Delphi Technique 731
Photo by: Yuri Arcurs

The Delphi technique is an intensive and fairly specialized group problem-solving method used to harness and reconcile the knowledge and judgment of several experts. There is no single way of conducting a Delphi study; hence the concept, also known as the Delphi method, refers to a general process of having experts formulate solutions to problems through several cycles of revision based on each other's feedback. Ideally, the end result is a better solution than any of the experts could have arrived at individually. Originally developed as a forecasting tool, it was created by the U.S. military around 1955 for the purpose of estimating the probable effects of a massive atomic bombing attack on the United States. In the mid1960s, researchers began applying the technique to technological forecasting. Since then, its use has spread into other disciplines as a method of identifying and solving problems.


In the early 1950s, there existed considerable concern in the United States over possible atomic bomb attacks by Russia. Experts were interested in what would happen if such attacks did occur. Therefore, the Air Force, which was responsible for defending the nation's skies, decided to ascertain what damage would be incurred by the country if it were attacked via air. The Air Force commissioned the Rand Corporation to question experts regarding what would happen as a result of these theoretical bombing raids. Thus was born the Delphi technique.

Guessing at what would happen as the result of an atomic bomb raid was dependent on the collection and analysis of large amounts of data. Today, that would be considerably easier than it was in the 1950s. Computers were just coming into popular use then, and data collection techniques had not been refined. Even if large amounts of data were collected, project members would have had to rely heavily on subjectivity, since information on Soviet intelligence, military strength, economic policies, etc., was not readily available. Rand Corporation sought an alternative collection and analysis method.


The founders of the Delphi technique based it in part on traditional philosophical premises. They looked to philosophers such as Gottfried Leibnitz, John Locke, Immanuel Kant, Georg Hegel, and Isaac Singer for the foundations of the techniques. They developed a basic set of questions on which to base the Delphi inquiry methods. Questions included:

These were by no means all the questions the technique's founders asked, but they represented a start. They were simply looking to find all the strengths and weaknesses involved in their new technique. They applied the philosophical lessons to the realities involved in the method.


Delphi technique developers wanted to delineate sharply between theory and reality. Ultimately, what they wanted was for Delphi panel members to conceptualize in realistic terms the problems on which they were working, rather than in theoretical terms. Theoretical findings, after all, were of limited value unless they were matched to reality. Since no atom bomb raids had actually been conducted by the Russians, the Delphi practitioners had no reality on which to base their findings. Yet they were expected to develop accurate results of a true bombing. The question, then, was how to arrive at those findings without relying too heavily on theory.

They perceived a step-by-step process through which the conceptual process could be carried out. Basically, it consisted of preparing the materials, interpreting the responses, integrating the insights, and presenting the results. The key was in repetition of these steps. For example, once the results of a round of questions were presented to Delphi members, they could reformulate their answers based on the new input. The founders believed that through the constant iterations of the questions based on new information, the answers the participants ultimately presented would be as realistic as possible.


The Delphi technique comprises several steps involving participants who may or may not meet face to face. The participants (or panel) might be employees of a specific company conducting a Delphi project or experts selected from the outside for their in-depth knowledge of a given academic discipline or manufacturing process. In the original technique, no matter who they were or what their specialties, the experts were to be kept isolated from one another. Later elaborations of Delphi methods eased such restrictions when circumstances seemed to warrant it; indeed, in some cases Delphi panels met in person in a highly structured conference setting.

The purposes of keeping them apart, or at least tightly structuring their communications, were to restrict undue influence that individuals may wield in group situations and, if necessary, to protect participants' anonymity. The technique was designed so that the participants' physical presence in the process is unnecessary throughout the multi-step process.

The essential steps in a traditional anonymous Delphi study are as follows:

  1. Articulate a problem.
  2. Ask participants to provide potential solutions through a series of carefully designed questionnaires. There may be a specific time horizon over which the solutions are to be based, e.g., for 50 years in cases of scientific breakthroughs.
  3. Participants complete the first questionnaire anonymously and independently.
  4. The results of the first questionnaire are tabulated at a central location, transcribed, reproduced, and forwarded to each participant.
  5. After reviewing the initial results, members submit new solutions.
  6. They may make new estimates, often based on a 50-50 probability, as to whether each breakthrough will occur earlier or later than predicted by the other participants. Or, they might suggest that the events predicted may not occur at all and explain why they feel that way. In any case, each round of results invariably triggers new resolutions.
  7. The submission of new solutions based on reiterations of the questionnaire process is repeated until the participants reach a consensus.

Generally, three to four cycles of the process result in a consensus among the participants.


There are several advantages to the Delphi technique. One of the most significant is its versatility. The technique can be used in a wide range of environments, e.g., government planning, business and industry predictions, volunteer group decisions. Another important advantage lies in the area of expenses.

For example, the Delphi technique saves corporations money in travel expenses. They do not have to gather participants from several points of the globe in one place to resolve a problem or predict the future, yet they still can generate relevant ideas from the people best suited to offer their expertise. This is particularly beneficial to multinational corporations, whose executives and key personnel may be based in cities as far apart as Melbourne, New York, Tokyo, Buenos Aires, and London.

