Driving a Global Revenue Pivot with Power BI & SQL
| Type | Self-initiated portfolio project |
|---|---|
| Dataset | 5 unstandardised CSV files — global brand, $9.32M gross revenue |
| Tools | Python (Pandas) · MySQL · Power BI |
| Outcome | Executive dashboard revealing a balanced acquisition channel mix and a data-backed case to pivot budget toward emerging markets |
The scenario: a global brand generating $9.32M in gross revenue, with a marketing team flying blind — data trapped in five separate, unstandardised CSV files. The mission: transform this fragmented data into a strategic command center that dictates where the next million dollars should be invested.
The Technical Architecture
A dashboard is only as good as the plumbing behind it. Rather than "plug and play" the raw files, I built a scalable data pipeline:
| Step | What Happened |
|---|---|
| 1 — ETL with Python (Pandas) | Cleaned and standardised five distinct datasets — handling missing values, aligning date formats, ensuring currency consistency across global markets |
| 2 — The MySQL Warehouse | Exported the cleaned data into a MySQL database. Setup took longer than anticipated — proving that data engineering is never as easy as it looks — but was crucial for data integrity and refresh speed |
| 3 — Power BI Visualisation | Connected Power BI directly to the SQL server to build a high-contrast, executive-level dashboard designed for rapid decision-making |
Key Discovery — The "Anti-Fragile" Channel Mix
One of the most surprising findings was the Acquisition Channel Mix. Usually, brands lean too heavily on one source (typically Paid Search). Here, the data revealed a remarkably balanced distribution: Affiliate, Email, and Paid Search each held a 22% share.
The Strategic Pivot — Beyond the US Market
While the US was the top-performing country, Monthly Revenue Trends showed a massive, unexplained spike in Q4. Whether a seasonal holiday surge or a viral moment, the momentum was clear.
The real "aha" moment came from comparing the US to countries ranked 2 through 4: India, UK, and Brazil.
Lessons Learned
- Infrastructure matters. The MySQL setup struggle proved that solid data modelling upfront saves hours of troubleshooting in the visualisation phase.
- Data tells you the "where," not always the "why." The dashboard showed when revenue spiked in Q4, but that sparked a deeper conversation with the marketing team to investigate the root cause — bridging data and human intuition.