This case study has been condensed and referenced from a published paper detailing the process in great depth. Find the original article here for those with the necessary membership for viewing, or a downloadable draft here.
Ensuring Accuracy
The model prioritized real-world constraints for optimal decision-making. Firstly, total shipments were limited to the available inventory in the warehouse to avoid stockouts. Secondly, the model considered the established relationship between inventory levels and sales at each store for accurate forecasting.
Implementation & Deployment
Zara and researchers from UCLA and MIT collaborated to develop and integrate the AMPL model. AMPL’s solver performs separate optimizations for each item, resulting in 15,000 optimizations weekly. The two-year deployment involved rigorous testing and culminated in a user-friendly application for the warehouse allocation team, empowering them to explore different allocation strategies.
INDUSTRY
Fast Fashion & Retail
WORKFLOWS OPTIMIZED BY AMPL
Inventory Optimization & Distribution Planning
INTEGRATIONS
AMPL-based Optimization Model
KEYWORDS
AMPL
Supply chain optimization
Inventory allocation
Demand forecasting
Piecewise-linear approximation
The high-paced world of fast fashion thrives on optimizing a complex web of factors. Zara, a prominent player in this industry, exemplifies this challenge. Their success hinges on their ability to solve a critical optimization problem: inventory allocation across a vast network of stores.
Here, the core challenge lies in balancing several competing objectives:
Maintaining sufficient stock to meet customer demand for trendy styles, while avoiding overstocking and the associated risk of dead stock or markdowns. This requires accurate forecasting of demand for various styles and sizes across different store locations.
Optimizing the allocation of limited warehouse space to accommodate a diverse range of clothing items in various sizes. This involves maximizing space utilization while ensuring efficient picking and packing processes.
Balancing the need for timely delivery to stores with minimizing shipment costs. Optimizing shipping routes and consolidating shipments where possible are key factors.
Fast fashion thrives on the ability to react quickly to evolving trends. The allocation model needs to be adaptable to incorporate real-time sales data and adjust stock levels accordingly.
Zara, a global fashion powerhouse renowned for its trendy designs and rapid response to market trends, thrives on a meticulously optimized supply chain. Managing a vast network of over 1,500 stores worldwide requires a constant flow of the right clothing items in the right sizes to meet ever-evolving customer demands. This case study explores how Zara, in collaboration with researchers from UCLA and MIT, leveraged AMPL, a powerful optimization software tool, to transform their inventory allocation and distribution processes.
Traditionally, Zara’s inventory allocation relied on a manual approach:
Store managers and the warehouse allocation team juggled a massive amount of data, including assortment decisions, store inventory, past sales data, requested shipment quantities, and warehouse inventory levels.
These manual processes lacked the ability to comprehensively analyze and optimize shipments across their entire network.
Decisions needed to be made quickly based on sales data received just hours before shipment.
To address these challenges and achieve optimal inventory allocation, Zara implemented a data-driven approach powered by AMPL:
A new forecasting model incorporated into the system analyzes assortment decisions, past sales data, and requested shipment quantities.
AMPL's optimization engine acts as the core, handling demand forecasts, warehouse inventory, store inventory data, and past shipment information.
The AMPL model is built on a foundation of detailed data sets:
Sizes (including both a comprehensive set and a frequently used subset) and individual stores are clearly defined.
Real-time data on pre-shipment warehouse inventory, current store inventory for each size, and the value per item remaining in the warehouse are factored in.
Selling prices at each store and forecasted sales for each size, considering the relationship between inventory levels and sales through a piecewise-linear approximation, are integrated.
AMPL’s optimization engine focuses on two key decision variables:
The model determines the optimal number of each size to be shipped to each store to maximize overall benefit.
The model also forecasts expected sales for each size at each store.
The core objective of the model is to maximize two key factors:
Optimizing shipments to meet customer demand and drive sales growth.
Ensuring valuable merchandise remains strategically allocated within the warehouse.
The model incorporates essential constraints to ensure feasibility and accuracy:
Total shipments of each size cannot exceed the available inventory in the warehouse.
Sales and inventory levels at each store must adhere to the established piecewise-linear approximation.
The implementation of this innovative approach involved a collaborative effort:
Researchers from UCLA and MIT partnered closely with Zara staff to develop and integrate the model.
AMPL's mixed-integer linear solver performs separate optimizations for each item sold, resulting in approximately 15,000 optimizations every week.
The two-year implementation process involved rigorous testing and collaboration with Zara's corporate technology team. The final solution utilizes a user-friendly client application for the 60-person warehouse allocation team, allowing them to experiment with the value of warehouse items and investigate different allocation scenarios.
Zara’s success story with AMPL serves as a testament to the power of data-driven optimization in the fast-paced fashion industry. By leveraging AMPL, Zara has achieved:
Optimized shipments ensure the right products are delivered to the right stores at the right time.
Data-driven decisions lead to maximized sales opportunities and customer satisfaction.
The automated model streamlines processes and frees up valuable staff time for strategic tasks.
Inspired by Zara’s success? AMPL can transform your fashion supply chain by optimizing inventory allocation and distribution. Whether you’re facing challenges with managing stock levels, maximizing sales across your stores, or streamlining allocation processes, AMPL’s powerful capabilities can deliver significant results. Book a free demo today to see AMPL in action and discover how it can help you optimize your fashion supply chain, reduce costs, and achieve greater customer satisfaction. Alternatively, start your free trial to experience the power of AMPL firsthand!
As Zara continues to expand its global reach, AMPL’s optimization capabilities will undoubtedly play a vital role in maintaining their position as a leader in the ever-evolving fashion landscape.