“No machine can possible do as good a job as I can, with all my years of real-world experience at this property.” Sound familiar? Quite probably – but it’s highly unlikely, according to research by Michael Davern, Professor of Information Systems at the Stern School of Business, New York University.

Although it’s still being finalized, Professor Davern’s research indicates that humans are not genetically programmed for the intuitive understanding of statistical probabilities. Our natural gambling instincts like to believe in a more deterministic world that we can influence, rather than accepting the actual statistics of reported results; even though we can predict the overall behavior of a roulette wheel, we still like to believe that we can “know” where a particular ball is going to land . To quote his presentation earlier this year:

“The research suggests that the nature of human cognition (the psychology of human judgment and decision making) imposes limits on users' ability to realize the full value potential offered by revenue management systems. More specifically, my work has demonstrated that under certain conditions, users have difficulty appreciating the value and applicability of computer-based revenue management models, adversely influencing their reliance on the recommendations of revenue management systems.”

Professor Davern quotes an example of a major chain hotel that did not achieve the same RM benefits as its peers. Analysis showed that the GM could have produced an extra 6% revenue per day by following the systems recommendations, but she consistently overrode it for what had always seemed valid reasons in the past, such as current weather patterns or an airline strike. What she hadn’t taken into account was that similar events – not necessarily identical ones, but ones with similar impact - had occurred in the past, and the RM system had already taken their effect into account.

This is not to say that humans should never override an RM system. People may not have an inherent statistical ability, but they do have the intelligence to interpret the effects of a competitor’s hotel opening. The trick is to leave the number-crunching to the computer, then add your local knowledge to that for the major outside factors. But keep an eye on your history.

-----------------------

Source: Davern, M.J., "Perceptions, Performance, and Reliance on Technology: The Value of Information From Revenue Management Decision Support Systems". Presentation at the Research Seminar Series, Center for Research in Information Systems, Stern School of Business, New York University, February 11, 1998.

Back to Main Article