CRM Sales Analytics Platform
Production-grade CRM sales analytics stack on Snowflake with dbt, Dagster orchestration, Elementary observability, and CI/CD via GitHub Actions.
Timeline
Q1 2026
Role
Data Engineer
Team
Solo Project
Status
CompletedTechnology Stack
Key Challenges
- Building reliable CRM analytics layers that remain auditable across source, transform, and mart outputs.
- Operationalizing data quality with both dbt tests and Elementary observability artifacts.
- Integrating orchestration and CI gates to secure production-like delivery workflows.
Key Learnings
- Layered dbt modeling (staging, intermediate, marts) improves maintainability and KPI consistency.
- Dagster asset orchestration provides clearer dependencies and easier failure diagnostics.
- CI validation on dbt and Dagster significantly reduces regression risk before merge.
Executive Summary
This project implements a modern end-to-end analytics platform focused on CRM sales opportunity analysis.
It combines Snowflake, dbt, Dagster, Elementary, and GitHub Actions to produce trusted sales KPIs and operational dashboards with a production-oriented engineering workflow.
Business Outcomes
- Better visibility on pipeline volume, team performance, and product win rates.
- Consistent KPI outputs for quarterly trends and agent-level execution.
- Improved reliability through automated tests, freshness checks, and CI validation.
Architecture Overview
The platform flow is organized as follows:
- Ingestion: CSV sources loaded into Snowflake
RAWschema. - Transformation: dbt layered modeling (
staging->intermediate->marts). - Quality: dbt tests and Elementary metadata for observability.
- Orchestration: Dagster assets and scheduled validation jobs.
- Delivery: GitHub Actions workflows for CI checks on PRs and pushes.
Data Model and KPI Layer
The model set includes:
- Staging: cleaned CRM sources (
stg_*) - Intermediate: enriched opportunities (
int_opportunities_enriched) - Facts and Dimensions:
fct_sales_opportunities,dim_accounts,dim_products,dim_sales_agents - Marts:
mart_sales_team_performance,mart_sales_agent_performance,mart_product_win_rates,mart_quarterly_trends
These marts answer core business questions around team performance, agent outliers, quarter-over-quarter trends, and product-level conversion quality.
Dashboards
Sales Opportunities Dashboard
Teams and Agents Performance
Data Quality and Operations
Quality and SLA controls include:
- dbt model and data tests executed in scheduled jobs
- Source freshness checks using
LOAD_TSthresholds - Elementary runs after dbt builds to persist observability metadata
- Dagster alerting with optional Slack webhook notifications
CI/CD and Governance
GitHub Actions workflows validate the platform before deployment:
- dbt build and tests
- Dagster integration checks
- Secrets-based environment configuration for Snowflake in CI
This creates a defensible release process for analytics code and infrastructure logic.
Repository Highlights
dbt_sales_analysis/models/- staging, intermediate, and marts modelsdbt_sales_analysis/tests/- custom data testsorchestraction/dagster_project/- orchestration definitions.github/workflows/- CI pipelinesdata/- CRM CSV sources and ingestion helper