Guided Surveys – User Manual

What the Platform Does

Guided Surveys helps you collect rich, qualitative feedback through short voice conversations instead of long, boring forms. You create a conversation brief that tells the system what you want to learn, share a unique link with customers or employees, and an AI assistant interviews them on your behalf. The platform records the audio, generates a transcript, and produces structured summaries so your team can review conversations quickly and confidently.

Quick Start Checklist

• Sign in with your Google account so we can create or connect your workspace.
• Create a project to group related briefs and conversations (for example, "Onboarding Research" or "Q2 NPS").
• Add a brief, either from a template or from scratch, and fill in greeting, goodbye, topics, and purpose.
• Choose a tone and voice for the assistant so it sounds like your brand.
• Decide whether you want an access code, IP restrictions, or contact info collection for this brief.
• Set the brief to active so it can accept conversations.
• Copy the /conversation/{slug} link and share it with participants by email, chat, or QR code.
• Watch new conversations appear in the dashboard and review transcripts, audio, and summaries.

Roles

There are two main types of people using the product. Business users are the people on your team who sign in, set things up, and review results. They sign in securely with Google, create projects and briefs, share conversation links, and use the dashboard to read transcripts, listen to recordings, and pull insights. Participants are the customers or employees you invite into conversations. They never need an account; they simply open the link you send, grant microphone access, and talk naturally with the assistant.

The Assistant (Your AI Interviewer)

The assistant is the AI voice that runs each conversation. It follows your brief closely: it reads your greeting, guides the discussion through the topics you define, and uses your goodbye message to wrap things up. The assistant asks one clear question at a time, listens without interrupting, and only uses follow‑up questions when it needs more detail. It avoids repeating questions and gently nudges people back on track if they drift away from your topics. From the participant’s perspective it feels like a calm, focused interviewer that is clearly “working for” your team.

Glossary of Core Terms

Business user: A member of your team who signs in with Google, creates projects and briefs, shares links, and reviews results in the dashboard.

Participant: A customer, employee, or other end user who joins a conversation via a shared link and talks with the assistant without needing an account.

Assistant: The AI voice that runs the interview, asks questions based on your brief, and follows your greeting, topics, purpose, and goodbye instructions.

Project: A container in the dashboard used to group related briefs and conversations, such as a specific research study, product area, or customer account.

Brief: The conversation blueprint that defines the greeting, goodbye, topics, purpose, tone, voice, and access settings the assistant should use for a specific use case.

Conversation: One recorded interview between the assistant and a single participant, including audio, transcript, analysis, and any captured contact details.

Conversation link / slug: The unique URL (for example /conversation/abc123xyz) you share with participants that points to a specific brief’s conversation page.

Access code: An optional passcode participants must enter before starting a conversation, used to keep briefs private or restricted to certain audiences.

Topics: Short, natural‑language descriptions of what you want to talk about; the assistant turns these into concrete questions during the interview.

Purpose: A description of the business goal behind the brief, which guides both how the assistant steers the conversation and how summaries and key points are generated.

Tone and voice: Settings that control how the assistant sounds; tone affects style (friendly, professional, empathetic), and voice selects the actual AI voice (such as alloy or nova).

Temperature: A creativity setting for the model; lower values keep questions more focused and consistent, while higher values allow more exploratory follow‑ups.

Contact info modes: The options that control whether and how you collect participant contact details: notrequested (never asked), requested (asked but optional), and required (must be provided to continue).

Transcript and summary: The text version of the conversation and its condensed overview, which you use to scan what was said and identify key insights without listening to the full audio.

Getting Started as a Business User

To get started, sign in with your Google account. If it is your first time, an account is created for you automatically. Once you are in the dashboard, create a project to group related briefs and conversations, such as a specific product line, research study, or client.

Next, create a brief in the Briefs section. You can start from a template (for example, an NPS survey or general customer feedback) or build one from scratch. In the brief, you configure the greeting and goodbye messages, the topics you want the assistant to explore, and the overall purpose so the analysis knows what “success” looks like. You also pick a conversational tone such as friendly, professional, or empathetic, and choose a voice like alloy, echo, fable, onyx, nova, or shimmer so the audio matches your brand.

