A 3-page Power BI dashboard simulating end-to-end affiliate marketing analytics for a blockchain hardware wallet company — covering the full funnel from ad impressions through approved conversions, plus affiliate/network profitability analysis.
This project analyzes affiliate acquisition performance across multiple dimensions — country, device, campaign type, affiliate network, and traffic source — to answer the kind of questions a Data Analyst (Acquisition) role would own:
- Which affiliates and networks drive the most revenue and profit, and at what cost?
- Where does the acquisition funnel leak the most (impressions → clicks → landing → checkout → purchase → approval)?
- How does conversion and approval rate vary by device, campaign type, and affiliate?
- What's the ROI/ROAS by affiliate, and which partners are worth scaling vs cutting?
1. Executive Overview High-level KPIs (Net Revenue, Profit, ROAS, ROI, New Customers), revenue trend, revenue by affiliate type/device/country, and new vs returning customer mix.
2. Affiliate Performance Per-affiliate and per-network breakdown — Top 5 affiliates by revenue/ROAS/purchases, approved vs rejected conversions, revenue vs cost scatter, and a full sortable performance table with ROAS/ROI per affiliate.
3. Acquisition Funnel Step-by-step funnel (Impressions → Clicks → Landing Visits → Configurator → Checkout → Purchases → Approved) with drop-off % at each stage, conversion/approval rate by device, campaign type, and affiliate.
- Funnel drop-off is steepest between Impressions → Clicks (95% loss), suggesting top-of-funnel ad targeting or creative is the biggest lever for scale.
- Approval rate is consistently ~93% across devices, affiliates, and campaign types — indicating conversion quality is stable regardless of acquisition channel, and the conversion rate (13.4%) rather than approval quality is the real optimization target.
- Revenue is fairly evenly distributed across affiliate types (Community, Influencer, Review Site, Cashback, Media, Comparison — all within a 44-47M band), so no single channel dominates; diversification looks intentional rather than risky concentration.
- Top 5 countries (Germany, UAE, Australia, Canada, Singapore) each contribute similarly (22-24M), suggesting geographic performance is well-balanced rather than dependent on one market.
- Power BI (Power Query, DAX measures, data modeling)
- Python (pandas, NumPy — data cleaning/preprocessing)
- Data cleaning: null handling in browser/affiliate rating fields, calculated columns for date hierarchies
Tangem__Market_analytics.pbix— full Power BI file- Screenshots of all 3 pages (above)