Triage and Route Customer Feedback to the Right Teams Automatically

Automatically classify customer feedback with AI and route it to the right department, turning raw reviews into actionable insights instantly.

Workflow Overview

What is this?: This workflow leverages a Large Language Model (LLM) to automatically analyze and triage incoming customer reviews. It first classifies feedback as positive or negative, then performs a second-level analysis on negative comments to pinpoint specific issues like after-sales or transportation, ensuring a nuanced understanding of customer sentiment.
What problem does it solve?: Are you spending hours manually reading, tagging, and forwarding customer feedback to different departments? This workflow eliminates that bottleneck, preventing valuable insights from getting lost or delayed and freeing up your team to focus on resolving issues instead of sorting them.
What will you get?: You will automate the entire feedback triage process to save hours of manual work each week, ensure every piece of customer feedback is routed to the correct department in real-time for faster responses, and gain a structured, immediate understanding of customer sentiment to separate praise from critical operational issues.

Apps Included

How to Use

Prerequisites

An active Dify account (cloud or self-hosted).

An OpenAI API key (or another compatible LLM provider) configured in your Dify workspace.

The specific API endpoint URLs for your internal systems (e.g., marketing, after-sales, transportation department webhooks).

Setup Steps

Import and Open the Workflow

Start by importing the workflow file into your Dify workspace and opening it in the editor to view the graph.

Configure the LLM Classifiers

Verify that both 'Question Classifier' nodes are linked to your preferred LLM provider. No changes are needed if you have already set up a default OpenAI provider in Dify.

Update Positive Feedback Endpoint

Select the 'HTTP Request' node connected to the 'Positive review' branch. Replace the placeholder 'https://www.example.com' in the URL field with the actual API endpoint for your brand marketing department.

Update Negative Feedback Endpoints

Locate the two 'HTTP Request' nodes connected to the 'After-sales issues' and 'Transportation issue' branches. Update each node's URL with the corresponding API endpoint for your after-sales and transportation departments.

Publish and Test

Click the 'Publish' button in the top-right corner to save and activate the workflow. Test the application by providing different types of customer reviews to ensure they are routed to the correct destinations.

Pro Tips

To make the notifications more actionable, modify the HTTP Request nodes to pass the original customer review text in the request body to the target system.

Consider adding a new classification branch, such as 'Feature Request', to route product ideas directly to your product management team's backlog.

For enhanced reliability, implement a proper error-handling mechanism. You could add a branch that triggers a notification to a Slack channel if any of the HTTP requests fail.

Back

Information

Published date8/11/2025
Last updated8/11/2025

Platform

Dify