Real work. Real data.

Built an end-to-end ETL pipeline using Python and SQL to consolidate raw marketing spend, impression, and revenue data into a single Power BI model — giving stakeholders a clear view of which campaigns were profitable and which were burning budget.

• ✓ Engineered ETL pipeline converting raw multi-source data into a clean analytical model
• ✓ Identified highest and lowest ROI campaigns across channels, enabling budget reallocation
• ✓ Delivered interactive Power BI dashboard with drill-through by channel, campaign, and period

Used Python, SQL, and Power BI to analyse Netflix's global content library and streaming performance data — automating an ingestion pipeline to surface content survival rates, genre performance by region, and production investment patterns.

• ✓ Automated data pipeline ingesting and cleaning multi-year global streaming records
• ✓ Surfaced content survival rates and genre performance patterns across regions
• ✓ Delivered strategic programming recommendations backed by quantitative trend analysis

Applied BI tools to transform raw point-of-sale transaction data into a strategic roadmap for a retail café operator — identifying peak trading windows, underperforming SKUs, and inventory inefficiencies.

• ✓ Identified top and bottom performing products by margin, not just revenue
• ✓ Mapped peak trading hours to surface staffing and inventory optimisation opportunities
• ✓ Built Tableau dashboard enabling ongoing self-serve reporting for the store owner

Built a single-page executive Power BI dashboard on a $9.32M-revenue dataset across 6 countries and 5 acquisition channels — surfacing channel mix, geographic concentration, AOV, and a Q4 seasonal dependency pattern.

• ✓ Identified Q4 seasonal revenue concentration — flat 9 months, spike 3 months — as a dependency risk
• ✓ Surfaced near-equal channel mix distribution as an "anti-fragile" acquisition strategy
• ✓ Delivered executive single-page dashboard with KPI scorecards, geographic breakdown, and trend line

A deep dive into the e-commerce conversion funnel — diagnosing a silent CartFlows/WooCommerce plugin conflict and building a fully validated GA4 + GTM event suite to show exactly where and why customers drop off before buying.

• ✓ Mapped the full visitor-to-purchase funnel to identify the highest drop-off stages
• ✓ Diagnosed a silent tracking failure caused by a third-party checkout plugin conflict
• ✓ Validated the complete event suite end-to-end in GA4 DebugView

Cleaned six months of messy web and ad platform data for a modelled portable toilet rental business, then turned it into a 3-page Looker Studio dashboard and a plain-language owner report that shows which campaigns earn bookings and where visitors drop off.

• ✓ Fixed data quality issues that had turned 6 real campaigns into 18 phantom ones
• ✓ Found a nearly 4x gap in cost per lead between channels (R27.61 vs R108.62)
• ✓ Pinpointed the biggest funnel leak: only 17% of sessions reached a pricing or booking page

Cleaned and reconciled four linked data sources for a modelled South African skincare brand, then built a 5-page Looker Studio dashboard and a one-page findings report showing which ads make sales, where shoppers drop off, and how much of the order value is real revenue.

• ✓ Showed only 51% of orders completed, so R872K in order value was really R448K in revenue
• ✓ Found an 18x gap in cost per purchase across five campaigns (R13.59 vs R239.08)
• ✓ Located the funnel leak after the cart: 69% cart abandonment, and email converting at 25.7%