For many fashion retailers, prices of new products are typically based on some combination of merchants’ intuition, competitors' pricing, and percentage markup on cost; given the difficulty of forecasting demand for fashion products, though, these approaches often lead to either stock-outs early in the season or large amounts of unsold inventory at the end of the season. We worked in partnership with the online flash sales retailer Rue La La to explore a data-driven approach to pricing new fashion products, developing machine learning and optimization algorithms that use Rue La La’s abundance of data to better predict demand and set prices. We conducted a five-month field experiment to test the impact of our algorithms’ price recommendations; results show an increase in revenue by approximately 10% with minimal impact on cost.

Kris Johnson Ferreira
Assistant Professor of Business Administration
Assistant Professor of Business Administration
Online retailers have several advantages over traditional brick-and-mortar retailers; for example, online retailers have the ability to view real-time consumer purchase decisions, and they can typically change prices frequently at negligible cost. Given these capabilities, we develop a novel machine learning algorithm that changes product prices over the course of a selling season in order to maximize total revenue.
Since the retailer does not know consumer purchase probabilities at different prices, they face a tradeoff commonly referred to as the exploration-exploitation tradeoff. Towards the beginning of the selling season, the retailer may offer several different prices to try to learn the purchase probability at each price (“exploration” objective). Over time, the retailer can use estimates of these purchase probabilities to set a price that maximizes revenue throughout the remainder of the selling season (“exploitation” objective). To further complicate matters, the retailer is constrained by limited inventory and thus pursuing the exploration objective comes at the cost of diminishing valuable inventory. Our algorithm addresses these challenges and optimally sets prices to balance exploration and exploitation in order to maximize revenue.
Kris Ferreira is an assistant professor of business administration in the Technology and Operations Management (TOM) Unit. She teaches the Supply Chain Management course in the MBA elective curriculum, the Operations & Supply Chain Management course in the Harvard Business Analytics Program, and in various Executive Education programs. She has previously taught the core TOM course in the first year MBA required curriculum.
In her research, Professor Ferreira focuses on helping e-commerce companies make more profitable revenue-management decisions, including pricing, assortment planning, ranking, etc. For much of this work, she partners with online retailers to design new machine learning and optimization algorithms that use the massive amount of data at their fingertips to improve their decision making in this domain. Most recently, Professor Ferreira has started studying the characteristics that are critical in the success or failure of an analytics tool’s development and implementation. Her work has been awarded first place for the Revenue Management and Pricing Section Practice Award, and two-time finalist for the M&SOM Best Paper Award and Innovative Applications in Analytics Award.
Professor Ferreira earned her PhD in operations research at the Massachusetts Institute of Technology and her BS in industrial and systems engineering at the Georgia Institute of Technology. Before entering graduate school, she was a supply chain consultant for Alvarez & Marsal and a project manager for UPS Supply Chain Solutions.
- Featured Work
-
For many fashion retailers, prices of new products are typically based on some combination of merchants’ intuition, competitors' pricing, and percentage markup on cost; given the difficulty of forecasting demand for fashion products, though, these approaches often lead to either stock-outs early in the season or large amounts of unsold inventory at the end of the season. We worked in partnership with the online flash sales retailer Rue La La to explore a data-driven approach to pricing new fashion products, developing machine learning and optimization algorithms that use Rue La La’s abundance of data to better predict demand and set prices. We conducted a five-month field experiment to test the impact of our algorithms’ price recommendations; results show an increase in revenue by approximately 10% with minimal impact on cost.
Online retailers have several advantages over traditional brick-and-mortar retailers; for example, online retailers have the ability to view real-time consumer purchase decisions, and they can typically change prices frequently at negligible cost. Given these capabilities, we develop a novel machine learning algorithm that changes product prices over the course of a selling season in order to maximize total revenue.
Since the retailer does not know consumer purchase probabilities at different prices, they face a tradeoff commonly referred to as the exploration-exploitation tradeoff. Towards the beginning of the selling season, the retailer may offer several different prices to try to learn the purchase probability at each price (“exploration” objective). Over time, the retailer can use estimates of these purchase probabilities to set a price that maximizes revenue throughout the remainder of the selling season (“exploitation” objective). To further complicate matters, the retailer is constrained by limited inventory and thus pursuing the exploration objective comes at the cost of diminishing valuable inventory. Our algorithm addresses these challenges and optimally sets prices to balance exploration and exploitation in order to maximize revenue.
