The Delphi Method is a structured expert-consensus technique that uses anonymous rounds of input, feedback, and revision to develop informed judgment when direct evidence is limited or uncertain.
Definition
The Delphi Method is a structured method for gathering and refining expert judgment through multiple rounds of anonymous input. Participants respond independently, a facilitator summarizes the results, and experts revise their views after seeing group feedback. The process continues until the group reaches sufficient convergence or the decision deadline is met.
It is useful when data is limited, the topic is uncertain, experts are geographically dispersed, or direct group discussion may be distorted by status, personality, politics, or anchoring.
History
The Delphi Method was developed in the mid-20th century by researchers at RAND for forecasting and expert judgment under uncertainty. It was designed to improve group estimation while reducing the problems of dominant voices and face-to-face pressure.
Since then, it has been used in technology forecasting, policy planning, healthcare consensus, risk assessment, product planning, quality strategy, and organizational decision making.
When to Use
Use the Delphi Method when expert judgment is needed but hard evidence is incomplete, when the team needs a defensible consensus, or when direct debate would likely be biased by hierarchy. It can support risk prioritization, future-state planning, strategic assumptions, competency models, quality criteria, project selection, and emerging-technology assessment.
Do not use it when direct process data is available and sufficient. Delphi is a structured judgment method, not a substitute for measurement, experimentation, or gemba observation.
Step-by-Step
- Define the question. State the decision, forecast, criteria, or ranking problem clearly.
- Select the expert panel. Choose people with relevant experience, diverse perspective, and enough independence to challenge assumptions.
- Design round one. Ask open or structured questions that collect estimates, risks, requirements, causes, or priorities.
- Collect responses anonymously. Protect independence and reduce status pressure.
- Summarize feedback. Report themes, statistics, ranges, rationale, and areas of disagreement without attributing comments.
- Run additional rounds. Ask participants to reconsider their responses in light of group feedback.
- Measure convergence. Look for stability, narrowing ranges, agreement thresholds, or clear unresolved differences.
- Document the decision. Capture panel composition, assumptions, rounds, results, dissenting views, and how the output will be used.
- Validate when possible. Compare the consensus with later data, pilots, customer feedback, or process results.
Examples
- Risk planning: A cross-site expert panel ranks likely failure risks for a new process before launch data exists.
- Training design: Master Black Belts and operations leaders define the competencies needed for a Green Belt development path.
- Healthcare quality: Clinicians and quality staff develop consensus criteria for escalation when evidence is mixed.
- Technology roadmap: Engineers estimate which automation opportunities are feasible within three years and what constraints must be solved.
- Supplier strategy: Experts anonymously evaluate supplier risk factors before a sourcing decision.
Common Pitfalls
- Poorly framed questions. Ambiguous prompts produce vague consensus.
- Weak panel selection. A panel with narrow experience or hidden conflicts can create false confidence.
- Too few rounds. One survey is not a Delphi study; feedback and revision are essential.
- Forcing consensus. Persistent disagreement may be important information, especially in high-risk decisions.
- No facilitator discipline. Leading summaries, selective reporting, or biased wording can steer the result.
- Treating judgment as proof. Delphi outputs should be validated with data when data becomes available.