The technique also protects participants' anonymity. Thus, they feel better protected from criticism over their proposed solutions and from the pitfalls of "groupthink", i.e., the withholding by group members of different views in order to appear in agreement. On the other hand, the technique has its drawbacks.


The Delphi technique is somewhat time consuming, which renders it ineffective when fast answers are needed. It might also be deficient in the degree of fully thought-out resolutions. People acting together in a group setting benefit from others' ideas. Thus, there might be more insightful and pragmatic resolutions to problems offered by people in interactive settings, e.g., through the nominal group technique, in which participants are gathered in one place but operate independently of one another. However, in situations where time is not of the essence or group interaction is not important, these disadvantages diminish in importance.

Another drawback to using the Delphi technique is that it can be difficult for researchers to design an effective study. As with survey and other respondent-dependent research designs, the results from a Delphi study are determined in large part by how they are framed and conducted. For example, the study coordinator may inadvertently railroad dissenters on the expert panel into accepting the consensus view before allowing them to express potentially important ideas that might otherwise change the consensus. Similarly, if the study coordinator is summarizing each participant's responses, care must be taken that the full breadth and depth of each expert's comments is recorded for the others to respond to.


There exist variations of the original Delphi process, which was applied primarily to technical topics. For example, there is the Policy Delphi, which marked the first major deviation from the original technique.

In the original process, the goal was to seek a consensus among homogeneous groups of experts on a given topic. Several years after the Delphi technique came into use, proponents introduced the Policy Delphi, which was based on a somewhat different approach.

The Policy Delphi tries to generate the strongest possible opposing views on the potential resolutions to a problem. Policy Delphi is based on the premise that the person (or people) making the ultimate decision in the process does not want the group to generate the decision. Rather, the idea is to have an informed group present all the options and leave the ultimate resolution to the decision maker.

Other adaptations of the Delphi technique focus on getting faster results from the expert panel and minimize the number of times they are asked to reevaluate the problem based on their peers' opinions. These abbreviated Delphi methods focus on reaching a consensus quickly, perhaps at the expense of details and nuances, and may only involve two rounds of responses from the experts.


The Delphi technique is applicable to many situations requiring group solutions to problems or prognostications about future events. Because of its nature as a rigorous forum for advancing educated guesswork, the Delphi technique is best suited to problems that require evaluative, qualitative answers rather than precise, quantitative results. Generally, Delphi studies are also most useful for assembling groups of experts who would otherwise probably never come into contact. A company would not be likely to need the Delphi method to gather ideas from a group of coworkers within the same office, for example.

Many pioneering academic uses of the Delphi technique occurred in the late 1960s and early 1970s. For example, in 1970, Professors Alan Sheldon of the Harvard Medical School and Curtis McLaughlin of the University of North Carolina Business School led a Delphi project on the future of medical care. The two applied a somewhat different approach in their project (another indication of how flexible the Delphi technique can be). They combined the events evaluated by the respondents into scenarios in the form of typical newspaper articles. They asked the respondents to propose additions or modifications to the scenarios and give their reactions to the scenarios as a whole. Inadvertently, they introduced a new way of conducting Delphi projects. Their concept of utilizing the vote on individual items and group events in scenarios classed by such things as likelihood and/or desirability became a standard technique in the project.

Bell Canada was one of the first businesses to adapt the Delphi technique to its forecasting activities. In the late 1960s, the company developed a study plan to evaluate future trends in the visual and computer communications fields. There existed a clear lack of qualitative data on future prospects for these fields as they pertained to Canada. Company leaders decided to use the Delphi technique to examine the future of these fields.

Bell Canada conducted a series of wide-ranging Delphi studies to determine the future course of technology and its applications in diverse areas of life. The company developed extensive questionnaires asking a variety of questions examining the future of technologically advanced applications from a user's point of view. The questions touched on social issues such as value changes in North American society and who would be most likely to use new communications products. Products included in the study included computerized library systems and computer-aided instruction systems for education, remote physiological monitoring and computer-assisted diagnosis for medicine, and terminals and data processing in the business environment. The study addressed questions such as evolutions in school design, trends in the medical profession, and changes in business procedures. Bell Canada derived an impressive amount of information from the studies.

The data Bell Canada derived resulted in a significant increase in the company's store of qualitative data used in the planning and forecasting processes. Perhaps more importantly, both for Bell Canada and other Delphi users, the studies led to widespread modifications from the original Rand Corporation approach, particularly in the emphasis on analyzing participants' comments and establishing threshold levels of acceptance.

More recent applications of the Delphi technique have addressed the emergence of electric and hybrid gas-electric automobiles, predicted tourism in certain locales, and identified the potential for sophisticated technology waste-management systems.



Information derived from Delphi studies can be highly valuable to businesses. For example, entities involved in Delphi projects can trade results with other interested parties, e.g., members of a trade association or competitors. They can furnish the results to specialists in their organizations, e.g., engineers or scientists, to stimulate innovation. Or, the results can be used as an educational tool for senior managers who are attempting to predict a company's future course via long-range planning.