There has been a recent trend by both brick-and-mortar and online retailers to change their assortment of products offered to customers more and more frequently. Several factors likely go into this decision: limited shelf space, enabling trend detection, etc. We introduce and study another important factor – the value of concealment – that should play a role in this decision. Namely, by distributing its seasonal catalog of products over multiple assortments rotated throughout the season – as opposed to selling all products in a single, fixed assortment – a retailer effectively conceals a portion of its full product catalog from consumers, injecting uncertainty into the consumer’s relative product valuations. Consumers may respond by purchasing more products, thereby generating additional sales for the retailer: the value of concealment.Brick-and-mortar retailers in shopping malls are well aware of the impact of a snazzy window display case, enticing customers walking through the mall to enter the store. Online retailers can rank (sort) their products to have a similar effect: when customers see the first several products in the ranking, they decide whether or not to continue browsing. We design online learning algorithms for product ranking that maximize the number of customers who browse/engage with the site. We develop a class of online learning-then-earning algorithms that prescribe a ranking to offer each customer, learning from preceding customers’ clickstream data to offer better rankings to subsequent customers. Our algorithms balance product popularity with diversity: the notion of appealing to a large variety of heterogeneous customers. We partner with Wayfair and find that our algorithm yields a significant increase (5-30%) in the number of customers that engage with the site. - Journal Articles
-
- Ferreira, Kris J., and Joel Goh. "Assortment Rotation and the Value of Concealment." Management Science (forthcoming). View Details
- Ngwe, Donald, Kris J. Ferreira, and Thales Teixeira. "The Impact of Increasing Search Frictions on Online Shopping Behavior: Evidence from a Field Experiment." Journal of Marketing Research (JMR) 56, no. 6 (December 2019): 944–959. View Details
- Ferreira, Kris J., David Simchi-Levi, and He Wang. "Online Network Revenue Management Using Thompson Sampling." Operations Research 66, no. 6 (November–December 2018): 1586–1602. View Details
- Ferreira, Kris J., Bin Hong Alex Lee, and David Simchi-Levi. "Analytics for an Online Retailer: Demand Forecasting and Price Optimization." Manufacturing & Service Operations Management 18, no. 1 (Winter 2016): 69–88. View Details
- Johnson, Kris, David Simchi-Levi, and Peng Sun. "Analyzing Scrip Systems." Operations Research 62, no. 3 (May–June 2014): 524–534. View Details
- Working Papers
-
- Ferreira, Kris, Sunanda Parthasarathy, and Shreyas Sekar. "Learning to Rank an Assortment of Products." Working Paper, July 2020. View Details
- Ferreira, Kris, and Emily Mower. "Demand Learning and Dynamic Pricing for Varying Assortments." Working Paper, May 2019. View Details
- Aouad, Ali, Adam Elmachtoub, Kris Ferreira, and Ryan McNellis. "Market Segmentation Trees." Working Paper, January 2020. View Details
- Ferreira, Kris J., Joel Goh, and Ehsan Valavi. "Intermediation in the Supply of Agricultural Products in Developing Economies." Harvard Business School Working Paper, No. 18-033, October 2017. View Details
- Cases and Teaching Materials
-
- Toffel, Michael W., Natalie Epstein, Kris Ferreira, and Yael Grushka-Cockayne. "Assessing Prediction Accuracy of Machine Learning Models." Harvard Business School Technical Note 621-045, August 2020. (Revised September 2020.) View Details
- Ferreira, Kris, and Srikanth Jagabathula. "JOANN: Joannalytics Inventory Allocation Tool." Harvard Business School Case 621-055, September 2020. View Details
- Ferreira, Kris, Joel Goh, Dawn Lau, and Tuan Phan. "GHN and AhaMove: Last-Mile Delivery in Vietnam." Harvard Business School Case 619-051, June 2019. (Revised December 2020.) View Details
- Ferreira, Kris, and Karim R. Lakhani. "Flashion: Art vs. Science in Fashion Retailing." Harvard Business School Case 617-059, March 2017. (Revised October 2017.) View Details
- Research Summary
-
In her research, Professor Ferreira focuses on helping e-commerce companies make more profitable revenue-management decisions. Although the overarching goal of revenue management is relatively well-defined – typically to maximize either revenue or profit – companies have numerous operational and marketing “levers” to do so, such as pricing, assortment planning, and targeted advertising. Many of these decisions are made over time in tandem with other decisions, such as inventory management. The key challenge in optimizing these decisions is that consumer-level demand, a function of these decisions, is often unknown and hard to predict; thus, accounting for each decision’s impact on demand is difficult. Over the last two decades, the domain of revenue management has become prominent in applied operations research. In practice, the airline and