> ## Documentation Index > Fetch the complete documentation index at: https://docs.voxworks.ai/llms.txt > Use this file to discover all available pages before exploring further. # Question settings > The settings every survey question shares — title, type, key, voice prompt and piping, alternative phrasings, the Voice panel and the Behaviour panel — what the interviewer does with each answer, and the variables every question records. ## Overview Select a question in the **Outline** on **Setup → Questions** and its settings open in the inspector on the right. This page covers what the question types have in common. Each type's own page covers the settings only that type has. The inspector only shows settings that mean something for the selected type — change the **Type** and the form changes with it. *** ## The top of the inspector | Setting | What it does | Default | | ---------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------ | | **Required** | The switch at the top right. A required question always nudges a refusal at least once before accepting it, even if the **Decline** cap is 0. The question can still end up skipped. Not shown on Display. The outline marks required questions with `*`. | Off | | **Title** | The heading at the top of the pane. It names the question in the outline, on the logic canvas and in Results. Never spoken. | Empty (placeholder *"New question"*) | | **Type** | **Display**, **Single choice**, **Rating**, **Number**, **Long text**, **First name** or **Phone number**. See [the type pages](#question-types). | **Single choice** | | **Question key** | How rules, piping and results refer to the question. *"Rules reference questions by key. Blank uses Q1"* (the next free number). | `Q1`, `Q2`, … | If you rename a key that rules use, the inspector warns *"… Renaming updates them too."* and the rules are updated when you save. ### Question types | Type | Page | | ----------------- | ------------------------------------------------------------------------------------------------------------------------------------ | | **Display** | [Display](/surveys/question-types/display) | | **Single choice** | [Single choice](/surveys/question-types/single-choice), and [Battery (grid)](/surveys/question-types/battery) when **Battery** is on | | **Rating** | [Rating](/surveys/question-types/rating), and [Battery (grid)](/surveys/question-types/battery) when **Battery** is on | | **Number** | [Number](/surveys/question-types/number) | | **Long text** | [Long text](/surveys/question-types/long-text) | | **First name** | [First name](/surveys/question-types/first-name) | | **Phone number** | [Phone number](/surveys/question-types/phone-number) | Other question types can arrive through an import (multiple choice, short text, email, date, matrix and so on). The voice interviewer can't ask them: the inspector shows *"Not supported for voice. The question can be read and deleted; its type cannot be changed here."* Delete them or rebuild them as a supported type. *** ## Voice prompt **Voice prompt** is the question as the interviewer says it — *"What the agent reads out"*. It's required: the save button reads *"Add a voice prompt to save."* until there is one. Write it the way you would say it on the phone. A few rules of thumb: * **End a question with a question mark.** When a Single choice, Rating or Long text prompt has no `?`, the interviewer adds *" — what would you say?"* so the respondent knows it's their turn — unless **Statement prompt** is on (see below). * **Don't read the options in the prompt** on a Single choice or Rating question. The interviewer reads them after the prompt, introduced by the **Options lead-in**. * **Keep it to what a person can hold in their head.** Long preambles belong in a section's **Intro line** or a [Display](/surveys/question-types/display) question. ### Inserting earlier answers and variables (piping) *"Type / to insert an earlier answer, a variable or a contact field."* The picker offers: | Source | Token | Example spoken value | | ---------------------------- | ------------------------------------------------------------------------------------------- | -------------------- | | An earlier question's answer | `{{survey..result}}` | `Very satisfied` | | A survey variable | `{{survey.var.}}` | `Labor` | | A contact field | `{{contact.first_name}}`, `{{contact.last_name}}`, `{{contact.phone}}`, `{{contact.email}}` | `Sam` | Only questions **before** this one in the survey are offered, and Display questions are never offered because they record nothing. A piped value is spoken as text. If the value is empty on the call — the earlier question was skipped, or the contact has no first name — the respondent hears the sentence with a gap. [Readiness](/surveys/readiness) warns about piping that can come up empty. See [Survey variables and piping](/surveys/survey-variables). *** ## Alternative phrasings **Alternative phrasings** — other ways of asking the same question, one per line, up to 20. *"One per line. The agent always asks the first time in the prompt above; when it has to ask again — after a question, or a re-read — it picks between the prompt and these, so it does not repeat itself word for word."