Photo credit: www.sciencedaily.com
New Forecasting Model Enhances Demand Estimation for Businesses
Washington State University researchers have unveiled an innovative forecasting model aimed at helping businesses more accurately gauge customer interest in products, addressing the challenge of missing key data.
Featured in the journal Production and Operations Management, the research presents a mathematical modeling technique that extends beyond conventional methods, allowing companies to assess customer interest by considering factors beyond completed transactions. This approach enhances demand understanding, operational efficiency, and decision-making capabilities.
Lead author Xinchang Wang, an assistant professor of operations management at WSU’s Carson College of Business, explained, “Many businesses can only perceive a fragment of demand; they are aware of who purchases but lack insight into how many potential customers considered buying and ultimately chose not to.” The model aims to fill in these gaps, giving companies a more holistic and dependable estimate of demand.
Industries such as travel, hospitality, retail, and e-commerce have faced ongoing challenges in accurately forecasting demand. Many have relied on sweeping assumptions, often estimating total market size based on their market shares. Wang argues that these conventional methods frequently fall short of capturing genuine customer behavior, which can result in imprecise sales forecasts and lost revenue opportunities.
Wang and co-author Weikun Xu, a PhD student in management science at Carson, devised a new method that not only estimates sales but also determines the overall number of customers contemplating a purchase. Their model, by thoroughly analyzing actual sales data, sheds light on the number of customers who might have turned away due to issues like pricing or timing.
To create their model, the researchers employed a technique known as the sequential minorization-maximization algorithm, which enhances forecast accuracy. Unlike traditional approaches, which may yield multiple potential demand estimates with ambiguity regarding the best one, this algorithm ensures a singular, most accurate prediction under specific data conditions. “Eliminating uncertainty enables businesses to make bolder pricing decisions,” Wang noted.
Since the model was specifically designed to function with incomplete data, its potential applications are broad and cross-industry.
Although the study utilized airline ticket sales data for testing, Wang believes the methodology can be applied across various sectors experiencing similar uncertainties in demand. For instance, hotels may leverage the model to better predict bookings when potential guests are browsing without finalizing reservations. Similarly, retailers and grocery stores could utilize the model to gauge total market demand, even as some shoppers prefer competitors. E-commerce platforms might also gain insights into shopping cart abandonment, allowing them to refine their sales strategies.
“This model offers a formidable tool for sectors grappling with incomplete data challenges,” Wang stated. “By enhancing demand forecasting, businesses can improve their planning processes, optimize operations, and ultimately increase their competitive edge.”
Source
www.sciencedaily.com