> ## Documentation Index
> Fetch the complete documentation index at: https://docs.caplena.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Create a survey

> Start a survey from a short brief. The survey agent asks a few follow-up questions and drafts a complete questionnaire you can refine.

> [!NOTE]
>
> Surveys is in **closed beta** and not yet available to all customers. [Join the waitlist](https://caplena.com/en/ai-survey-tool) to get access. See [Surveys overview](https://docs.caplena.com/surveys/overview.md) for current limitations.

## Start a new survey

**Step 1**

Open a project and go to **Collect** → **Surveys** in the sidebar.

**Step 2**

Click **New survey**. Alternatively, on the **Data sources** page click **Add data source** and then **Create survey** in the **Create a survey** box at the top (highlighted below).

![The "Create a survey" option in the Add data source dialog](https://mintcdn.com/caplena-32172960/8VLjyD540Rp1inTq/images/surveys/add-data-source.png?fit=max\&auto=format\&n=8VLjyD540Rp1inTq\&q=85\&s=231503c001913ab71cf1ba4e42aa37ba)

Caplena creates a draft survey and opens the **brief**. In a new project without data, **Create a survey** is offered right away when you open the project.

> [!TIP]
>
> A survey writes its responses into the project it belongs to. Create the survey in a new, empty project if you want its data kept separate.

## 1. Write the brief

The brief screen asks **"What are we creating today?"**. Describe what you want to learn, in your own words. A hunch, a business question or a metric that's off is enough. For example:

> Understand why customers who bought in our online shop in the last 3 months didn't buy again, and what would bring them back.

Press **Enter** to continue. The brief is saved as the **survey goal** and guides everything the agent does afterwards.

![The brief screen](https://mintcdn.com/caplena-32172960/8VLjyD540Rp1inTq/images/surveys/brief.png?fit=max\&auto=format\&n=8VLjyD540Rp1inTq\&q=85\&s=e34416ff94f360cd7315ead35681da21)

> [!TIP]
>
> **Already have a questionnaire?** Paste the full questionnaire into the brief. The agent builds it exactly as written instead of designing its own, and tells you about anything it had to adapt (for example, CES questions always use a 1–7 scale).

If you'd rather build everything yourself, click **Skip the brief and start from a blank survey**.

## 2. Answer the follow-up questions

Next, the agent asks **a few follow-ups before it starts building**. These are generated from your brief, for example about the target audience or which metric matters most. For each question, pick an option, write your own under **Other...**, or choose **Decide for me**.

You're always asked:

* **Which tone and wording should the questions use?**
  * **Professional**: formal address with respectful, business-appropriate phrasing.
  * **Friendly**: informal address with a relaxed, conversational tone.
  * **Playful**: informal, casual address with upbeat phrasing and emojis ✨
* **Anything else we should know?** (optional)

![Follow-up questions before the survey is built](https://mintcdn.com/caplena-32172960/8VLjyD540Rp1inTq/images/surveys/refinement.png?fit=max\&auto=format\&n=8VLjyD540Rp1inTq\&q=85\&s=d231e44e3df9576c6b880630790926b6)

Click **Build my survey**.

## 3. Let the agent draft the survey

The survey builder opens and the agent drafts your survey. A typical first draft contains:

* 5 to 12 questions, with an introductory message if it helps.
* One headline metric, such as NPS, CSAT, CES, a star rating or a Likert question.
* An open-text "why" question with [AI follow-up questions (probing)](https://docs.caplena.com/surveys/logic-probing-and-parameters.md#ai-follow-up-questions-probing) turned on.
* A matrix when several items are rated on the same scale.
* A survey name, a start screen title and a closing message.

When it's done, the agent summarizes what it built and points out anything worth reconsidering, for example if the survey is longer than the time you asked for.

## 4. Refine with the agent

The **Agent** panel on the left stays available while you edit. Ask it for changes in plain language, for example:

* "Add a question about delivery speed after the likelihood question."
* "Only show the question about what would bring them back to people who answered 1–3."
* "Make the tone more casual and shorten all questions."
* "Turn off voice mode."
* "Make the last question required."

![Asking the agent to make a question required, and the result in the editor](https://mintcdn.com/caplena-32172960/8VLjyD540Rp1inTq/images/surveys/agent-refine.png?fit=max\&auto=format\&n=8VLjyD540Rp1inTq\&q=85\&s=1f6d66a19b22c1c1ca7f6356565364cb)

The agent edits the draft directly and shows an **Updating survey** indicator while it works. You can stop it at any time. Click **What could I ask?** for suggestions.

The agent can change questions, logic and survey settings, and fix individual translations. It **can't add or remove languages**; do that in the [Translations](https://docs.caplena.com/surveys/languages-and-translations.md) view. The agent only ever changes the **draft**. It never publishes, archives or deletes a survey.

> [!TIP]
>
> You can also start a survey from the project [Agent](https://docs.caplena.com/insight-and-reports/insight-agent/insight-agent.md): ask it to create a survey, for example one that digs deeper into a finding, and it drafts one in your project.

To clear the chat history, click **Reset conversation**. Your questions stay as they are.

## Next steps

> [card] [Edit your survey](https://docs.caplena.com/surveys/build-your-survey.md)
>
> Fine-tune questions, settings, branding and the closing screen.

> [card] [Question types](https://docs.caplena.com/surveys/question-types.md)
>
> All question types and their settings.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.