Maximising profit through modelling customer behaviour based on Generalised Unavailability and Generalised Nines

Abstract

This paper discusses two related concepts of “Generalised Unavailability” and “Generalised Nines” as the foundation of a versatile framework for modelling customer satisfaction. Those concepts were introduced in a previous publication. A service is modelled as a network of maintained components. The annual cost of providing a service is modelled as a simple function of the reliability and restoration time of every component. The performance of a service is modelled as a function of network architecture, and the reliability and restoration time of each component. Hence, the causal chain from cost to service performance to customer satisfaction to anticipated revenue is modelled, and a method for maximising the profit of providing the service is developed. Necessary conditions on the reliability and restoration time of each component are derived. Simple analytic results are derived for special cases. A simple cloud computing service is studied. A small pilot study indicates that customers give greater weighting to restoration time than to reliability. Hence the design method is used to optimise profit according to the sampled customer preferences. The optimal reliability and restoration times for all components are calculated to yield maximum profit. In another example, an alternative model is used, with the demand curve parameterised by quality and hence customer satisfaction. This model allows the prediction of the number of customers who will buy a service of given quality at a given price. Illustrative results are derived, demonstrating a choice between discrete alternatives, and also showing the application of the framework to cases where concurrent outages are considered more problematic than independent outages.

Keywords:

• Reliability
• restoration
• resilience
• availability
• customer preferences
• optimisation
• profit
• cloud computing

Acknowledgements

Thanks go to Roland Padilla, PhD candidate at the University of Melbourne, for his insights and advice on cloud computing services and their customers. Thanks also to the reviewers for their observations and suggestions.