📚 Automate Survey Analysis with AI

Reading Time: 4 minutes

Hello AI Enthusiast,

Survey responses pile up faster than anyone can read them. Some are gold mines of actionable feedback, while others are just "Great!" or "Good product." Today, we're showing you how we built an AI automation that analyzes every survey response the moment it arrives, classifies its sentiment and usefulness, and delivers insights straight to Slack, so our team can focus on responses that actually matter.

The Problem

Customer feedback forms pile up in spreadsheets while you're busy with actual work. By the time someone reviews them, half the team has forgotten what you even asked. Meanwhile, genuinely useful feedback—the person who explained exactly why your pricing confuses them—gets buried under dozens of "nice product" comments.

You end up manually reading every response (nobody has time) or ignoring feedback entirely (missing valuable insights). You need a system that automatically spots which responses deserve attention and which are just noise.

How We Built It: A Step-by-Step Guide

We created an automation using Make that watches our Tally survey form, analyzes each new response with ChatGPT, and posts a formatted summary to Slack. Here's exactly how we did it.

Step 1: Set Up Your Survey Form Trigger

First, configure Make to watch for new survey responses. We used Tally, but this works with any form builder that integrates with Make (Google Forms, Typeform, etc.).

1. Create a new scenario in Make

2. Add your form platform as the trigger module (we used "Tally: Watch New Responses")

3. Connect your form and select which survey to monitor

4. Test the connection to confirm Make can access your responses

The automation now watches your form 24/7. Every time someone submits a response, Make captures it immediately and kicks off the analysis process.

Step 2: Configure ChatGPT to Analyze Survey Data

We send each survey response to ChatGPT with specific instructions about how to analyze it. The key is giving ChatGPT a clear framework for what makes feedback useful versus generic.

Add an OpenAI module (Generate a Completion) and use this prompt:

You are a product feedback analyst. Analyze the following survey responses and provide:

1. Sentiment classification (choose one):
   - Positive
   - Negative  
   - Neutral
   - No Answer

2. Usefulness determination:
   Consider feedback USEFUL if it contains specific information about:
   - Product features, pricing, or capabilities
   - Technical aspects or performance
   - Specific suggestions for improvement
   
   Consider feedback NOT USEFUL if it's just:
   - Single words like "Great," "Cool," "Bad"
   - Generic praise without specifics
   - Unclear or off-topic comments

3. Analysis summary:
   Provide a brief reason for your classification without repeating the customer's words. Examples:
   - "Customer appreciates new features but finds pricing too high"
   - "Vague positive feedback without actionable details"  
   - "Specific technical complaint about load times"

Format your response in JSON format with the following structure:
Sentiment: [classification]
Usefulness: [Useful/Not Useful]
Analysis: [your brief reason]

Survey Questions and Responses:
Rating: {{survey_rating}}
What they liked or not: {{detailed_feedback}}
What could improve: {{improvement_suggestions}}

Map your actual survey fields to the placeholders in the prompt. In our case, we mapped the satisfaction rating, detailed feedback, and improvement suggestions from Tally.

Don’t forget to select JSON Object as response format and select Yes to parse JSON directly in the OpenAI module. You can find these settings under “Advanced settings”

Step 3: Send Formatted Results to Slack

Now that ChatGPT has analyzed the response, send a clean notification to your team. Add a Slack module (Send a Message) and format it with the analysis results.

Here's our message template:

🔔 *New Survey Response*

⭐ Rating: {{survey_rating}}

📊 *AI Analysis:*
- Sentiment: {{chatgpt_sentiment}}
- Usefulness: {{chatgpt_usefulness}}  
- Analysis: {{chatgpt_analysis}}

💬 *Customer Comments:*
- Details about the rating: {{detailed_feedback}}
- Could Improve: {{improvement_suggestions}}

_Analyzed automatically by ChatGPT_

The Slack notification gives your team everything they need at a glance: the original response, ChatGPT's classification, and a quick explanation. No need to open spreadsheets or dig through raw data.

Step 4: Test Your Automation End-to-End

Before going live, test the complete flow with sample survey responses:

1. Submit a test response with detailed, specific feedback

2. Submit another with generic praise like "Good product"

3. Verify Make captures both responses

4. Check that ChatGPT correctly classifies one as useful and one as not useful

5. Confirm Slack receives properly formatted messages

If ChatGPT's classifications seem off, adjust your prompt to be more specific about what counts as useful feedback in your context. The classification accuracy improves significantly when you give clear examples of what you're looking for.

What Makes This Different From Manual Review

Every response gets analyzed immediately using the same criteria. When someone explains exactly why a feature isn't working, that shows up in Slack within seconds with a "Useful" tag. You can act on it while the feedback is fresh instead of discovering it three months later during a quarterly review.

Your Turn

Pick one feedback form you're currently using. Set up this automation following our steps, then test it with both useful and generic responses. Adjust the ChatGPT prompt until it consistently spots the difference between "Great product!" and actual actionable feedback.

Once it's working, you'll wonder how you ever manually reviewed surveys before.

Want to get even more practical? Explore hands-on AI learning with AI Academy:

• AI Academy Membership: Get 12 months of access to all our cohort-based programs, live webinars, on-demand courses, and tutorials.

• AI Agent Bootcamp: Accelerate processes and solve business problems by mastering prompts and building AI Agents, without coding.

• Corporate Training: Equip your team with the skills they need to unlock the potential of AI in your business.

• Practical Introduction to ChatGPT: A free course on using ChatGPT confidently, understanding its workings, and exploring its potential.

We'll be back with more AI tips soon!