|
|
 Business Processes
Decisions made in business are often complex and require an understanding of tradeoffs and likely outcomes from alternative decisions. We help businesses chart their activities, learn about the likelihood of successes and failures, and conduct cost benefit analyses to support decision processes.
In many instances, we find that there’s a cycle of decisions. The first step impacts the second, which in turn impacts the third. The cascading effects of decision making lead to a need for business systems that help make decisions as part of an overall process, where current outcomes are the best predictors of what needs to come next, and where there isn’t a lot of time to ponder what to do. We help by establishing business rules to optimize outcomes and systems that learn as they progress, so that decision-making improves over time.
|
Case Examples:
For a major hotel chain, we were asked to develop a forecasting model that would allow them, for each property, for each night, predict 10-30 days out whether that property for a specific night would be sold out. If it was going to be sold out, was it then safe to turn off some or all of the incentives offered to allow the property (and thus the chain) to maximize its room revenue. This is a much more difficult problem than one would expect because the forecast is subject to a wide variety of factors, it covered hundreds of hotels for 365 nights each, and so there were literally thousands of forecasts being made every single day in the system we developed.
Making the problem more complex is the problem that the forecasts themselves are intercorrelated, and a truly efficient system of forecasting would also capitalize interrelationships to improve the quality of the forecasts. Another complication was that there were multiple incentive programs offered and so turning each one on or off had different impacts on the outcomes. With ten incentive programs, there are 1,024 different forecasts that can be made with turning options on and off, and each incentive program had a different discount, so cost-benefit analyses for just over 500 properties, 365 nights, 1,024 forecasts resulted in over 500 thousand forecasts for a single night and over 200 million forecasts in a year. The system also needed to be able to incorporate information from each property manager and external economic data. Such a system not only had to be accurate, it had to be efficient so that it didn’t get so weighted down in detail as to become incomprehensible to the hotel chain and its many users. We developed a system that interacted with their reservation system and also their accounting and reporting systems for forecasting and monitoring purposes.
|
|
