Hybrid Workplace Intelligence

An end-to-end Data Analytics & Business Intelligence project designed for a modern SaaS company (Deskbird).
This project solves the challenge of fragmented sales and product usage data by creating a single source of truth for strategic decision-making and automated reporting.

---

Problem Statement

In most SaaS environments, critical data sources such as:

• CRM systems
• Web event tracking
• Product usage logs

are siloed and disconnected.

This makes it difficult to answer high-impact business questions like:

Which company segment is most likely to convert?
How can we optimize workplace utilization for customers?

Solution

This project builds a robust ELT data pipeline that:

• Unifies disparate data sources
• Ensures data quality and integrity
• Enables predictive and prescriptive analytics
• Supports revenue growth and product strategy

---

Architecture & Workflow

The pipeline follows a modern ELT (Extract, Load, Transform) approach using dbt for transformation management.

Extract & Load (EL)

• Synthetic datasets generated using Python (Pandas, Faker)
• Data includes:
  • Sales data
  • CRM data
  • Web events
  • Product usage logs
• Loaded into MySQL (Data Warehouse)

---

Transform (T - dbt)

Transformations are structured into modular layers:

Staging Layer

• Cleans and standardizes raw data
• Maintains source-level integrity

Intermediate Layer

• Resolves relationships early
• Joins core entities (Users, Companies)
• Example: int_user_company_map

Marts Layer

• Final analytical datasets
• Optimized for BI and analytics

Examples: growth_funnel_mart,office_utilization_mart

---

Analyze

• Advanced analytics performed in Python
• Real-time insights for business stakeholders

---

Technologies Used

Category	Tools / Technologies
Data Transformation	dbt
Data Warehouse	MySQL
Data Processing	Python
Analytics	Pandas, Statsmodels
Visualization	Streamlit

---

Key Insights & Analytics

Sales Propensity Modeling

• Logistic Regression reveals:
  • Enterprise leads → Highest conversion probability
  • Mid-Market leads → 72% lower likelihood to convert

---

Funnel Optimization

• Identified major bottleneck:
  • Mid-Market segment
  • High Average Deal Value ($10,636)
  • Low conversion rate

Action: Requires targeted sales strategy improvement

---

Product Utilization Insights

• Non-traditional usage patterns discovered:
  • Enterprise weekend usage
  • High SMB demand for meeting rooms

Impact:
Enabled creation of a new Utilization Advisory Service

---

A/B Testing Results

• New free trial offer:
  • +37.5% conversion uplift
  • Statistically significant improvement

---

Built by Aklilu Abera | Data Analyst