Long-range forecasting is essential to any business that hopes to survive in the increasingly competitive global environment. While executives may not have enough time to gather data via Delphi for short and intermediate planning, they can use it for long-range forecasting. Delphi projects are ideal for such a purpose.

Many companies have utilized the Delphi technique as a forecasting tool. Among them are TRW, IBM, AT&T, Corning Glass Works, Goodyear Tire and Rubber Company, and the Alaska Department of Commerce and Economic Development, which used the Delphi technique in the early 1980s to assess the state's energy, economic, and resource development future.

The billion-dollar company Bharat Heavy Electricals, Ltd. (BHEL), India's largest heavy electrical equipment company, demonstrated the value of Delphi in long-range planning in the mid-1980s. The company explored the future direction of power development in India, particularly in the fields of electric energy and electric transportation. BHEL's products create systems for electric power generation through thermal, nuclear, and hydro sources, and for power transmission to India's industrial and transportation processes. To forecast the company's needs in these areas, management solicited data from 286 employees in a variety of engineering disciplines.

In the first round, participants received an open-ended questionnaire. The purpose was to gather as much information as possible regarding major technological breakthroughs that conceivably could be developed over the next few decades. Participants were asked to estimate when these breakthroughs would occur.

The compiled list of ideas and predictions was returned to the participants in the second round. The respondents were asked to reassess their earlier estimates in view of the new information. They were also requested to assign a priority ranking for each projected technological development. Once they completed this round, a consensus began to emerge. In the third round, the participants submitted final estimates based on the results of the second round, along with their rationales for their ideas. The results were positive.

The process identified the likely development of 19 new energy sources. It also provided estimates as to when each of the new sources would be available. BHEL's management found these predictions extremely helpful in its long-range forecasting plans—which is exactly what the Delphi technique is designed to do.


One of the most common misuses of Delphi-generated information is to assume that the results are to become official company policy. That is not the purpose of the Delphi technique. The information gathered is intended to be used simply as an aid in the problem-solving or forecasting process, not as the foundation of company policy.

Similarly, corporate public relations representatives sometimes treat the results of Delphi projects as information that is to be applied immediately to the production of goods and/or services. Hence, in their minds, what Delphi participants have projected to occur in 30 or 40 years is to be implemented immediately. They release the results to the world at large in order to tout their company's innovative philosophies and progressiveness, only to discover that the information is of no use in the present. In the process, they might inadvertently leak trade secrets to competitors and harm their own company's competitive advantage. Worst of all, they can damage their company's reputation.

One last misuse of Delphi results is in the interpretation of the findings. In many cases, results are fed into computers for analysis and verification. The computer programs that are used may be making "best guess" predictions regarding the data. For example, in the initial Rand study, any computerized analysis of the data analyzed by a computer could not have been based on actual events, since they never happened. Thus, any results of Delphi data analyzed by the computer would have been based on projection. It is easy, then, for people to misinterpret computer-generated analyses as events that will occur exactly as predicted. Mistakes such as this can render the results of Delphi projects unreliable and unusable. Despite the potential for misuse, Delphi procedures are still valuable tools for researchers.


The Delphi technique has evolved dramatically since its first application in the 1950s. Researchers have expanded its uses and modified the procedures through which they gather information. The evolution of computers and their applications have simplified the decision-making part of the Delphi process. Computer models can now make more efficient use of the data gathered through basic techniques and generate highly realistic projections and results of future events.

The modifications and enhanced computers have by no means banished Delphi to the scrap heap of forecasting history. Indeed, the opposite is true. The Delphi technique will remain a viable technique in the business world for the foreseeable future. In fact, it will take on added importance as global competition expands and finite sources of raw resources diminish. Corporations will seek new replacements as resources such as oil, coal, and minerals reach extinction. They will look for more efficient production techniques that will enable them to improve their profitability ratios, e.g., their net profit margins and return on investment, in order to remain competitive with domestic competitors and manufacturers in emerging nations. It is not inconceivable that Delphi projects will be used to promote innovative ideas regarding the future of space travel, colonizing other planets, etc. That is simply because Delphi is a technique designed primarily to deal with long-range forecasting.

SEE ALSO : Forecasting

[ Arthur G. Sharp ]


Boks, Casper, and Erik Tempelman. "Future Disassembly and Recycling Technology: Results of a Delphi Study." Futures, June 1998.

Drucker, Peter F. Technology, Management and Society. New York: McGraw-Hill, 1970.

Gupta, Uma G., and Robert E. Clarke. "Theory and Applications of the Delphi Technique: A Bibliography (1975-1994)." Technological Forecasting and Social Change, October 1996.

Linstone, Harold A. and Murray Turoff, Editors. The Delphi Method: Techniques and Applications. Reading, MA: Addison-Wesley Publishing Company, 1975.

Ng, H.K., J.L. Anderson, and D.J. Santini. "Electric and Hybrid Vehicles: A 25-Year Forecast." Automotive Engineering, February 1996.

Wheelright, Steven C. and Spyros Makridakis. Forecasting Methods for Management. New York: John Wiley & Sons, 1985.

User Contributions:

Comment about this article, ask questions, or add new information about this topic: