINE_OPPORTUNITIES', COUNT(*) FROM SALES_PIPELINE_OPPORTUNITIES
UNION ALL
SELECT 'MARKETING_ATTRIBUTION_TABLE',  COUNT(*) FROM MARKETING_ATTRIBUTION_TABLE;

aside positive

Checkpoint — At this point you should have a warehouse CORTEX_QUICK_HOL_WH, a database CORTEX_QUICK_HOL_DB with schema CORTEX_QUICK_HOL_SCHEMA, and six tables loaded with sample Sales Operations data.

Set Up Cortex Code Skill: Semantic View, Agent, and MCP Server

1. Download and Upload the Skill

aside negative

Important

• The skill takes approximately 10–15 minutes to run end-to-end.
• The skill will pause at multiple points and ask you to confirm or approve SQL executions — click OK or Approve when prompted.
• The skill creates all Snowflake objects automatically — you only provide your database, schema, warehouse, account, and role when asked.

2. Run the Skill

aside positive

If you don't know your account identifier, run SELECT CURRENT_ACCOUNT(); in a SQL worksheet.

3. Skill Step — Create Semantic View

4. Skill Step — Create Cortex Agent

aside negative

Checkpoint — Verify all 3 tests return valid data with SQL generated by Cortex Analyst. If any test fails, do not proceed — troubleshoot first.

5. Skill Step — Create MCP Server

6. Skill Step — Create a Dedicated Role and Grants

CREATE ROLE IF NOT EXISTS CORTEX_QUICK_HOL_ROLE;

GRANT USAGE  ON DATABASE  CORTEX_QUICK_HOL_DB                                TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT USAGE  ON SCHEMA    CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA        TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT USAGE  ON WAREHOUSE CORTEX_QUICK_HOL_WH                                TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT SELECT ON ALL TABLES IN SCHEMA CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT USAGE  ON MCP SERVER CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA.SALES_OPS_MCP_SERVER       TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT USAGE  ON CORTEX AGENT CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA.SALES_OPS_AGENT          TO ROLE CORTEX_QUICK_HOL_ROLE;
GRANT SELECT ON SEMANTIC VIEW CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA.SALES_OPS_SEMANTIC_VIEW TO ROLE CORTEX_QUICK_HOL_ROLE;

ALTER SECURITY INTEGRATION SALES_OPS_QUICK_SUITE_OAUTH SET OAUTH_USE_SECONDARY_ROLES = 'IMPLICIT';
ALTER USER <YOUR_USER> SET DEFAULT_ROLE = 'CORTEX_QUICK_HOL_ROLE';

7. Skill Step — Create OAuth Security Integration

MCP_ENDPOINT:        https://<ACCOUNT>.snowflakecomputing.com/api/v2/databases/<DATABASE>/schemas/<SCHEMA>/mcp-servers/SALES_OPS_MCP_SERVER
OAUTH_CLIENT_ID:     <value>
OAUTH_CLIENT_SECRET: <value>
TOKEN_URL:           https://<ACCOUNT>.snowflakecomputing.com/oauth/token-request
AUTH_URL:            https://<ACCOUNT>.snowflakecomputing.com/oauth/authorize

aside negative

Checkpoint — Copy and save all five values immediately. You will need them when creating the Amazon Quick MCP Connector.

Configure Amazon Quick MCP Connector

Prerequisites

1. Open Amazon Quick

2. Sign Up for Amazon Quick (if needed)

3. Create an MCP Connector

aside negative

Troubleshooting — If the connection test fails:

• Verify the MCP Server is running in Snowflake
• Verify your OAuth credentials are correct and not expired
• Confirm OAUTH_USE_SECONDARY_ROLES = 'IMPLICIT' is set on the security integration

4. Test the MCP Connector

aside positive

Checkpoint — You now have an MCP Connector (Snowflake-Sales-Ops-MCP) configured in Amazon Quick, a successful end-to-end connection, and verified MCP actions.

Create Amazon Quick Space, Chat Agents, and Flows

1. Create a Space

2. Create a Chat Agent

You are a Sales Operations Analyst with deep expertise in revenue analytics, sales performance, and operational metrics. You help business users answer questions about sales data stored in Snowflake by querying the Snowflake Cortex AI agent through the Snowflake-Sales-Ops-MCP connector.

