Smart Surveys: AI-Powered Conversations

Smart Surveys are Revuloop's AI-powered conversational survey format. Instead of a fixed set of questions, Smart Surveys use AI to dynamically generate follow-up questions based on each respondent's answers, creating a personalized interview experience.

What Makes Smart Surveys Different

Traditional vs Smart Surveys

• Traditional: Fixed question list, same for every respondent
• Smart: AI adapts questions in real-time based on previous answers
• Result: Deeper insights, more engaged respondents, shorter surveys

Key Capabilities

• Adaptive Question Generation: AI creates relevant follow-up questions
• Sentiment Tracking: Real-time sentiment and engagement scoring
• Conversation Quality: Tracks depth and quality of each conversation
• Pattern Detection: Identifies themes and root causes across sessions

When to Use a Conversational Survey

Smart Surveys shine when the most valuable answer is one you could not have predicted, when you would normally schedule a customer interview but cannot interview everyone. Use a traditional survey when you need structured, comparable data across every respondent (ratings, rankings, choices). Use a conversational Smart Survey when you need the "why" behind behavior.

Teams typically use conversational surveys for:

• Churn and cancellation interviews: dig beneath "too expensive" to the real blocker
• NPS detractor "why" deep dives: turn a low score into a specific, fixable experience
• Post-support follow-up: confirm the fix actually worked after a ticket closes
• Product discovery and user research: first-round interviews at scale
• Employee exit and engagement interviews: honest patterns behind departures
• Win/loss analysis: what actually tipped the buyer's decision

Each of these has a full deep dive (recommended objective and opening question, a sample conversation, analytics guidance, and distribution tips) in the Conversational Survey Use Cases article.

Creating a Smart Survey

1. Click "Create New Survey"
2. Select "Smart Survey" mode
3. Configure your survey objective and initial context
4. Set the maximum number of questions per session
5. Preview and test the conversation flow
6. Launch your Smart Survey

Configuration Options

• Survey Objective: Describe what you want to learn. This is the single biggest quality lever: the AI uses it to decide what to probe. Be specific about the decision the answers should inform (see the objective examples above).
• Opening Question: The first question every respondent sees. Make it easy to answer and broad enough to branch from; the AI takes it from there.
• Initial Question Type: The format of the opening question (text, rating, NPS, sentiment, scale, slider, or date).
• Max Questions: Caps questions per session (default: 10). We recommend 8-10; shorter caps cost less but may end conversations before the AI reaches root causes.
• Tone: Conversation style (default: professional). Match it to your audience, casual for consumer products, professional for B2B.
• Target Audience and Industry: Optional context that helps the AI use the right vocabulary.
• Question Types: Restrict which formats the AI may use for follow-ups.
• Adaptive Logic and Emotional Analysis: Toggles for answer-driven branching and real-time sentiment tracking (both on by default).

Writing a Good Objective

• Name the decision: "Decide whether to build X" beats "learn about X"
• List the dimensions: pricing, onboarding, support; the AI will cover each
• State who is answering: "customers who cancelled in the last 30 days"
• Avoid leading: describe what to learn, not what you expect to hear

How Sessions Work

Each respondent gets a unique conversation session:

1. Session Start: Respondent opens the survey, a session is created
2. AI Generates Questions: Based on context and previous answers
3. Respondent Answers: Free-text responses analyzed in real-time
4. Follow-up Generated: AI creates the next question based on the answer
5. Session Complete: After max questions reached or conversation naturally ends

Session Data Tracked

• Full conversation transcript
• Per-question sentiment scores
• Engagement level over time
• Conversation quality score
• Emotional journey analysis

Token Costs

Smart Surveys use tokens per interaction:

Typical Session Cost: A standard 8-question session costs approximately 41 tokens.

Optimizing Token Usage

• Set reasonable question limits: 8 questions balances insight vs cost
• Use fallback questions: Pre-defined questions cost less than AI generation
• Monitor session costs: Track per-session token usage in analytics
• Preview mode: Test without using tokens

Smart Survey Analytics

After collecting responses, access Smart Survey analytics:

• Session Overview: Total sessions, completion rates, and quality scores across all conversations
• Conversation Quality Score: A 0-1 score per session measuring depth and engagement: how substantive the answers were, not just whether questions were answered
• Sentiment Journey: How each respondent's sentiment moved across the conversation (improving, declining, stable, or volatile), plus trends across all sessions
• Conversation Insights: Common themes, patterns, and root causes detected across sessions
• Session Details: Full transcripts with per-question sentiment, quality, and engagement scores

Smart Survey data lives in its own analytics view (Analytics → Smart) because conversations do not aggregate like fixed questions. Transcripts are also available via the API for teams that pipe research into other tools.

Good to Know

• Distribution: Smart Surveys share via the same channels as regular surveys: link, email, QR, and embed.
• Skip logic and scoring do not apply: the AI itself decides what to ask next, so branching rules and point values are features of traditional surveys only.
• Privacy: Free-text answers are processed by AI for question generation and sentiment analysis. Avoid asking respondents for sensitive personal data in your objective or opening question.
• Respondent experience: Respondents see one question at a time in a chat-style interface and can finish early; completed answers are kept.

Plan Limits

Smart Surveys are available on every plan, including Free. The Free plan can

have 1 active Smart Survey at a time; Starter and above have no cap. Token

allowances differ by plan, and every session consumes tokens.

Minimum Requirements

• Minimum tokens to start: 20 tokens
• Low token warning: Displayed at 100 tokens remaining

Troubleshooting

Follow-up questions feel generic

Sharpen the Survey Objective. Generic objectives ("get feedback") produce generic probing. Name the dimensions you care about and the decision you are making.

Sessions end too early

Raise Max Questions, and check that your opening question invites elaboration; yes/no openers give the AI little to branch from.

Sessions are not starting

Check your token balance: sessions need a minimum of 20 tokens to start. Top up under Settings → Billing.

Costs are higher than expected

Lower Max Questions, and review the per-session token usage in Smart Analytics to find unusually long conversations.