The term decision-support is synonymous with Business Intelligence software applications. With these, managers are supposedly better able to make the right decisions to guide their organisation, but is this really true? What should users expect of a decision-support system? In this article, Michael Coveney, author of Strategy to the Max, gives his views on what to look for, based on a life-time in designing and implementing decision-support systems.
History of Decision Support
Advice on how to make the ‘right’ decisions has been with us for over 2,500 years, arguably starting with Confucius who said that decisions should be informed by benevolence, ritual, reciprocity, and filial piety. Things became more scientific in the 17th century when Blaise Pascal and Pierre de Fermat developed the concept of calculating probabilities for chance events. This was followed by Carl Gauss in the 18th century who developed a structure for understanding the occurrence of random events, which could be used in making investment decisions. But it wasn’t until the 1960s that the study of rational business decision-making and decision support gained formal recognition.
Ronald Howard, a Stanford University professor, who was instrumental in developing practices and tools in support of organisational decision-making, coined the term ‘decision analysis’ in 1964. His methods recognized and included uncertainties with a structured and rationally justifiable method for making the ‘right’ decision.
Decision analysis advocates making the decision whose consequences have the “maximum expected utility” or that “maximizes the probability of achieving the uncertain aspiration level”, all of which is the fundamental goal of business planning. In the 1965 landmark book ‘The Rational Manager: A Systematic Approach to Problem Solving and Decision-Making’, the decision-making process is defined this way:
- Establish objectives.
- Classify objectives and place in order of importance.
- Develop alternative options
- Evaluate alternative options against all the objectives.
- Make a tentative decision based on the option most likely to achieve all the objectives.
- Evaluate the chosen option against potential consequences.
- Implement the chosen option and additional actions required to prevent any adverse consequences from becoming problems.
The first computer-based systems to support managerial decision-making were introduced in the late 1960s. They allowed managers to assess potential outcomes based on quantifying individual actions and to analyse numeric data such as costs and sales that resulted from past decisions.
These systems initially came under the umbrella term of ‘management information’, but as new ways of presenting data became available that also allowed users to investigate to produce their own analyses, the term business intelligence (BI) took over.
The term ‘BI’ was defined by Howard Dresner in 1989 as the "concepts and methods to improve business decision making by using fact-based support systems”. This was later refined in a Gartner Research document as: 'The systems that help decision-makers throughout the organization understand the state of their company’s world. A set of methods that support sophisticated analytical decision-making aimed at improving business performance.'
BI technology systems have been around commercially since the early 1970s. They typically comprise of a multidimensional database (sometimes referred to as an OLAP—On-Line Analytical Processing—database), routines to load data from a transaction system into the database, and reports/analyses that display variances and trends. But is this enough, or is something missing to truly support organisational decision-making?
Requirements of a decision-support system
In looking at the decision-making process defined by Ron Howard, systems need to be able to do three main tasks:
1. Must be able to present data in context
Anyone who has read my other articles will know I go on about this a lot. There’s no point in presenting a set of numbers unless they are reported in the context of the business environment and the activities that took place that led to those numbers being produced. Without this context it is easy to be misled that an over target performance is good, when the assumptions made about the business environment were wrong when the target was set.
Similarly, if the activities required to achieve a target were not properly resourced or completed on time, then any under performance may not be due to a ‘wrong decision’. It could be simply that the decision was never implemented to begin with.
2. Must allow managers to assess alternative choices
I find many systems fail on this point. In order to properly assess a range of alternatives, the different activities involved must be combined in various combinations, with the forecast outcomes being available for side-by-side comparison. Alternatives are not simply a set of numbers but could also include changes to the organisation structure and the setting up of new initiatives. This will require the software to handle multiple organisational structures at the same time so comparisons can be made.
As well as the above, a true support system should allow initiatives to be moved back and forward in time, so that the effect of delaying a project or bringing one forward can be seen.
3. Must support the implementation and monitoring of decisions
Good decisions are useless if they can’t be implemented. Research tells us that between 50 to 60% of the potential within an operational plan is never realised due to failures in the planning and execution process. Therefore a decision-support system should support the allocation of resources to a chosen course of action, and then go on to monitor its implementation and success on corporate goals.
In looking at the above requirements, it would seem that very few systems are fully able to meet the needs of decision support. There are good systems for analysis, for budgeting and for reporting but to combine these around the requirements of making and tracking decisions requires a degree of integration and data manipulation we’ve not yet seen on the market.
Some of the reason for this has to do with the way in which systems have been developed in the past, which tend to be an incremental improvement on what they have always done. As a result budgeting systems tend to remain budgeting systems while KPI reporting tools remain as reporting tools. Sure, there may be some crossover functionality, but at their core they remain focused on one aspect of decision-support. If this were not so then software vendors would only need one solution for all client needs.
What is required is a complete rethink about the role of a decision-support system and to then develop one, from the ground up, for that role. It should be a complete system – not just supporting one part of the process. In the same way that smartphones today are more than just an incremental development of the phone, so the same needs to happen in decision support.
Buy Michael Coveney’s bestselling book “Strategy to the MAX”
Read Michael Coveney’s other related articles:
Strategy to the MAX - choosing the right measures
Performance management vs. performance measurement
