Situation
A leading home improvement retailer faced challenges managing inventory across 30 million store-SKU combinations. Flawed demand planning and an outdated inventory system led to:
- Rising lost sales due to stockouts.
- Escalating holding costs from excess inventory.
- Legacy systems based on unsuitable assumptions compounded inefficiencies, especially for slow-moving items tailored to diverse customer needs.
Challenge
Key inefficiencies emerged from the retailer’s complex demand patterns:
- Seasonality Variations: Product categories showed significant demand shifts by season.
- Customer Segments: Distinct behaviors between regular shoppers and bulk-buying contractors influenced demand.
- Supplier Variability: Erratic lead times disrupted stock consistency.
- Temporal Demand Changes: Weekday and weekend shopping patterns further complicated planning.
Approach
Mu Sigma developed a three-pronged strategy to enhance the seller experience:
- Defining Bad Selling Experiences (BSE)
- Analyzed historical buyer-seller interactions to construct a global BSE metric applicable across diverse markets.
- Monitored the metric weekly to evaluate the impact of platform-wide initiatives.
- Proactive Abusive Buyer Mitigation
- Designed a heuristic-based rule engine to detect and automatically suspend abusive buyers, ensuring a safe environment.
- Seller Trust Score & Refund Campaign
- Introduced a trust score metric, leveraging sellers’ transaction history.
- Rolled out a refund campaign for high-trust sellers, reducing friction and enhancing loyalty.
Impact
- $2M increase in revenue for small businesses, boosting seller success.
- 2,000 abusive buyers automatically suspended, fostering a safer platform.
- 10% reduction in seller detractors, improving seller satisfaction.
Business Impact
-
$2M
increase in revenue
-
10%
reduction in seller detractors
Let’s move from data to decisions together. Talk to us.

The firm's name is derived from the statistical terms "Mu" and "Sigma," which symbolize a
probability distribution's mean and standard deviation, respectively.