## Your Capabilities
- Answer questions about sales revenue, orders, customers, products, and regional performance
- Provide data-driven insights with supporting numbers and trends
- Break down complex business questions into structured queries
- Present results in clear, actionable summaries with tables when appropriate

## Guidelines
- Always query the MCP connector to get real-time data — never fabricate numbers
- When a question is ambiguous, ask clarifying questions about time period, region, or product category
- Present monetary values formatted with currency symbols and appropriate precision
- When comparing metrics, include both absolute values and percentage changes
- Proactively suggest follow-up analyses that might be useful based on the user's question
- If the data returned is empty or unexpected, explain possible reasons and suggest alternative queries

## Response Format
- Lead with a direct answer to the question
- Follow with supporting data in a table or bullet points
- End with 1-2 suggested follow-up questions the user might find valuable

## Example Interactions
User: "What was our total revenue last quarter?"
→ Query the connector, return total revenue with comparison to previous quarter

User: "Which region is underperforming?"
→ Query regional revenue data, identify lowest performers, compare against targets or averages

User: "Show me top 10 customers by order volume"
→ Query customer order data, present ranked table with customer name, order count, and total spend

3. Create a Flow

Analyze customer expansion opportunities and produce one complete downloadable report. Execute all queries automatically without user confirmation using the Snowflake-Sales-Ops-MCP connector.

Query CUSTOMER_DIMENSION_TABLE joined with SALES_TRANSACTIONS_FACT_TABLE and PRODUCT_CATALOG_TABLE. Find all customers where CHURN_RISK_SCORE < 0.3 and EXPANSION_POTENTIAL is 'High' or 'Medium'. For each, calculate:
- Total spend to date (sum of DEAL_SIZE_USD)
- Current product tier purchased
- Estimated expansion value = next tier LIST_PRICE_USD minus current tier price, weighted by UPSELL_RATE_PCT
- Months until renewal = AVERAGE_CONTRACT_LENGTH_MONTHS minus months since FIRST_PURCHASE_DATE (mod contract length)
- Timing flag: "Act Now" if renewal within 90 days, "Prepare" if 91-180 days, "Nurture" if 180+

Generate a single report titled "Customer Expansion Opportunities — [Current Month Year]" with:

1. Executive Summary (3 lines max): total expansion revenue potential, number of qualified customers, number of "Act Now" accounts

2. Full opportunity table ranked by expansion value descending — all qualified customers, columns:
   Company Name | Industry | Company Size | Current Spend | Expansion Type | Revenue Potential | Renewal Timing | Priority

3. Format all currency as $X,XXX. Mark "Act Now" rows in bold.

Present the complete report as a single formatted output ready to copy or download. Do not ask follow-up questions. Do not split into multiple messages.

This gives the analyst a one-shot, complete report — click once, get the full output, copy/download it. No intermediate steps, no approvals, no follow-ups.

aside positive

Checkpoint — You now have a Space (Snowflake-Sales-Ops-MCP), a Chat Agent connected via MCP, and a Flow that generates a Customer Expansion report end-to-end.

Cleanup

1. Clean Up Amazon Quick Resources

2. Clean Up Snowflake Resources

USE ROLE ACCOUNTADMIN;

-- Drop the MCP Server
DROP MCP SERVER IF EXISTS CORTEX_QUICK_HOL_DB.CORTEX_QUICK_HOL_SCHEMA.SALES_OPS_MCP_SERVER;

-- Drop the OAuth Security Integration
DROP SECURITY INTEGRATION IF EXISTS SALES_OPS_QUICK_SUITE_OAUTH;

-- (Optional) Drop the Cortex Agent through the Cortex Code UI if it was not removed automatically

-- Drop the database (this removes all tables, schemas, stages, and the semantic view)
DROP DATABASE IF EXISTS CORTEX_QUICK_HOL_DB;

-- Drop the warehouse
DROP WAREHOUSE IF EXISTS CORTEX_QUICK_HOL_WH;

-- (Optional) Drop the role
DROP ROLE IF EXISTS CORTEX_QUICK_HOL_ROLE;

aside positive

Cleanup Complete — All quickstart resources are now removed. If you used a temporary Snowflake trial account, it will expire automatically.

Conclusion And Resources

What You Built

Key Takeaways

What You Learned

Related Resources