Products
Contact Support Additional Learning Monthly Product Updates
Menu

Forsta Survey Designer Support

All the topics, resources needed for Forsta Survey Designer.

Dynamic Conversations

  • Last updated:

Articles in this section

See more
In this article

    Prerequisites

    • This option is only available on Azure sites (not available on EURO or US Rackspace).

    • This option is only available if you are using Forsta's JavaScript engine (controlled from survey settings).

    • This option is only available if you are using a Responsive Rendering V2 layout (e.g. “Public Responsive Whitespace Layout V2”).

    • This feature requires the new Linux-based engine.

    • Access to this feature is controlled via a feature toggle.

    1: Overview

    Dynamic Conversations (DC) introduces a new way to design adaptive survey experiences by allowing AI to determine which questions to ask each respondent.

    Instead of following a fixed path, a Dynamic Conversations container groups a set of questions and uses generative AI to select the most relevant ones in real time. Based on the provided context and previous responses, the AI dynamically chooses which question to ask next, up to a maximum number defined by the survey author.

    This enables more personalized, efficient, and engaging surveys while ensuring respondents are only asked the most relevant questions. Dynamic Conversations containers can also include benchmarked questions, allowing you to ask a smaller, more targeted subset while still supporting benchmarking needs.

    2: Features

    2.1: AI-Driven Question Selection

    • Dynamic Conversations evaluates all questions within the container and selects the most relevant one to ask next based on prompt context and previous respondent answers.

    • This process continues until the maximum number of questions is reached or no further suitable questions remain.

    2.2: Adaptive Survey Paths Without Manual Logic

    Dynamic Conversations removes the need for complex routing or branching logic. Survey authors define the pool of questions and context, and the AI determines the optimal path for each respondent.

    2.3: Optional Personalisation

    Dynamic Conversations can enhance the respondent experience with:

    • A short personalized introduction before each selected question

    • A summary at the end of the DC section

    These features help create a more conversational and engaging interaction.

    3: Setting Up Dynamic Conversations

    You can configure Dynamic Conversations within Forsta Survey Designer by following these steps:

    3.1: Add the Dynamic Conversations Container

    1. Navigate to the Questions menu

    2. Select "Dynamic Conversation"

    3.2: Add Questions to the Container

    Add supported question types into the container. These will form the pool of questions that the AI evaluates and selects from.

    4: Configuration

    In the Dynamic Conversations configuration menu, you can:

    • Define Maximum Number of Questions

      Set how many questions the AI can select and present.

    • Provide Prompt Context

      Supply contextual information to guide the AI in determining relevance and shaping the overall conversation.

      This field supports script piping, allowing you to include prior answers or respondent attributes.

    • Add AI Disclaimer Text

      Inform respondents that AI is being used in the experience.

    • Enable Question Personalisation (Optional)

      Add a short, AI-generated lead-in before each selected question.

    • Enable Summary Personalisation (Optional)

      Generate a summary of the interaction at the end of the DC section.

    Best Practice:

    Clear, concise context significantly improves question selection. Write context as if explaining the survey goal to someone unfamiliar with it.

    Here are some examples:

    Example 1: Generic VoC Survey

    Prompt context:

    The respondent is completing a relationship survey about their overall experience with our company. In a previous question, they provided the following feedback: ^f('LTR_comment')^. Prioritise questions that help clarify what influenced their experience, including specific interactions, products, or services, and whether their experience was positive or negative.

    Example 2: Healthcare – Billing Experience

    Prompt context:

    The respondent is a patient who recently received care and has provided the following feedback about their billing experience: ^f('billing_feedback')^. Prioritise questions from the available set to better understand their billing experience. Focus on identifying issues related to clarity of charges, insurance processing, payment experience, or unexpected costs, and prioritise questions that will help uncover specific problems or areas for improvement.

    Example 3: Automotive – Post-Service Experience

    Prompt context:

    The respondent recently had their vehicle serviced and provided the following feedback about their experience: ^f('service_feedback')^. Prioritise questions that help understand key aspects such as service quality, timeliness, staff interactions, communication, and overall satisfaction, based on the details in their response.

    And some poorly formed examples:

    • Too vague: "Ask relevant questions about the respondent’s experience."

    • Trying to control order: "First ask about staff, then ask about pricing, then ask about overall satisfaction."

    • Lacking respondent data: "This survey is about billing. Ask good follow-up questions."

    5: Supported Question Types

    The following question/node types can be added inside a Dynamic Conversations container:

    • Text

    • Text list

    • Open Assist

    • Informational text

    • Date

    • Single choice

    • Table lookup / hierarchy

    • Multi-choice

    • Numeric

    • Numeric list

    • Ranking

    • Grid

    Custom questions based on these supported types can also be used.

    6: Engine Behaviour

    • Respondents cannot use the Back button during the DC experience

    • Once completed, respondents cannot re-enter the DC container

    • Redirecting into a DC container will restart the experience and trigger a new AI-driven selection, or continue if already started, or go to the next question if already completed

    • Question-level and answer-level masking is supported

    • All DC questions are asked one by one, one question per page

    • Questions inside the container should not depend on each other (as not all will be asked)

    • Avoid piping answers from other DC questions, as they may not have been answered

    • Survey authors do not have explicit control over selection order (e.g. cannot force question sequences)

    7: Data and Variable Outputs

    Each Dynamic Conversations container generates the following variables (where [qid] is the container name):

    • [qid]_status

      Status of the DC section:

      empty (not started)

      incomplete (started but not finished)

      complete (finished)

      error (an error occurred)

    • [qid]_last

      The QID of the last selected question

    • [qid]_summary

      The AI-generated summary shown at the end of the experience

    • [qid]_dcc

      A comma-separated list of selected question QIDs

      [qid]_dcp

      An open text list containing the AI-generated prompt shown before each selected question

    These variables are available in survey data, Hubs, and downstream analytics.

    8: Testing Behaviour

    In External Test, Dynamic Conversations behaves the same as in production (AI-driven selection)

    In Survey Designer Quick Test, AI is not used and random questions are selected instead

    9: Responsible AI and Transparency

    We place strong emphasis on responsible and transparent use of AI in Dynamic Conversations.

    • Respondents can be informed when AI is used via disclaimer text

    • AI-driven decisions are guided by clearly defined context

    • All selected questions and prompts are captured for full visibility and auditability

    This ensures trust, traceability, and control over how AI is used in your survey experience.

    10: Security Information

    The OpenAI service is a part of Forsta’s private Azure instance, and Microsoft is listed as sub processor here: Forsta Sub-Processor List.

    • No data is stored elsewhere and only comments data is submitted. We do not include any metadata.

    • Your prompts (inputs) and completions (outputs), your embeddings, and your training data:

    • are NOT available to other customers.

    • are NOT available to OpenAI.

    • are NOT used to improve OpenAI models.

    • are NOT used to improve any Microsoft or 3rd party products or services.

    • are NOT used for automatically improving Azure OpenAI models for your use in your resource (The models are stateless, unless you explicitly fine-tune models with your training data).

    • Your fine-tuned Azure OpenAI models are available exclusively for your use.

    • The Azure OpenAI Service is fully controlled by Microsoft; Microsoft hosts the OpenAI models in Microsoft’s Azure environment and the Service does NOT interact with any services operated by OpenAI (e.g. ChatGPT, or the OpenAI API).