Guided Story + Segmentation (plugin type unified_single_multiple) combines a voice narrative with an AI-driven single- or multiple-choice classification: the AI reviews the spoken answer and picks (or scores) it against a set of choice options you define, in the same frame as the recording. It's wizard-driven.
What respondents see#
The same greeting-and-microphone recording flow as Guided Story. After the respondent answers, the AI reads the open-ended response and selects one or more of your predefined choice options based on what was said — the respondent doesn't manually pick from a list themselves.
When to use it#
- When you need both an open narrative and a structured classification of that narrative — e.g. "tell us why, then we'll bucket you into a segment" — without a separate follow-up question.
- Replacing a spoken answer + a manual follow-up choice question with one AI-scored step.
- Segmentation or driver studies where the segment should be derived from language, not a forced-choice click.
Configuring it in the builder#
Drag Guided Story + Segmentation from the StorySync AI Qual group onto your canvas — the wizard opens with Questions, Display Settings, and Text for Alerts. In addition to the shared greeting/sub-question/guardrails fields, the Questions step has a dedicated Question to Answer From Open End section:
- Question Type — Single Select or Multiple Select (whether the AI picks one or several choices).
- Question to Answer — the classification question the AI uses to extract an answer from the open-ended response (e.g. "Which of the following types of vacation did the respondent like most?").
- Choices — the answer options the AI can choose from, with support for an Exclusive option (e.g. "None of these"). Use Add choice for one more line, or Bulk Edit to paste or edit the whole list at once and set recode values. Choices are plain text: Bulk Edit here has no Image URL column and no link, tooltip, piped text or formatting buttons, because the AI reads only the words. A choice you add gets the next free recode value; the values already set never move. This is the only place the choices are edited.
Settings reference#
| Setting | What it controls | Default |
|---|---|---|
| Question Type | AI picks one choice or several | Single Select |
| Question to Answer | The classification prompt the AI applies to the narrative | — |
| Choices | Answer options the AI can select, with an optional exclusive/"none" option. Edited here, or all at once with Bulk Edit | Five example choices, ending with None of These (Exclusive) |
| Show Image/Video in Question | Reveals the media fields (Media Type, image count, image URLs 1–4, video URL) for a stimulus at the top of the question | Off |
| Initial Instructions | Text shown before the first sub-question to set context; supports piped text | Seeded example prompt |
| Voice Mandatory | Voice input required, no text fallback | Off |
| Show Transcription | Live transcript while the respondent talks | On |
| Default to Voice | Voice recording is the default input method (only shown when Voice Mandatory is off) | On |
| Request Voice or Text Preference | Ask respondents whether they'd rather record or type (only when voice isn't mandatory) | Off |
| Tap to Zoom Image | Allow respondents to zoom the stimulus image | Off |
| Enable Image Scrolling | Let respondents scroll through multiple images | Off |
| Enable Auto Probing | Automatically probe on any question if the response is too short or lacks detail | Off |
| Use Whole Conversation | Feed the entire prior conversation into AI-generated probes | Off |
| Is Loaded from Iframe | Marks the question as embedded in an iframe inside an external survey | Off |
| Enable Voice Attribute Detection | Detect speaker gender/age/language/accent/human-vs-AI | On |
| Honor voice-permission globals | Follow survey-level voice/mic-permission setting | On (recommended) |
| Concept Image Height | Height of the stimulus image as % of screen | 50% (range 30–70) |
| Question Font Size | Font size for question/instruction text | 17px (range 10–40) |
| Maximum Recording Duration | Cap on voice recording | 5 minutes (range 3–10) |
| Select Language | Speech-recognition language | English (+ 6 others) |
| Select Text Direction | LTR or RTL layout | Left to Right |
| Button Color | Primary action-button color | #34ac8b |
Using the choices in logic, quotas, carry forward and loops#
Each Guided Story + Segmentation question comes with a hidden choice question. It sits right after the Guided Story question and is named Segmentation choices from <tag> (Guided Story), where <tag> is the Guided Story question's tag. It has the same choices. Respondents never see it. When the AI picks a choice, that choice is saved as this question's answer.
You don't see it as a card in the builder. You find it wherever you pick a question as a source:
- Display, skip, end-of-survey and branch logic — pick Segmentation choices from <tag> (Guided Story). Its choices are listed with Is Selected (Any), Is Selected (All), Is Not Selected (Any), Is Not Selected (All), Is Answered / Is Not Answered and the other choice operators. Rules on the same page as the Guided Story question work too.
