## Introduction
For product teams, gathering and analyzing customer feedback is crucial for improving features, UX, and overall satisfaction. However, manually reviewing large volumes of feedback can be time-consuming and error-prone. Automating the process of categorizing or tagging feedback based on sentiment helps teams rapidly identify critical issues, prioritize feature requests, and measure customer happiness.
In this tutorial, we’ll create an automated workflow using n8n — an open-source node-based workflow automation tool — to auto-tag customer feedback by sentiment. This streamlines how product managers and customer success teams handle feedback from multiple sources, such as Zendesk, Google Forms, or Slack.
## What Problem Does This Solve?
– **Manual triage of feedback is slow and inconsistent.**
– **Sentiment-based categorization helps prioritize negative feedback for urgent response.**
– **Automated tagging enables segmentation for analytics or notifying relevant teams.**
## Who Benefits?
– Product Managers looking to understand customer sentiment trends.
– Customer Support for quicker escalation of issues.
– Marketing and UX teams leveraging sentiment data for improvements.
## Tools and Services Integrated
– **n8n:** Core automation platform.
– **Feedback Data Source:** Example using Google Sheets or incoming webhook.
– **Sentiment Analysis API:** We’ll use external API such as Azure Text Analytics or Google Cloud Natural Language API.
– **Slack:** To notify product team of negative feedback.
– **Google Sheets:** To store tagged feedback for reporting.
—
# Step-by-Step Tutorial: Building the Sentiment Auto-Tagging Workflow in n8n
### Prerequisites
– n8n instance running (cloud or self-hosted).
– API key for sentiment analysis service (Azure, Google NLP, or alternative).
– Access to your feedback data, e.g., Google Sheets or webhook input.
– Slack workspace for notifications.
————
### Step 1: Define Trigger Node
We begin by fetching new feedback. Depending on your data source:
– **Google Sheets Trigger:** Use the “Google Sheets Trigger” node to watch for new rows added.
– **Webhook:** Use the “Webhook” node to receive feedback submitted through forms or apps in real-time.
**Example:** Add a Webhook node named `Receive Feedback` configured with POST method. The payload contains fields like:
“`json
{
“feedback_id”: “123”,
“user_email”: “customer@example.com”,
“message”: “I love the new update, but the app crashes sometimes.”
}
“`
### Step 2: Extract Feedback Text
Use the **Set** or **Function** node to extract the relevant text from the payload.
Example: Extract the `message` field to process.
### Step 3: Connect to Sentiment Analysis API
Use the **HTTP Request** node to send the feedback text to a sentiment analysis API.
– **API Options:**
– Azure Text Analytics `/sentiment` endpoint
– Google Cloud Natural Language API `/analyzeSentiment`
**Configuration Example for Azure**:
– Method: POST
– URL: `https://
– Headers:
– `Ocp-Apim-Subscription-Key`: your_api_key
– `Content-Type`: application/json
– Body (JSON):
“`json
{
“documents”: [
{
“id”: “1”,
“language”: “en”,
“text”: “{{$json[“message”]}}”
}
]
}
“`
**Handle Response:** Parse response to extract sentiment label (e.g., positive, neutral, negative) and confidence scores.
### Step 4: Assign Tags Based on Sentiment
Add a **Function** node to evaluate the sentiment and assign corresponding tags.
Example logic:
“`javascript
const sentiment = $json[“documents”][0][“sentiment”];
let tag = “neutral”;
if(sentiment === “positive”) tag = “Positive Feedback”;
else if(sentiment === “negative”) tag = “Negative Feedback”;
else tag = “Neutral Feedback”;
return [{ json: { tag } }];
“`
### Step 5: Save Tagged Feedback
If using Google Sheets:
– Add a **Google Sheets** node configured to append a new row with columns such as `feedback_id`, `user_email`, `message`, `sentiment_tag`.
If using CRM or database, configure the respective node to update feedback records.
### Step 6: Notify Product Team on Negative Feedback
To ensure urgent issues are caught early, add a condition branch:
– Add a **IF** node after the Function node to check if the tag is “Negative Feedback”.
– If YES, use a **Slack** node to send a message to a channel or user.
Slack message example:
“`
New Negative Feedback received:
User: {{$json[“user_email”]}}
Message: {{$json[“message”]}}
Sentiment Score: {{$json[“documents”][0][“confidenceScores”][“negative”]}}
“`
### Step 7: Error Handling and Retries
– Use **Error Trigger** node in n8n to catch any node failures.
– Implement retry logic in HTTP Request node settings (e.g., retry on 500 errors).
– Validate API responses to handle unexpected data.
### Step 8: Test Your Workflow
– Send sample feedback through the trigger.
– Confirm sentiment tag is assigned accurately.
– Ensure Google Sheets or storage is updated.
– Verify Slack notification fires for negative feedback.
### Step 9: Deploy and Schedule
– For sheet triggers: Set polling interval.
– For webhook triggers: Ensure URLs are exposed and integrated with feedback forms.
### Step 10: Scaling & Adaptation Tips
– Support multiple languages by configuring sentiment API accordingly.
– Add more nuanced tags, e.g. separating “compliments” from “feature requests.”
– Integrate with ticketing systems like Zendesk to auto-create tickets for negative feedback.
– Store enriched data in BI tools or dashboards for trend analysis.
– Handle large volume by batching requests to the API if supported.
## Common Errors & Troubleshooting
– **API Quotas:** Monitor sentiment API usage to avoid hitting limits.
– **Incorrect Sentiment:** Tune or switch APIs if sentiment accuracy is insufficient.
– **Invalid Data:** Ensure incoming feedback contains valid text fields.
– **Slack Permissions:** Confirm Slack bot has message posting rights.
## Summary
In this guide, we built a robust, scalable n8n workflow that automatically tags incoming customer feedback by sentiment. This empowers product teams to triage feedback effectively, react quickly to negative inputs, and generate organized data for analytics.
Automating sentiment analysis integrates smoothly with existing tools like Google Sheets and Slack, creating a powerful loop of feedback gathering, analysis, and response.
## Bonus Tip
Consider enriching your workflow by incorporating additional NLP features such as keyword extraction or intent recognition. This enables multi-dimensional tagging, such as identifying feedback about performance issues vs feature requests, further improving your team’s prioritization capabilities.
—
Now that you have the foundation, you can tailor the workflow to your product’s unique feedback channels and grow your automation ecosystem with n8n.