Delphi Technique
The Delphi Technique is a structured way to bring a panel of experts or stakeholders to a considered judgement through several rounds of individual, usually anonymous, answers, with a summary of the group's responses fed back between rounds so people can revise their view. It was developed at Project RAND during the early Cold War by Olaf Helmer, Norman Dalkey and Nicholas Rescher, originally for forecasting, and Dalkey and Helmer published an experimental application in 1963. The classic form is written and remote and runs over several weeks; the duration given here refers to the face-to-face workshop variant described by Bob Dick in 1991.
Facilitation script
- 1
Face-to-face variant. Present the question and the process, and form small groups of people with similar roles or perspectives.
10 min - 2
Individual work: everyone writes their own answers to the question in silence.
10 min - 3
Small groups combine their answers and agree a ranked list of their most important items.
20 min - 4
Whole group: compile all items on flip chart paper, merge duplicates, then run a multi-vote to produce a ranked group list.
25 min - 5
Individuals note what they would change in the group list. Small groups then compare it with their own list and prepare evidence for any change they want.
20 min - 6
Groups report in turn under a strict time limit, stating whether they accept the list or presenting their evidence for a change. No debate during reports.
15 min - 7
Vote again on the revised list. Repeat the review, report and vote cycle if clear disagreement remains.
25 min - 8
Record the final ranking, the points of remaining dissent and the reasons given for them.
10 min
Tips
The quality of a Delphi depends on the summaries between rounds, so plan enough time to write them and report the reasons behind outlying answers as well as the averages.
Keep the question narrow; a panel cannot converge on a topic, only on a question.
Tell panellists at the start how many rounds to expect and how long each takes, because drop-out between rounds is the most common way a Delphi fails.
Common pitfalls
Feeding back only averages or medians, so panellists drift towards the middle without having heard a single argument and the consensus is an artefact
Selecting a panel of like-minded experts, which produces fast agreement that reflects the selection and not the question
Asking a broad or double-barrelled question in round one, so the answers cannot be compared and later rounds argue about wording
Losing panellists between rounds without noticing, so the apparent convergence comes from who stayed, not from anyone changing their mind
Reporting the result as a single consensus figure and leaving out the dissent, which hides exactly the uncertainty decision-makers need to see
Variations
Classic remote Delphi: questionnaires by email or survey tool, three or four rounds over several weeks, panellists never meet. Face-to-face Delphi (Bob Dick): one workshop of two to four hours in which small groups of similar stakeholders stand in for individual panellists and evidence is presented under strict time limits, for example one minute per report. Real-time Delphi: an online form with no fixed rounds, where panellists see the running group result and can revise at any time. For estimates in small teams, Planning Poker applies the same idea of independent answers followed by discussion of the outliers.
Where it fits
When to use it
A judgement is needed on a question where data is thin and informed opinion is the best available evidence
The people whose view counts are spread across organisations or time zones and cannot be brought together
Seniority or reputation would distort an open discussion, and anonymity is the only way to get independent answers
A topic is contested and you want people to respond to arguments without knowing who made them
You need a documented process showing how a set of priorities or criteria was reached
When not to use it
A decision is needed this week; the classic form takes weeks, so use Nominal Group Technique in a single session
The question can be answered with existing data or a quick experiment; collect the data
The aim is to build relationships or shared ownership among participants; anonymous written rounds do neither, and a World Café or Open Space Technology serves better
You want new ideas, not convergence on a judgement; run an ideation method such as 1-2-4-All
The panel is small and trusts each other; a facilitated discussion with Gradients of Agreement is faster and loses little
Related methods
Frequently asked questions
What is Delphi Technique?▾
The Delphi Technique is a method for reaching a considered group judgement through repeated rounds of individual, usually anonymous, answers. After each round a facilitator summarises the responses and the reasons behind them and returns the summary to the panel, who may then revise their answers. It was developed at Project RAND for forecasting and is now used widely for priority setting and consensus building.
How many rounds does a Delphi need?▾
Classic applications use two or more rounds, and three or four is common. Stop when answers have converged or have stopped changing between rounds; further rounds then only cost you panellists.
How long does a Delphi take?▾
A remote Delphi with three or four rounds is likely to take several weeks, since each round needs time for answers and for writing the summary. The face-to-face variant fits into a workshop of roughly two to four hours. That workshop figure is an estimate; the sources give no fixed duration.
Why are the answers anonymous?▾
Anonymity stops well-known or senior panellists from steering the group and lets people change their mind without losing face. Panellists respond to the arguments in the summary, not to the person who made them.
What is the difference between Delphi and Nominal Group Technique?▾
Both start with independent individual answers before any group influence. Nominal Group Technique happens in one live session with the participants in the room. Delphi keeps panellists apart and anonymous across several rounds, which suits dispersed experts and sensitive topics but takes much longer.
Plan your next workshop with AI
Workshop Weaver helps you combine methods like Delphi Technique into a complete, timed agenda in minutes.
Try it freeMethod descriptions on Workshop Weaver are original content written by our team, based on established facilitation practices. This method was inspired by work from Project RAND (Olaf Helmer, Norman Dalkey, Nicholas Rescher); face-to-face variant by Bob Dick.