* The first read is always the voice prompt exactly as written. Every later read — a re-ask after silence, the return after an FAQ or a "can you call me back?" detour, a clarify line or a nudge — picks at random from the prompt and these phrasings. Two or three good alternatives per question make the interviewer sound far less robotic. Not shown on Display. See [Avoiding repetition](/surveys/repetition). *** ## The Voice panel Click **Voice** to expand it. It holds how the interviewer listens and confirms. | Setting | What it does | Default | Shown on | | -------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------- | --------------------------- | | **Read every answer back** | Reads the answer back for a yes before recording it. On option questions the hint reads *"Off: only answers matched from a mishear are checked"*. See [Answer confirmation](/surveys/answer-confirmation). | Off on Single choice, Rating and Number; on for Long text, First name and Phone number | All but Display and Battery | | **Screen out if the question cannot be run** | If the question's own setup fails on the call, the respondent is screened out rather than moved on. Only needed on a screener question. | Off | Single choice, Rating | | **Say a moving-on line after it commits** | After an answer is recorded, says a short acknowledgement such as *"Noted."*, *"Got it."* or *"Next one."* | Off | All but Display | | **Statement prompt** | *"Spoken as written; no ‘what would you say?’ is added when the prompt has no question mark."* The prompt is spoken once, and silence or an unmatched reply moves on quietly. Use it for a notice authored as a question. It applies to Single choice and Rating questions, other than batteries. | Off | All but Display | | **Look back for it before asking** | Scans what the respondent has already said and reads a value back instead of asking. Off by default because a long transcript is full of names and numbers that belong to other questions. | Off | First name, Phone number | | **Context for the matcher** | Guidance the interviewer uses to interpret answers — never spoken. Placeholder: *"What the scale means, how to treat half answers, volunteered words"*. Also used to answer "what does that mean?". | Empty | All but Display | | **Keyterms** | *"Place, council and brand names. Never a mishear — those belong on the option."* Words the speech recogniser should expect on this question. Aim for a handful, not dozens. | Empty | All but Display | | **Often heard as** | *"What the recogniser tends to produce for this question's own wording."* | Empty | All but Display | ### Writing context Write **Context for the matcher** as matching rules, not as a description of the question. For example: > "Every day" is Daily; anything from a few times a week down to rarely is Less often. "Not sure" is the separate Don't know answer. Refusing is a recorded answer, not a screen-out. If you leave it empty, any help text or interviewer instruction that came in with an import is used instead. ### Keyterms versus "Often heard as" * **Keyterms** pull the recogniser towards a spelling. Use them for real words it might not know — a council, a street, a product. * **Often heard as** lists what the recogniser writes when it gets a word wrong — for example *"can send"* for *concerned*. Never enter a mishearing as a keyterm. A keyterm makes the recogniser more likely to produce that exact text, so a misspelling added as a keyterm makes the mistake more common, not less. The survey-wide **Keyterms** in the [Agent section](/surveys/agent) apply to the whole call; a question's keyterms apply while that question is being asked. *** ## The Behaviour panel Click **Behaviour** to expand it. | Setting | What it does | Default | | ----------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------- | | **Inherit survey default** | On: the question uses the survey's attempt caps from the Agent section — *"The survey’s own Behaviours, from the Agent section. Nothing is stored on this question."