Sign inRequest early accessEarly access

Early access for one study

Run a Delphi study you can defend in review.

Design, run, analyze, and audit a modified Delphi study in one workspace, then download formula-safe CSV and a Markdown summary.

  • Set the rules before launch.
  • Run the rounds.
  • Export the tables and the audit log.
Request early accessSee how a study runs

Email info@delphistudio.org. Include your name, institution, role, and when round 1 needs to open. No price is published. The message does not create an account.

Example interface. The study and the counts are not real. Round 2 is still open, so this picture does not show consensus or a decision.
Two jobs

Each one ends in a file you can hand someone.

Quality improvement, policy, and consortia use the same two files.

A guideline group

Lock the scale and the consensus rule before round 1, invite the panel, remind people who have not answered, and export the consensus table and the audit log.

An education team

Set who may see identities, collect ratings and comments across two rounds, and hand a co-investigator a CSV and a Markdown summary.

The research reality

Your tools were built for surveys.
Not consensus methodology.

A second round that shows a person their previous answer beside an anonymised group summary, a rule declared before the data, and a record of what was done.

The usual workaround
With Delphi Studio
Spreadsheets, email threads, and disconnected files
Versions get lost, and the methods section is assembled at the end.
One study workspace
Ratings, decisions, and the audit stay with the study, from the protocol through the rounds.
Manual round setup, reminders, and closeout
Someone has to remember who has not answered, and chase them by hand.
A schedule, a queue, and a count
Rounds can open and close on a schedule. Reminder emails are queued for people who have not submitted, when sending is turned on. The round header shows how many have submitted.
Statistics in a side script
Median and IQR are computed somewhere else, and the file is not the study record.
Statistics after the round closes
The workspace computes median, IQR, agreement, and Kendall's W, and keeps that version sealed. Kendall's W runs from 0 (no agreement of orderings) to 1 (identical orderings). It does not decide the round. Stability between rounds can be written down. It is not calculated yet.
Decisions with no trail
The reason an item was kept or dropped lives in a note.
A hash-chained audit you can export
Rule changes and exports are written to a hash-chained study audit. A principal investigator can download it.
A methods section assembled at the end
Tables are copied out of a spreadsheet into a manuscript by hand.
CSV and a Markdown summary
Download formula-safe CSV for responses, consensus, and participation, and a Markdown summary of rounds and outcomes. A Word report is not available.
Delphi Studio gives consensus researchers the structure their methodology requires.
See how a study runs
Scale and zones lock at launchResults sealed after closeCSV and Markdown
How a study runs

Set the rules, run the rounds, then export.

Modified Delphi percentage-and-IQR rules, and RAND/UCLA appropriateness classification on a 1 to 9 scale. The stopping percentage is advice. It does not end the study.

  1. Design
    Set the items, the scale, and the thresholds before launch.
  2. Panel
    Build the panel and record consent. Invitations leave only when sending is turned on.
  3. Open
    Open a round and collect ratings and comments.
  4. Count
    The round header shows how many people have submitted. Reminder emails are queued for people who have not, when sending is turned on.
  5. Feedback
    Comments are coded by two researchers and an adjudicator. The team chooses what the next round shows.
  6. Export
    Download formula-safe CSV and a Markdown summary of rounds and outcomes.
Consensus emergence

See expert agreement converge across rounds.

Each bar shows an interquartile range: how widely the illustrated ratings are spread. A narrower bar is not a decision. Retain, Uncertain, and Exclude are the words the app uses after a round closes.

Illustration only. These items and figures are not a Delphi Studio study. The bars are on a 1 to 9 scale.

