Minimize Uncertainty with Robust Decisions
It is not sufficient to have good ideas. Ideas are cheap; the hard part is choosing the right ideas to develop. During early innovation stages information is uncertain, incomplete, evolving and conflicting. Yet, making a robust decision during this early stage is essential to transform innovation into successful actions.
What are “robust decisions?” The name and methods begin with the field of “robust design” popularized by Dr. Genichi Taguchi in the 1990s. The essence of robust design is a two-step optimization process: 1) find the alternative that is the least sensitive to noise (uncertainty) and 2) bring the design to its performance target. This differs from what is generally taught and practiced – do deterministic analysis (ignore uncertainty) and then take into account the effects of uncertainty. Taguchi’s point is that if you do not account for the effects of uncertainty from the beginning, you may end up with a product that is great if everything goes right, but that may behave poorly if there are any changes in the environment in which the product operates, as the product ages or when moving from one example to the next.
Robust decision making extends this philosophy to general decision making with uncertainty considered from the beginning: controlling what uncertainty you can and finding the best possible solution that is as insensitive as possible to the remaining uncertainty. A robust decision is the best possible choice, found by eliminating all the uncertainty possible within available resources, and chosen with known and acceptable satisfaction and risk.
The end goal is to make the best possible choice. The “best possible” may vary with different stakeholders. (A sub-goal is to support the development of best-choice buy-in by all stakeholders.) Improve the quality of the information used in making a decision with the optimal use of available people, time and money. With all reasonable uncertainty eliminated, a decision is made that is as insensitive as possible to the remaining uncertainty, conflicting opinions, inconsistent viewpoints about what is important and information incompleteness. The result of decision making is an option chosen with known and acceptable satisfaction and risk. This implies that (a) you have a way to measure the satisfaction that you believe you can obtain with an alternative and that (b) you know the level of risk that this may not be achieved.
Many methods can help you achieve robust decisions. They range from simple mnemonics to computer supported methods designed to support distributed teams. With any of these methods, a decision checklist can help ensure you are following the best possible decision-making process.
- Realize that there is a decision to be made.
- Form the issue as a single sentence. An issue is the statement about the problem being solved or the question being answered.
- Develop multiple alternatives for resolving the issue.
- List the stakeholders. The stakeholders include everyone who will be affected by the decision.
- Work to generate a set of discriminating criteria and their targets. Identify targets as 1) qualitative, 2) quantitative (<, >, =) and/or 3) 5- point level of agreement or degree. Discriminating criteria are measures that help to separate the good and features of the alternatives. For example, “I want a car with four wheels” is general. “I want a car with four wheels that gets more than 30mpg” provides more detail about the same statement. Typically, there are 8-10 discriminating criteria to drive a particular decision.
- Assign the appropriate criteria to the appropriate stakeholder(s).
- Evaluate the alternatives relative to the criteria by 1) level of evidence, 2) level of certainty and 3) rationale. The level of evidence is a measure of how well an alternative meets a criterion and the level of certainty, as the name implies, judges an alternative’s certainly level. The two levels must be identified in order to make a robust decision.
- Manage and visualize the evaluation uncertainties (using belief maps, for example).
- Fuse the evaluations using a structured method to find 1) alternative satisfaction and 2) alternative risk. In this instance, fusing means to combine the alternative evaluations in order to view the whole situation.
- Base future activity on a what-to-do-next evaluation. A what-to-do next evaluation is the effort to understand the cost and benefit of doing more work to improve the decision.
- Document the decision and the reasoning behind it.
Following these principles can help you achieve more robust decisions, which in turn reduce overall uncertainty and, therefore, facilitate the innovation process.
Dr. David G. Ullman is the President of Robust Decisions Inc., Emeritus Professor of Mechanical Design, Oregon State University and a product designer whose text, The Mechanical Design Process, is used at many universities. His book Making Robust Decisions was published in 2006. Contact Dr. David G. Ullman at ullman (at) robustdecisions.com or visit http://www.robustdecisions.com.