Skip to main content

What is the Agent section?

Agent is the first of the fixed sections at the top of the Outline on the Questions tab. It is described as “Who is calling, how the study is conducted, and the floor every question inherits.” Every survey has one, it cannot be removed or reordered, and it carries the lock icon and Fixed tag of the other fixed sections. Everything here applies to the whole survey: the interviewer’s identity and voice, the character it speaks with, the limits on the call, and the default attempt caps and step levels that each question uses unless it sets its own. To edit it, open the survey’s working draft, click Agent in the Outline, change the fields and click Save. The confirmation is Agent saved. On a published or imported version the form is read-only and shows “This version is read-only — create a draft to change it.” (see Versions and drafts).

Identity

Choose a voice

A survey has no default voice. You must choose one here before the survey can place or answer calls. Without one, Readiness warns “No voice is chosen — the compiled script will carry none and cannot place a call.” The warning does not stop a compile, so check for it on the Readiness tab before you place calls. The picker lists the voices in your account — the same list as the Voice Library. If a voice saved on the survey has since been removed from your account, the form keeps it and says “The recorded voice (…) is not in this account’s list — it is kept until another is chosen.” If the list cannot load, it shows “The voice list could not be loaded — the recorded id is kept.”

Character

Leave both empty to use the platform default. Keep them short and about manner: the questions, options and FAQs carry the content.

Call

Keyterms

Keyterms help the speech recogniser hear words it might otherwise get wrong — your study’s name, the organisation, local place names, brand names. Add the word as it should be written. Never add a mishearing as a keyterm. A keyterm pulls the recogniser towards that spelling, so adding a mishearing makes it more likely, not less. Mishearings go in the Often heard as list on the option, FAQ or behaviour they belong to. Questions have their own keyterms too, in the question’s Voice panel (see Question settings).

The moving-on line

When this is on, the interviewer says a short acknowledgement after each answer is recorded — lines such as “Moving on.”, “Noted.”, “Got it.” or “Thanks for that.” — before the next question. Praise words such as “Perfect” or “Great” are never used, so respondents are not steered by the interviewer’s reaction. Each question can also turn its own moving-on line on or off with Say a moving-on line after it commits in its Voice panel.

Attempts

“The floor every question inherits unless it states its own.” Each value is from 0 to 50. A question uses these values unless its own Behaviour panel turns off Inherit survey default and sets its own. What each cap controls on a call, and how the per-question override works, is covered in Attempts and step levels. If a question is Required but Decline resolves to 0, Readiness warns that it “is required but accepts the first refusal (attempts.decline is 0).”

Step levels

Three timing and reasoning levels for every question step. Each is Template default, Low, Medium or High. Template default keeps each question template’s own setting. Questions can override these levels individually, and FAQ answers and the busy and stop closes have their own separate sets on the FAQs and Behaviours sections. See Attempts and step levels for guidance on choosing levels — for example, Medium idle survey-wide gives people time to think on a rating question.
  • Gatekeeper — the greeting and consent question, which use the agent name and organisation.
  • FAQs — answers to “who is calling?” and “how long will this take?”.
  • Readiness — the voice warning and every other check.
  • Running a survey — why a voice is required before calls can be made.