Expert directory and recruitment
The Expert directory is an organization-scoped, pseudonymous participation index and future-recruitment registry. It helps research teams identify people who may fit a new study while preserving a strict boundary between prior participation, identity access, and permission for future contact.
It is not a general-purpose contact database. Participation in one study never implies consent to another study or permission to receive future recruitment messages.
Three records with different purposes
| Record | Purpose | What it does not imply |
|---|---|---|
| Panelist | A person's relationship with one study | Permission for contact about unrelated future studies |
| Organization expert | Stable, pseudonymous link across studies in one organization | Membership in a study or access to the researcher workspace |
| Directory decision | Versioned evidence of future-contact preference | Consent to participate in any particular study |
A single expert may have multiple study-specific panelist relationships. Ratings, consent, invitation state, withdrawal, and participation history remain study scoped.
What appears in the directory
The /experts workspace has two deliberately different sections:
Available for recruitment
An expert appears here only when all of the following are true:
- the record belongs to the viewer's organization;
- a stable expert record is linked to at least one study panelist;
- an encrypted identity record is available for operational contact;
- the latest directory decision permits future recruitment; and
- any temporary-unavailability period has elapsed.
Email suppression and target-study access are evaluated again when an authorized researcher attempts to add or contact the expert.
Participation index
This broader section includes authorized study participation even when future recruitment is not permitted. It can therefore contain:
- experts who were never offered a directory choice;
- experts who opted out or are temporarily unavailable;
- experts already participating in one or more studies; and
- legacy panelist records with no linkable identity.
An entry marked No linked identity remains a pseudonymous participation record. Delphi Studio does not infer or reconstruct identity from study activity, and the record cannot be recruited directly.
Directory decisions
Future-contact decisions are distinct from study consent and are stored as append-only evidence. Supported states include:
| State | Meaning |
|---|---|
| not offered | No future-contact choice has been presented |
| invited to directory | A versioned notice was successfully sent |
| opted in | Future recruitment contact is permitted |
| temporarily unavailable | Contact is paused until the recorded date |
| opted out | Future recruitment contact is not permitted |
The current status is a projection of that decision history. A later decision adds evidence; it does not overwrite the previous choice.
Directory-choice links are separate, purpose-bound links tied to the expert, organization, source study and panelist, notice version, expiration, and one-time identifier. A study rating or consent link cannot be reused to make a directory choice.
How experts enter the registry
There is no generic Create expert button in the directory. New registry relationships arise through controlled workflows:
- Study panel add or bulk import — adding a panelist with an identity creates or reuses the organization's pseudonymous expert match. The initial directory status remains not offered unless a separate choice is recorded.
- Organization-scoped signup candidate — a confirmed candidate can enter a target study only after organization assignment, explicit versioned recruitment sharing, and administrator approval for panel availability.
- Migration of existing participation — records with safely matchable identities can be linked; identity-less records remain unlinked rather than being guessed.
This design avoids converting historic participation into marketing permission.
Adding an opted-in expert to a study
The Add to study action appears when organization recruitment is enabled and the viewer has panel-management permission for at least one eligible target study.
The researcher must document:
- the target study;
- the versioned inclusion and exclusion criteria in effect;
- a study-specific selection rationale;
- the intended stakeholder role; and
- the conflict-of-interest assessment.
On confirmation, Delphi Studio creates a pending panelist relationship and immutable selection evidence. It does not send an email.
:::info Adding and contacting are separate decisions After the panelist record is created, the researcher reviews the pending relationship and explicitly sends the invitation. Eligible resends are a third, separate command. This separation prevents a selection click from silently contacting a person. :::
For the same study, an existing panelist relationship is reused rather than duplicated. Declined and withdrawn relationships are not reset through directory recruitment.
Organization feature control
Direct expert recruitment is controlled per organization. When the feature is disabled:
- the directory remains available as a read-only participation index;
- add, send, and resend recruitment actions are blocked; and
- existing panelist, decision, selection, email, and audit history remains intact.
An administrator controls the organization flag from Admin → Users → Expert recruitment. Enabling the flag does not create eligible experts or infer opt-in; it only allows authorized actions for records that already satisfy the recruitment conditions.
Identity and anonymity boundaries
Routine directory and analytic views use stable workspace pseudonyms. Real names and email addresses are encrypted separately from ratings.
An authorized panel manager may reveal identity only when:
- a manageable study context is selected;
- a non-empty operational reason is provided; and
- the reveal is committed to the audit trail before identity is returned.
The response is not cached, and identity is not added to general-purpose directory or reporting data. Panelists never see peer identities or peer-level responses.
Qualifications and selection evidence
Qualification records are provenance-bearing claims, not a universal expert score. Claims can be created, verified, updated, or retired while retaining their history.
Study selection freezes the criteria and decision used at that moment. This lets a later reviewer distinguish:
- what the study required;
- what was known about the expert;
- why the expert was selected or excluded;
- which stakeholder role and COI status were assigned; and
- which panelist relationship resulted.
Researchers should screen for fit with declared criteria, not for a candidate's expected agreement with the desired conclusion.
Candidate recruitment
Approved signup candidates are shown only inside a target study's controlled panel-recruitment workflow. The candidate must have:
- confirmed control of the submitted email address;
- explicitly opted into the versioned recruitment-sharing notice;
- been assigned to the organization by an administrator; and
- been marked available to panels by an administrator.
The authorized panel manager still records the study-specific criteria snapshot, rationale, stakeholder role, and COI decision. The candidate becomes an expert and panelist only when that transaction succeeds.
Reporting and audit evidence
The study recruitment report is pseudonymized and includes:
- criteria snapshots and selection provenance;
- stakeholder and COI composition;
- recruitment and invitation outcomes; and
- round participation rates for recruited experts.
Export requires study export permission, is non-cacheable, and creates an audit event. Plaintext identity and the server-side matching key are never included.
Recommended operating practice
- Define panel eligibility and perspective targets in the study protocol.
- Enable organization recruitment only after reviewing notice text, email suppression, and authorized roles.
- Use the recruitment pool only for explicitly available experts.
- Record a study-specific rationale and COI assessment for every selection.
- Treat add, send, and resend as separate decisions.
- Reveal identity only for a necessary operational task and give a meaningful reason.
- Report recruitment flow and attrition without publishing identities.