Background
MyRunway is a South African discount e-commerce retailer dedicated to making fashion accessible through daily deals from over 500 local and international brands, with frequent promotions of up to 70% off nationwide delivery/returns. With web and mobile apps, the platform offers a local, convenient experience (no customs) featuring quick delivery and easy returns. It is currently a South African value leader in fashion, offering free shipping on orders over R650. MyRunway has been one of South Africa’s few profitable e-commerce companies, enabling it to invest in data, BI, and cross-cloud analytics to drive growth.
Challenges
The client lacked in-house GCP and dbt expertise after a prior vendor disengaged, leaving no clear ownership or knowledge transfer. The costs associated with BigQuery have increased approximately 1.5 times year over year, with unused pipelines and legacy tables contributing to the escalation and obscuring the allocation of expenditures. Data problems caused incorrect core metrics on executive dashboards, such as CTR/CTSR, leading to a lack of stakeholder trust. The disorganized data models led to slow and costly large DBT tables.
Solution
Platform review & remediation
- Audited dbt models and BigQuery tables; refactored “large” models for efficiency.
- Traced dashboard data lineage end-to-end, fixed null handling, and metric formulas.
Cost & hygiene actions
- Investigated usage with BigQuery logs and Looker references; closed an unused data stream and deleted two obsolete benchmark tables across dbt, Terraform, and BigQuery to prevent auto-recreation.
- Compiled a watchlist of the largest, most expensive tables for phased optimization.
BI stabilization & net-new analytics
- Repaired Commercial Compass and Brand Insights Report dashboards in Looker Studio.
- Built a new “MyRunway Brand Insights” dashboard from scratch (designed in Figma, implemented in Looker Studio), consolidating multi-table inputs.Compiled a watchlist of the largest, most expensive tables for phased optimization.
Roadmap alignment
- Planned future setup to keep BigQuery as the warehouse while evaluating Amazon QuickSight for dashboarding/AI-assisted insights, with secure cross-cloud connectivity.
Features
Brand Insights:
- Customer geography (city-level map)
- Gender breakdown (GA)
- Department views (Women/Men/Kids/Unisex)
- Device mix (iOS/Android/Tablet/Web Desktop/Web Mobile)
- Category interest (e.g., shoes, tops, accessories)
- Brand analytics: orders by brand, orders with number of unique Brands, revenue contribution, bucket size (items per order)
- Product dynamics: fastest/slowest selling, most/least clicked
Stabilized legacy dashboards:
- Commercial Compass and Brand Insights metric corrections (CTR/CTSR formulas, averages, impressions), null handling, and brand/supplier drill-downs.
Data platform hygiene:
- dbt model refactors for performance
- Removal of legacy benchmark tables end-to-end (dbt, Terraform, BigQuery)
- Closure of an unused data stream after dependency verification (Logs, Looker refs)
Outcome
- Operational Cost Savings: 65% reduction in operational costs
- Future-proof Payment Ecosystem 3x faster time-to-market for new payment methods. 100% coverage of local payment methods across 15+ countries.
- Improved Performance & User Experience: 40% faster frontend performance across key regions.