Weitere Kapitel dieses Buchs durch Wischen aufrufen
Executives tend to view new approaches to decision-making with a healthy dose of skepticism: why bet an enterprise’s future (and personal reputations) on the decision test drive method? When exposed to new analytic methods, executives and managers typically pose several questions. Sections 14.1 through 14.6 of this chapter address the following important concerns about the decision text drive method: Why should we trust this new method? What about “garbage in, garbage out” (GIGO)? What about unknown unknowns such as rare “black swan” events? What return on investment (ROI) does this new method produce? What collateral benefits does the new method provide? What guarantees does the new method offer?
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
Van Vliet [ 13].
Common sources include case studies from business schools, books, and papers from academic journals, trade and professional associations. For example, the Society for Human Resources Management publishes guidelines for downsizing strategies such as voluntary and forced layoffs and employee retirement incentive programs (cf. Introduction). Cascio [ 1].
Decision models can be instrumented with logic to flag such violations of logical constraints in a report, and block users from projecting decision scenarios until they resolve these problems in the input data.
Fully automated (AI) systems (e.g., expert systems, machine learning algorithms) have their own well-known problems relating to trust and GIGO. Lenat et al. [ 5], https://www.wired.com/story/greedy-brittle-opaque-and-shallow-the-downsides-to-deep-learning/.
The range would have a confidence factor as well, such as 95%. Hubbard [ 3].
As described in Chap. 8, Monte Carlo trial results can be plotted with cumulative (frequency) histograms. This allows estimation of a cumulative distribution function and of the probability of achieving outcomes below or at a specified (threshold) value for the desired output parameter.
http://archive.defense.gov/Transcripts/Transcript.aspx?TranscriptID=2636. Rumsfeld was responding to a question about the lack of evidence that Saddam Hussein’s regime in Iraq was actively working to supply weapons of mass destruction to terrorist groups such Al Qaeda.
Schoemaker [ 9].
Comparing projected outcomes across dynamic scenarios with the same assumptions essentially subtract out inaccuracies or errors in judgments, like cancelling out noise from a signal. (This assumes that the decision model does not contain dynamics that are highly non-linear).
A few intelligence analysts actually anticipated a major Al Qaeda attack on domestic soil, but they did not know the group’s exact targeting strategy (i.e., hijacking multiple aircraft) and targets (i.e., major symbolic targets including the World Trade Center and the Pentagon). https://www.newyorker.com/magazine/2002/01/14/the-counter-terrorist).
Our approach is a variant of regression testing, where you continually expand your suite of test cases that your program or method previously processed successfully with new tests. van Vliet [ 13]. It also exemplifies a strategy of continuous process improvement.
Silver [ 10] makes a similar point about forecasting, comparing meteorologists, who predict weather conditions over hours, days or weeks, to climatologists that forecast weather over years to decades.
The calculation can be refined to include a pro-rated share of cost to develop and validate a test drive model. These are one-time costs. ROI will obviously drop somewhat when including the full costs for a decision model built from scratch (if an existing decision model is unavailable). Pro-rating these start-up costs over multiple uses provides a standard basis for comparing ROI across different types of decisions.
Costs incurred to frame a critical decision, collect market intelligence and other inputs, and define decision options should not be included. These costs would be incurred even if the test drive method wasn’t used.
ROI would be much larger if the drug had higher base sales or if the risk of a flawed strategy was more substantial. Also this simple calculation assumes that the test drive reallocates rather than reduces the marketing mix budget, which can run tens to hundreds of millions of dollars a year for branded drugs.
Recall from Chap. 13 that the dominant cause of failure in executing transformational decisions is a failure to communicate and sustain a sense of urgency. Interactive decision models alleviate this issue.
Sharing scenarios requires careful security preparations to protect highly sensitive data.
Engagement is critical because these stakeholders have a limited amount of time—and attention—to conduct reviews. Think about whether you would rather read a report, view a slide deck or video, or interact with a dynamic model in real time (like SimCity) in a facilitated workshop. de Geus [ 2].
Similarly, Richard Thaler, the Nobel laureate behavioral economist, suggests that decision-makers “write stuff down”, including their decisions and why they thought they were a good idea at the time.” This practice counteracts hindsight bias, a contributor to CEO overconfidence. Javetski and Koller [ 4].
Oakland [ 6].
All URLs Accessed 07 Jul 2019.
Cascio, Wayne E. 2010. Employment Downsizing and Its Alternatives: Strategies for Long-Term Success. Alexandria, VA: Society for Human Resource Management Foundation.
de Geus, Arie. 1988. Planning as Learning. Harvard Business Review. 66(2): 70-74.
Hubbard, Douglas W. 2007. How to Measure Anything: Finding the Value of Intangibles in Business. Hoboken, NJ: John H Wiley and Sons.
Javetski, Bill, and Tim Koller. 2018. Debiasing the corporation: An interview with Nobel laureate Richard Thaler. McKinsey Quarterly. Available at https://www.mckinsey.com/business-functions/strategy-and-corporate-finance/our-insights/debiasing-the-corporation-an-interview-with-nobel-laureate-richard-thaler. Accessed 05 Jul 2019.
Lenat, Douglas B., Mayank Prakash, and Mary Shepherd. 1985. CYC: Using Common Sense Knowledge to Overcome Brittleness and Knowledge Acquisition Bottlenecks. AI Magazine. 6(4): 65-85. https://doi.org/10.1609/aimag.v6i4.510.
Oakland, John S. 1996. Statistical process control. Boston: Butterworth-Heinemann.
Popper, Karl. 1992. The Logic of Scientific Discovery. London: Routledge.
Salmon, Wesley P. 1966. The Foundations of Scientific Inference. Pittsburgh, PA: University of Pittsburgh Press.
Schoemaker, Paul J.H. 2002. Profiting from Uncertainty: Strategies for Succeeding No matter What the Future Brings. New York: Free Press.
Silver, Nate. 2012. The Signal and The Noise: why so many predictions fail – but some don’t. New York: Penguin Books.
Taleb, Nassim Nicholas. 2007. The Black Swan: The Impact of the Highly Improbable. New York: Random House.
___. 2014. Antifragile: Things That Gain from Disorder. New York: Random House.
van Vliet, Hans. 1993. Software Engineering: Principles and Practice. New York: John Wiley & Sons.
Richard M. Adler
- Chapter 14
Neuer Inhalt/© Stellmach, Neuer Inhalt/© Maturus, Pluta Logo/© Pluta