Turn raw survey answers in Google Sheets into clear insights your team can use. The flow groups similar answers, adds sentiment, and writes results back to a new insights sheet. It suits product and marketing teams that collect open text feedback and want patterns, not long walls of text.

Data loads from your sheet, then answers are paired with their questions and participants. Each answer is turned into a vector using OpenAI and stored in Qdrant with metadata. The flow works question by question, finds similar answers in Qdrant, and runs a K means clustering step. Small clusters are removed, then the remaining groups are summarized by an AI chat model and scored for sentiment. Results are appended to a new sheet so you can share and filter insights in seconds.

Setup needs Google Sheets, OpenAI, and a Qdrant endpoint. You can trigger it by hand in n8n or call it from another workflow. Teams can expect faster analysis with consistent scoring and traceable counts per group. Common uses include UX research, NPS comments, onboarding feedback, and open response surveys from events and beta tests.

• Imports survey rows and headers from Google Sheets and prepares data for analysis
• Builds question and answer pairs and attaches participant and survey metadata
• Creates embeddings with OpenAI and stores vectors in Qdrant
• Filters and retrieves answers by question directly from Qdrant using metadata
• Runs a K means clustering step to find groups of similar answers
• Skips empty results and removes clusters with fewer than three answers
• Fetches full payloads for cluster members to keep context and identity links
• Uses an AI chat model to summarize each group and assign sentiment
• Creates a new insights sheet and appends results for easy sharing

• Reduce manual review from hours to minutes by grouping similar answers automatically
• Keep counts and participant context so insights are traceable and reliable
• Handle thousands of responses by searching and clustering in Qdrant
• Improve insight quality by summarizing each cluster with sentiment scores
• Connect Google Sheets, Qdrant and OpenAI in one flow without copy and paste
• Protect progress with per question processing so partial runs still add value

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Google Sheets, OpenAI and Qdrant. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, create a Google Sheets OAuth credential or double click any Google Sheets node and choose Create new credential, then follow the prompts to grant access to your spreadsheet.
4. Open the Get Sheet Details and Get Survey Results nodes and select your spreadsheet and the correct sheet tab. Confirm the sheet loads sample rows when you click Fetch.
5. Open the Create Insights Sheet node and select the same spreadsheet. It will create a new tab for insights when the flow runs.
6. Create a Qdrant credential: double click a Qdrant node, choose Create new credential, enter your base URL for Qdrant and API key if your server requires it, then test the connection.
7. Check the Set Variables node and confirm the collection name matches your Qdrant setup. Create the collection in Qdrant if needed or allow the vector store node to create it when supported.
8. Create an OpenAI credential: double click the Embeddings or Chat node, choose Create new credential, paste your API key from the OpenAI account page, and save. Confirm models text embedding 3 small and gpt 4o mini are available in your account.
9. If triggering from another workflow, open the Execute Workflow Trigger node and note the expected input fields for sheetID and sheetName. Make sure your calling workflow sends these values.
10. Run a manual test. Start the workflow in n8n and watch the execution. You should see vectors created in Qdrant and a new insights sheet added to your spreadsheet.
11. Review outputs. Look for question, group summary, sentiment, and counts. If no clusters appear for a question, the flow will skip writing for that item which is expected.
12. Troubleshoot common issues: verify Google Sheets access if export fails, confirm Qdrant URL or API key if no points are returned, adjust batch size if OpenAI rate limits, and clear the insights tab before rerunning to avoid duplicate rows.