‘Deciding how to decide’ by Hugh Courtney, Dan Lovallo and Carmina Clarke, Harvard Business Review, November 2013
This is the first of two interesting articles on strategic decision making in this edition of Harvard Business Review. This first article has strong connections with McKinsey; Courtney worked for them before becoming a US business school professor and Lovallo advises McKinsey while also working as a professor of strategy at Sydney University.
The common underlying theme of both articles is that there are different types of strategic decision but managers continue to use the same approaches and techniques, without appreciating the fundamental differences. This first article suggests that the key elements of difference are whether you understand the ‘causal model’ – in simple terms the factors that will deliver success – and whether you can predict outcomes with reasonable certainty. If both these criteria are met, conventional techniques like discounted cash flow are suitable to evaluate the decision; if not, other approaches are required.
The article describes a number of potential situations, based on knowledge of these two variables. Where neither of these two criteria are not met, other tools are required; for instance where the causal model is well understood but predictions are difficult – maybe because competitive response is uncertain – complex modelling tools like Monte Carlo and Real Options analysis are required. These techniques enable the simulation of a number of possible outcomes and the use of statistical analysis to support the decision.
This led me to have my first reservations about the approach because our team at MTP has seen few day to day applications of these tools in our work with some of the world’s top companies; this could be because the high calibre managers of these companies are all falling into the trap suggested by the article but I think this unlikely. A more likely explanation is that, apart from one-off major complex projects, they have found these advanced tools to be less practical than conventional evaluation techniques and have instead adapted the latter by applying more flexible, dynamic analysis, in particular the asking of ‘what if’ questions. But despite these reservations I read on with as open a mind as possible.
The first suggestion is the application of Qualitative Scenario Analysis, in situations where the model is well understood but there is still a high level of uncertainty. This seems to be no more than looking at a number of likely outcomes and their consequences; hardly a great step forward. The most challenging situation is clearly where the causal model is not understood and where accurate predictions are not possible. An example of this would be entering a new, developing market – either organically or by acquisition – and fighting against competitors whose actions are impossible to predict. One answer might be to avoid such ‘opportunities’ because the risk is too high but instead the authors recommend what they describe as an underutilised tool – ‘Case Based Decision Analysis’. This approach involves the collection of information about analogous situations from past experience in other sectors and applying these to support the decision.
I would have liked to see more details of this approach as it clearly has merit in principle; however it assumes that there are analogous situations and that these really can deliver lessons to improve decision making. There have been many past criticisms of senior executives failing because they are trying to fight previous battles when situations have changed in subtle ways and there could be similar dangers with this approach. Are two complex business situations ever really alike? Do lessons from the past always have implications for the future? A number of business school cases that we have covered on our courses would suggest otherwise.
The article moves on to make some interesting points under the heading of ‘Complicating Factors’. One is the perennial problem of bias, which impacts any tool that tries to predict the future. The authors suggest that most decision makers are over-confident in their assessment of their own ability to predict; thus their decision tree to determine which tool is used will be distorted. Their answer is for the decision making to be open and transparent, with judgments about the level of predictability and the use of evaluation tools to be subject to challenge by peers. The authors admit that this may require a culture change in many organisations; how many CEOs would admit to their team that they do not understand the business model and cannot predict the outcome?
The other key point made at the end is that few decisions require only one evaluation tool; the best evaluations require a combination. An example would be discounted cash flow evaluation combined with sensitivity, range and probability analysis, maybe also a decision tree for multiple options. In our experience this is what most top companies are already doing for complex decisions and I see dangers in the suggestion that decisions can be pigeon-holed into particular situations and the use of specific tools. The best decisions are often the result of the use of multiple evaluations and tools which stimulate challenge and debate, leading to informed and balanced judgments.