Overview

Professor Ferreira's research primarily focuses on how retailers can use algorithms to make better revenue management decisions, including pricing, product display, and assortment planning. In the retail industry, anticipating consumer demand is arguably one of the biggest challenges for a retailer's decision making. Errors in demand predictions typically lead to poor revenue management decisions, e.g., prices being set too high or too low, leading to lost profit from either excess inventory or stock-outs. Over the last decade, with the growing amount of data retailers capture, retailers are now considering how they can use these data to help predict demand and improve revenue management decisions. Professor Ferreira develops new machine learning and optimization algorithms to predict consumer demand and improve decision making, with theoretical, numerical, and/or experimental evidence that each algorithm performs well in practice. Taken together, her work demonstrates how retailers can achieve great benefits from using data and algorithms to help them make key revenue management decisions. In addition to her algorithmic work, Professor Ferreira also conducts research that develops and analyzes mathematical models that challenge the common belief held by managers at many e-commerce companies that they should make it as easy as possible for customers to search their catalog of products and make purchases. Her work has led to two key insights. First, online retailers may benefit from making it more difficult for customers to find their most preferred products. Second, limiting purchases in an online marketplace can improve system-wide customer experience and engagement. These findings offer insights for both retailers engaging in e-commerce as well as for digital platforms providing online marketplaces. By engaging with many managers across different industries, Professor Ferreira has been exposed to how they are thinking about using data and algorithms to drive decision making. Many managers recognize that algorithms are able to process significantly more data than humans, uncovering important relationships that drive predictions and making recommendations that would be difficult for humans to identify. That said, many managers also understand that data and algorithms cannot capture all of the nuances of a decision-making problem, and that employees often have additional information or contextual knowledge that they should consider in tandem with algorithmic recommendations when making decisions. In Professor Ferreira's newest line of research, she focuses on a critical challenge faced by many managers: employees equipped with algorithmic recommendations to aid their decision making often make errors by either discounting or adhering to algorithmic recommendations when they should not do so. This is a big problem: if users do not effectively incorporate data and algorithms in their decision making, the organization will not realize the algorithms' full potential or limitations. Professor Ferreira seeks to understand the root causes underlying this poor use of algorithmic recommendations and provide advice to managers as to how the potential of those algorithms might be better realized.