- Quotas — build a quota on one of its choices (for example "stop after 50 respondents assigned Katniss").
- Carry forward, Loop & Merge and piping — use it like any choice question.
- Set Variable — a condition on its choices works like any choice condition.
Editing the choices. Change them in the Guided Story question's setup window and press Save. The hidden question follows: same choices, same order, same Exclusive flags. Recode values you set in Bulk Edit are kept. If the survey already has responses, renaming or removing a choice shows a warning: answers collected with the old wording no longer match it.
Copying, moving and deleting. Duplicate, move or delete the Guided Story question and the hidden question goes with it. Question randomization never separates them. The hidden question is stored right after the Guided Story question, after that question's own page break, and never gets a page break of its own; it is still answered and treated as displayed on the Guided Story question's page.
Older Guided Story questions. A Guided Story + Segmentation question made before this update has no hidden question yet. It keeps working as before. To use its choices in logic, quotas, carry forward, loops or piping, open its setup window and press Save once. Answers collected before that have no saved choice.
When the answer fits no choice, the AI saves your Exclusive choice (such as "None of these"). With no Exclusive choice, nothing is saved; a real choice is never saved just because it is first in the list. A new question's "None of These" choice is Exclusive already. The note under Choices says which applies, and How choices are picked explains every case.
When no choice is assigned (the microphone was denied, the recording failed, or nothing fit and no choice is Exclusive), the hidden question is left unanswered, so Is Not Answered is true for it, and Is Not Selected (Any) and Is Not Selected (All) are true for every choice. Carry forward and loops then read it like any unanswered choice question that was shown: Selected carries nothing (the follow-up is skipped) and a Selected loop runs zero times, while Not Selected carries every choice and a Not Selected loop runs once for each. If you would rather skip a Not Selected follow-up in that case, give it a display condition: the hidden question Is Answered. If the respondent goes back and records again, the new choice replaces the old one.
Rules on the Guided Story question itself. The Guided Story question stores the recording, not a choice, so a condition on it can only ask Is Answered, Is Not Answered, Is Displayed or Is Not Displayed; that is all the condition picker offers for it. Every rule about the chosen answer goes on Segmentation choices from <tag> (Guided Story). A choice rule left on the Guided Story question itself (from an earlier build of this feature) never runs: the condition row says so in red, and Publish names the rule and tells you to use <tag>_SEG instead.
Piping. Pipe the chosen answer from Segmentation choices from <tag> (Guided Story) (Selected Choices and the other choice options). The piping menu offers only Question Text for the Guided Story question itself, and a token that points at the Guided Story question's answer shows nothing (or its fallback) rather than the stored recording data. The same holds in exports and emails.
Publishing checks#
Publish stops if the hidden question was changed by hand (for example in the survey JSON editor) and no longer matches its Guided Story question. The message names the question and says: Open its setup window and press Save to repair it. Saving the setup window puts it right. If the Guided Story question itself was removed or changed by hand and its hidden question was left behind, Publish says so, and the hidden question has to be removed in the survey JSON editor.
Export and HM Admin#
- Your export gets the hidden question's column next to the Guided Story question's own column. It follows the usual single or multiple choice export rules: <tag>_SEG for Single Select (and for Multiple Select with Split multi-value fields into columns off), or <tag>_SEG_1, <tag>_SEG_2, … one per choice for Multiple Select with that option on (the default). It holds the choice the AI picked.
- In HM Admin the hidden question is a normal single- or multiple-choice question. Its answers, and the logic and quotas built on it, sync like any choice question, and it shows in Quant Analysis.
Good to know#
- The choice is assigned by AI (GPT-4o) from what the respondent said, not picked by the respondent. Write Choices precisely enough that open-ended language maps onto them reliably, and add an Exclusive "None of these" choice for answers that fit nothing.
- If the AI is unavailable, the choice is assigned by keyword: a choice is picked only when its full text appears in the answer. If none does, it is treated as "nothing fits". Short, distinctive choice texts keep that fallback useful.
Related guides#
- Question Types — overview and when-to-use table
- Guided Story — the base voice-only format
- StorySync AI Qual — all 9 StorySync formats
- Logic, Flow & Conditions — operators and condition builder
Was this page helpful?