ParthBadiger

Identified that Outerwear & Sweaters yielded the highest profit margin ($1.8M+) but suffered an 11.2% return rate, while Organic/Search acquisition channels produced 2.4x higher CLV compared to Social campaigns.

Identified that delivery speed and geographic proximity to logistics hubs were the strongest predictors of customer satisfaction — suggesting marketplace expansion should prioritise regions with existing fulfilment infrastructure.

Demonstrated that transfer learning with domain-specific fine-tuning can replace expensive manual sentiment labelling, reducing the cost of customer feedback analysis significantly.

Validated on an 80/20 train-test split with R² of 0.97 on held-out data — the strong fit is driven by low-noise pricing signals (mileage, brand, age) in a well-structured marketplace dataset, not overfitting.

Optimal allocation yielded £4.85M — a meaningful improvement over the existing ad hoc pricing policy — with the solution remaining robust under ±10% demand fluctuations, giving management confidence to implement.

Identified a high-value customer segment that was under-targeted in current campaigns, providing a clear rationale for reallocating marketing spend toward the most profitable audience.

Surfaced underperforming units that appeared efficient by traditional metrics but were exposed by DEA's multi-input/output framework — enabling targeted process improvement rather than blanket cost-cutting.

Revealed that mid-tier wearable brands were gaining market share faster than premium players in emerging markets — a trend not visible in aggregate industry reports but clear through drill-down regional analysis.

Enabled comparative country-level analysis that surfaced diverging development trajectories between nations with similar starting conditions — highlighting the outsized impact of health and education investment on long-term GDP growth.