Item
Round 1: IQR spread
→
Round 2: IQR spread
Outcome
Multimodal analgesia as defaultMedian 5→8 · IQR 2.8→0.8
Round 1: Wide spread, no consensus
Round 2: Tight agreement, high score
Retain
Preoperative risk stratificationMedian 6→7 · IQR 3.1→1.0
Round 1: Moderate spread, under discussion
Round 2: Agreement reached
Retain
Pre-op skin decolonizationMedian 5→5 · IQR 2.5→2.1
Round 1: Spread, mixed expert opinion
Round 2: Slight narrowing, needs another round
Uncertain
Antibiotic prophylaxis >24h post-opMedian 3→3 · IQR 2.2→1.8
Round 1: Low rating, mixed on exclusion
Round 2: Consistent low rating, excluded
Exclude
After a round closes

What the numbers are, and what is not computed yet.

Median, interquartile range, agreement, and Kendall's W are computed when someone runs results after a rating round closes, and that version is sealed. Kendall's W runs from 0 (no agreement of orderings) to 1 (identical orderings). It does not decide the round. Stability between rounds can be written down. It is not calculated yet.

Locked at launch

At launch, the scale, zones, and classification thresholds are locked. Changing the percentages later is a recorded amendment.

  • Modified Delphi thresholds, applied by the software
  • RAND/UCLA on a 1 to 9 scale
  • A later change is an amendment, not a silent edit

Coded by people

Comments are coded by two researchers and an adjudicator. The team chooses what the next round shows.

  • No model is called
  • The team chooses the feedback
  • The study can be labeled CREDES, ACCORD, or none

What you can hand someone

Download formula-safe CSV for responses, consensus, and participation, and a Markdown summary of rounds and outcomes.

  • A Word report is not available
  • The label does not produce a checklist or a manuscript
  • The principal investigator can export the audit

CREDES is a reporting guideline (EQUATOR (opens in a new tab)). ACCORD is a reporting guideline for consensus methods (PLOS Medicine (opens in a new tab)). Storing either label does not produce the checklist. RAND/UCLA appropriateness classification runs on a 1 to 9 scale.

An ethics office

What you can say, and the limits in the same section.

A study can record the review. It does not submit a protocol, store the approval file, or hold a certification.

Names

Panelists do not see each other's names. After enrolment, staff see a pseudonym unless a principal investigator reveals a name, which is recorded, or downloads an identifiable file with a written reason.

The ethics record

Record the ethics review, its number, and the names of the approval documents. Launch waits until that record is complete enough. The file itself is not stored yet.

The rules

Thresholds are stored with the round. The scale, zones, and classification thresholds lock at launch. The stopping percentage is advice, not an automatic stop.

The audit

The organization and study audits are hash-chained, and a principal investigator can export the study audit. Consent and withdrawal are recorded with the study.

Early access

Write about one study. There is no price on this page.

A principal investigator or a small methods group should write. Include your name, institution, role, and when round 1 needs to open. Mail info@delphistudio.org. That mailbox is not the study panel. Nothing here takes a card or opens an account.

01 · Look

See an illustration

The dark band is a picture of agreement tightening. The items and the figures are not a Delphi Studio study.

See an illustration
02 · Read

See how a study runs

Six steps, on this page: set the items and the thresholds, build the panel, open a round, count responses, share only the feedback the team chose, and export.

See how a study runs
03 · Join

Request early access

Request access and confirm your email to join the waitlist. Joining does not create an account or guarantee access.

Request early access

Data note

Delphi Studio is in early access. This page does not name a corporation, a postal address, or a price. Who operates a given deployment, where the data sits, and which agreements apply are not stated here.

info@delphistudio.org is the address for an early-access email. It is not the study panel. A message to it is not enrolment, and it does not create an account.

Identity fields in a study are encrypted with AES-256-GCM. The organization and study audits are hash-chained. The software starts a retention clock when a study is completed and does not delete the study by itself. To ask about deletion, write to the same address.

The site sets one session cookie, for signing in, and no tracking: no analytics, no advertising cookies, and no script on the pages panelists use. That is why there is no cookie banner. The privacy policy says the same.

This page does not claim HIPAA, a business associate agreement, or that the service is ready for an ethics board. A study can record an ethics review. The approval file itself is not stored yet.