Delphi Method

The Delphi method of predicting outcomes has been around for a long while but is not widely used in continuous improvement. It is the process of posing questions to many experts and using a summary of their results to further a discussion to predict a future outcome.

One such panel was formed by a television network to estimate how fast HDTV would be adopted. The panel accurately predicted delays, helping the network to postpone significant capital expense and avoid early adopter headaches.

Of note, the name “Delphi” originates from the Oracle at Delphi. In Greek mythology, this is the location in which prophecies from Apollo were handed down.

This logic behind the Delphi method holds true in continuous improvement applications, and even without experts. There is a training exercise in which people are asked to rank order a number of items that would benefit them on a life raft while adrift at sea. The approved answer comes from the U.S. Coast Guard. When the rankings are compiled for the entire group of participants, the aggregate answer invariably scores close to the top of the pile, if not as the top solution. The premise is that compounded ideas are better than individual answers. The challenge is in finding a way to meld the ideas. Delphi does that through its systematic approach.

The basic Delphi method is

1. Select a panel of experts. In most cases, the panel should be anonymous.
2. Give the experts a questionnaire to answer.
3. A facilitator combines the results and creates a new questionnaire.
4. The experts answer the new questions, with access to the previous compiled results.
5. This process is repeated for either a set number of iterations or until a threshold of consensus is reached.
6. The results are acted upon.

With some creativity, the Delphi method can be applied to your continuous improvement efforts. It can be used to predict outcomes. You could ask the materials team about potential performance of suppliers, or an engineering team to predict failures on a product. The results can then drive an action plan to prevent the problems.

In continuous improvement, the Delphi method can be extended beyond just making predictions. It can also be used to develop and winnow down possible solutions to a problem.

This kind of problem-solving tool tends to work best when the results are quantifiable, as they are easier to compile. For example, a panel of experts within an engineering team can estimate the time to launch a particular product. The discussion can be used to present ideas on why someone thought a particular way, which can then lead to a refined individual estimate, and presumably, a more accurate group consensus. It is harder to do this with things like a cell layout, or a website design. Harder, but not impossible. You can still use individual ideas overlaid onto one design.

Words of Warning When Using the Delphi Method

• There is a lower likelihood of getting good predictions when the topic is discussed in a room before making estimates. These conversations can lead to the individuals on the panel altering their beliefs before answering questions. In effect, the estimate becomes less a consolidation of all the views and more of a compilation of the ideas of the most vocal people on the panel. Anonymity prevents valuable ideas from being rejected before they are voiced.
• Be careful: a weak team can drag down a great idea of a star player if the facilitator lets that happen. While the Delphi method can lead to a good estimate, it may not be better than an idea of a prodigy on the panel. The problem is that you never know whose idea is the best one, but you can be fairly confident that the panel consensus will be a good one.
• For basic continuous improvement projects/problems, feel free to use a version of the Delphi method. If the problem is more sophisticated or the stakes are high, make sure you use an expert facilitator.