Brief Settings and What They Mean

Greeting and goodbye messages control how the conversation starts and ends. The greeting should set expectations about length and reassure people their feedback matters, while the goodbye should thank them and explain how their input will be used. Topics are short, natural‑language descriptions of the areas you care about (for example “onboarding experience”, “feature X workflow”, or “reasons for churn”). The assistant uses these to decide what to ask about, rather than following a rigid script.

The purpose field tells the system why you are running this interview, such as prioritizing feature work or understanding satisfaction drivers. This guidance affects both how the assistant steers the conversation and how summaries and key points are generated afterward. Temperature controls how exploratory the assistant can be: values around 0.5 keep it focused and consistent, while higher values make it a bit more creative and exploratory.

Access codes let you gate sensitive briefs. When an access code is set, participants must enter that code on the landing page before they can join the conversation, which is useful for internal interviews or invite‑only studies. IP restrictions go one step further and allow only specific networks or locations (for example, your company’s VPN or office range) to access the link.

The collect contact info setting controls how the assistant and page handle personal details. With notrequested, the system never asks for contact information and conversations remain fully anonymous from the participant’s point of view. With requested, the assistant will politely ask for contact details near the end, but participants can decline and still complete the conversation. With required, participants must provide at least their name and either an email address or phone number before they can continue; the form validates basic email and phone formats so you get usable data.

What Participants Experience

When someone clicks your link, they see a simple, branded landing page that shows the brief name, description, and any logo or background color you configured. If the brief is not active (for example, it is still a draft or has been paused), they see a clear message explaining that the conversation is not currently available. If you set an access code, they are asked to enter it before they can proceed.

Before starting, participants can expand a privacy and consent section that explains what will be recorded, how their data will be used, and how to end the conversation at any time. If contact information is required, they fill in their name and at least one contact method (email or phone) on a short form. Once they are ready, they grant microphone permission and press Start Conversation. The assistant then plays your greeting, asks questions one at a time based on your topics, and listens while they answer in their own words. When the assistant has heard enough, it announces the end of the interview using your goodbye message and the page shows a thank‑you state.

Reviewing Results in the Dashboard

After conversations complete, you review everything from the dashboard. The Conversations area shows a list of interviews with basic details like project, brief, timestamps, and completion status, making it easy to see overall progress at a glance. You can scan how many responses each brief has received and drill down into specific conversations that look interesting or unusual.

Opening a conversation shows the full transcript with clear speaker labels, along with the audio recording saved to S3 so you can replay the exact wording and tone. The page also surfaces a short summary, key points, and quality or sentiment indicators so you can understand the gist without reading every line. If you configured contact collection, you will see the participant’s shared details alongside the transcript, which is helpful for follow‑up research or outreach. You can download individual transcripts and audio files as needed and use project‑level views to compare themes across many conversations over time.

Tips for High‑Quality Briefs

A strong brief makes the assistant feel thoughtful, respectful, and on‑brand. Keep your greeting to one or two sentences that set expectations about length (for example, “about three minutes”) and reassure participants that their feedback will be used to improve the experience. When listing topics, think in terms of themes rather than a long questionnaire; the assistant handles turning those themes into conversational questions. Including both positive and critical angles (for example, “frustrations” as well as “favorite moments”) leads to more balanced insights.

If your goal is focused research, keep the temperature near 0.5 so the assistant stays consistent between participants. If you are exploring new ideas, consider nudging it slightly higher so it feels more exploratory and creative in follow‑ups. Reserve the required contact info mode for studies where follow‑up is truly essential, such as high‑touch customer success or in‑depth UX interviews. For most surveys, requested strikes a good balance between response rates and the ability to follow up with especially valuable participants.

Troubleshooting

If a participant cannot start a conversation, first confirm that the brief is active in the dashboard and that they are using the correct access code if one is configured. Briefs in draft, paused, or archived states will show an informational message instead of the normal start flow. If someone reports microphone issues, have them allow microphone access in their browser settings and refresh the page; without permission, the assistant cannot hear them.

When transcripts update but no audio plays, the browser may have blocked autoplay. In those cases, ask the participant to click anywhere on the page or use the play or unmute controls so audio can start. If a conversation gets stuck or ends unexpectedly, the system will attempt to reconnect, but it is usually easiest to refresh the page and start a new conversation; finished conversations are treated as complete records and are not resumed so your analytics stay clean. If you continue to see issues, note the brief link and approximate time of the problem so your team can investigate logs and environment settings.