hospitality industries were early adopters of revenue management; more recently, retail and other industries have started to adopt revenue-management techniques. Professor Ferreira’s research extends revenue-management theory and algorithms to the rapidly growing e-commerce industry, including traditional firms that are now expanding online. In particular, she exploits three key aspects of e-commerce – compared to the offline world –to improve revenue-management results. First, e-commerce companies typically collect a wider variety of data compared to their offline counterparts. For example, they may collect clickstream data that describes how each customer interacts with the products on their website, such as which products they view, for how long, and in what sequence. Second, e-commerce companies typically face lower costs to change many revenue-management decisions, including pricing and assortment decisions. For example, changing a product’s price online takes significantly less time and cost than doing so in (or across) physical stores. Professor Ferreira develops machine learning and optimization algorithms to use data and exploit cost-savings in order to help e-commerce companies make more profitable revenue-management decisions. Her work includes developing algorithms for dynamic pricing, dynamic product ranking, and personalized advertising, much of which is conducted in collaboration with e-commerce companies. A third key aspect of e-commerce is the increase, over the past decade, in new business models and marketplaces that are unique to e-commerce. Professor Ferreira looks beyond the data opportunity alone and identifies additional opportunities that operating in an online world permits. For example, firms in the online “flash sales” industry sell discounted apparel and accessories for extremely short time periods, typically 48 hours. Such a business model in which the entire assortment of products offered changes daily is too costly to exist in the offline world due primarily to high transportation and stocking costs. Professor Ferreira studies how firms can take advantage of new e-commerce business models when making revenue-management decisions.
- Teaching
-
This course explores topics such as supply chain design, demand forecasting, inventory management, distribution economics, and retailing operations. The focus is on managing material and information flows across both functional and organizational boundaries. Cases in the course illustrate that barriers to integrating supply chains often relate to behavioral and managerial issues (e.g., misaligned incentives or change-management challenges) and operational execution problems. The course makes it clear that suitable information technology and the effective application of analytical tools are a necessary, although not sufficient, requirement for effective supply chain management.
This course enables students to develop the skills and concepts needed to ensure the ongoing contribution of a firm's operations to its competitive position. It helps them to understand the complex processes underlying the development and manufacturing of products as well as the creation and delivery of services.
Topics encompass:
- Process analysis
- Cross-functional and cross-firm integration
- Product development
- Information technology
- Technology and operations strategy
- Awards & Honors
-
Finalist for the 2018 and 2019 M&SOM Best Paper Award from the Manufacturing & Service Operations Management (MSOM) Society with Bin Hong Alex Lee and David Simchi-Levi for "Analytics for an Online Retailer: Demand Forecasting and Price Optimization."Finalist for the 2019 Institute for Operations Research and the Management Sciences (INFORMS) Revenue Management and Pricing Section Practice Award with Matt Capizzi, Arthur Hong, and Emily Mower, for "Demand Learning and Dynamic Pricing for Varying Assortments: Algorithm Development and Implementation at Zenrez."Second Place Winner of the 2015 Innovative Applications in Analytics Award from the Institute for Operations Research and the Management Sciences (INFORMS) with Bin Hong Alex Lee, David Simchi-Levi, Murali Narayanaswamy, Philip Roizin, and Jonathan Waggoner for "Analytics for an Online Retailer: Demand Forecasting and Price Optimization."Finalist for the 2015 IBM Service Science Best Student Paper Award from the Institute for Operations Research and the Management Sciences (INFORMS) for "Online Network Revenue Management Using Thompson Sampling" (HBS Working Paper 16-031, September 2015) with He Wang, David Simchi-Levi.Winner of the 2014 Institute for Operations Research and the Management Sciences (INFORMS) Revenue Management and Pricing Section Practice Award with Murali Narayanaswamy, Philip Roizin, Jonathan Waggoner, Bin Hong Alex Lee, and David Simchi-Levi for "Analytics for an Online Retailer: Demand Forecasting and Price Optimization."Recipient of a 2013 Graduate Student Award for Excellence in Engineering Systems Teaching Massachusetts Institute of Technology (MIT).
- Additional Information
- Areas of Interest
- In The News
-