* Off: the caps below are this question's own. | On | | **Ask**, **Confirm**, **Spell** (First name only), **Clarify**, **Don't know**, **Decline**, **Acks** | How many times the interviewer re-asks, reads back, asks for a spelling, clarifies, nudges a "don't know", nudges a refusal, and lets a bare "okay" pass, before moving on. 0–50 each. | 5, 5, 5, 5, 1, 1, 3 | | **Eagerness**, **Idle**, **Thinking** | How quickly a pause ends the respondent's turn, how long a silence is tolerated before a prompt, and how much the interviewer reasons about each reply. **Template default** or low / medium / high. Left on **Template default**, the question follows the survey's **Step levels** in the Agent section. | Template default | Display questions show only the step levels. Each cap and level is explained, with guidance, in [Attempts and step levels](/surveys/attempts-and-step-levels). If **Required** is on and **Decline** is 0, the inspector warns: *"Required questions nudge a refusal at least once, but decline nudges is 0 in Behaviour. Raise it to 1 or turn Required off."* *** ## Saving * **Add question** (a new question) or **Save changes** — reads **Saving…** while it saves. Toasts: *"Question added"*, *"Question saved"*. * **Cancel** discards a new question; **Delete** removes an existing one (toast *"Question deleted"*). A question in use by rules or variables, or with recorded answers, can't be deleted — see [Questions tab](/surveys/questions-tab#deleting-a-question). On a read-only version, the inspector shows the stored values instead: *"This version is read-only. Create a Voxworks copy to edit it."* *** ## What happens when the interviewer asks Every answerable type follows the same pattern, with the details varying by type: | The respondent… | The interviewer… | | -------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Gives a clear answer | Records it and moves on (reading it back first if confirmation applies). | | Says "sorry?" or "say that again" | Re-reads the question in other words. This uses one **Ask** attempt. | | Says "okay" or "uh-huh" while the question is being read | Carries on — up to the **Acks** cap, this doesn't use an attempt. | | Says "I don't know" | Nudges once — on a choice question, for example, *"No worries — if you had to pick one, which would be closest?"* — then records the question as skipped. | | Declines to answer this question | Nudges once — on a choice question, for example, *"That's no problem at all — we can skip it. Unless one of those is close enough for you?"* — then records it as skipped. | | Gives an answer that doesn't fit | Asks again with a clarify line, up to the **Clarify** cap. | | Says nothing | Re-asks — *"Let me read that again — …"*, then *"No problem — once more: …"* — and skips the question when the **Ask** cap is spent. | | Asks "what does that mean?" | Gives a short, neutral explanation drawn from the question and its context, without suggesting an answer, then waits. | | Asks about the call, is busy, or wants to stop | Hands over to your [FAQs](/surveys/faqs) or [Behaviours](/surveys/behaviours). After an FAQ answer, the question resumes. | When a question is skipped the interviewer moves on gracefully — *"That's okay, let's move on."* If the question's own setup can't run on the call, the respondent hears a brief apology and the survey continues (or screens them out, if **Screen out if the question cannot be run** is on). How the interviewer tells an answer apart from an FAQ or a detour is covered in [Answers given early and interruptions](/surveys/riders-and-interruptions). *** ## What every question records Each question writes these custom variables on the call. `` is the question key in lower case, with spaces and punctuation turned into underscores — so a question keyed `Q3` records `survey.q3.result`. | Variable | Type | Example | Use | | ---------------------------------------- | ---- | ---------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `{{survey..result}}` | text | `Very satisfied` | The answer. Empty when the question was skipped or never reached. What it holds depends on the type — see each type's page. | | `{{survey..status}}` | text | `captured` | `captured` (answered), `skipped` (the respondent didn't answer, or attempts ran out), `invalid` (no usable answer, or the question couldn't run), `abandoned` (the call ended on this question after a busy or stop request), or empty if never reached. | | `{{survey..termination_requested}}` | text | `false` | `true` when this answer ended the interview (a screen-out). | Later prompts can speak `{{survey..result}}`, and routing rules read the answer directly. After the call, answers are stored against the survey session and appear in **Results** and in the CSV export's `answer` column. See [Survey variables and piping](/surveys/survey-variables) and [Responses and export](/surveys/responses-and-export). *** ## Next Steps * [Answer confirmation](/surveys/answer-confirmation) — when answers are read back. * [Attempts and step levels](/surveys/attempts-and-step-levels) — the Behaviour panel in depth. * [Single choice](/surveys/question-types/single-choice) — the most common question type.