Sentiment analysis for portal feedback

Sentiment analysis helps organizations understand how users feel about their experience on your Power Pages portal. By capturing and analyzing user feedback using tools like AI Builder, Power Automate, and Dataverse, you can make data-driven decisions to enhance your portal’s UX, resolve pain points faster, and prioritize features.

---

Why Sentiment Analysis Matters

When a user submits feedback—such as through a contact form, satisfaction survey, or support interaction—sentiment analysis can determine whether the feedback is positive, neutral, or negative. This analysis helps to:

• Automatically route negative feedback to support
• Trigger alerts for critical comments
• Score customer satisfaction
• Measure user sentiment trends over time
• Improve content and service design

---

Key Components Used

Component	Purpose
Power Pages	Collects user feedback through embedded forms
Dataverse	Stores feedback records with user metadata
Power Automate	Triggers flows when new feedback is submitted
AI Builder	Performs sentiment analysis on the feedback text
Power BI (optional)	Visualizes feedback sentiment over time

---

Step-by-Step Implementation

---

Step 1: Create a Feedback Form in Power Pages

1. Open Power Pages Studio
2. Add a Custom Page named Feedback
3. Insert a Basic Form connected to a custom Dataverse table, e.g., Portal Feedback
4. Include columns like:
   • Feedback Text (multiline text)
   • Submitted By (lookup or contact)
   • Submission Date (auto-filled)
   • Sentiment Score (calculated later)
   • Sentiment Category (Positive / Neutral / Negative)

---

Step 2: Create the Custom Table in Dataverse

In Power Apps:

1. Navigate to Tables > New Table
2. Table Name: Portal Feedback
3. Add columns:
   • FeedbackText (Text, Multiline)
   • SentimentScore (Number, Decimal)
   • SentimentCategory (Choice: Positive, Neutral, Negative)
   • Any user metadata (Contact ID, Email, Page URL)

---

Step 3: Enable AI Builder Sentiment Model

1. Go to AI Builder in Power Apps
2. Select Prebuilt Models
3. Choose Sentiment Analysis
4. You do not need to train a model—AI Builder uses Microsoft’s prebuilt AI

---

Step 4: Create a Power Automate Flow to Analyze Sentiment

1. Go to Power Automate
2. Create a new Automated cloud flow
3. Trigger: When a row is added in Portal Feedback table
4. Next step: AI Builder – Predict
   • Model: Sentiment analysis
   • Input: FeedbackText field
5. Extract values from the output:
   • Sentiment Score (numeric confidence)
   • Sentiment Label (Positive, Neutral, Negative)
6. Add step: Update Row in Dataverse
   • Fill in the calculated sentiment values
   • Optionally send email to admins if sentiment is negative

---

Step 5: Show Sentiment to Admins or Users

In Power Pages:

• You can use a List Component connected to Portal Feedback
• Display feedback entries along with their sentiment label and score
• Use conditional formatting (e.g., red for negative, green for positive)

---

Step 6: Create a Power BI Dashboard (Optional)

To track sentiment trends:

1. Connect Power BI to Dataverse
2. Create visualizations:
   • Pie chart for sentiment distribution
   • Line chart for sentiment over time
   • Heatmap by page or submission source
3. Embed dashboard into admin-only page in Power Pages

---

Optional Enhancements

1. Trigger Alerts for Negative Feedback

• If feedback is “Negative,” send an email or Teams alert using Power Automate
• Include the feedback content, user info, and submission time

2. Use Copilot to Generate Action Suggestions

• In Power Automate, use Copilot (preview) to generate actions based on feedback sentiment
• E.g., auto-create a support case or escalate to a human agent

3. Add Feedback Context

• Capture page URL or context of where the feedback was submitted
• Helps link sentiment to specific portal sections or features

---

Use Case Example

Let’s say a user on your government services portal submits feedback:

“This page is confusing and I couldn’t complete my application.”

Your sentiment analysis flow will detect this as Negative, score it (e.g., 0.15), and automatically:

• Update the record in Dataverse
• Notify your service desk team
• Flag it in Power BI

---

Benefits Recap

• Automation: No need to manually read every comment
• Real-Time: React instantly to negative feedback
• Insights: Understand how different portal sections perform
• Engagement: Show users their feedback is being acted upon

---

Licensing Notes

• AI Builder sentiment analysis consumes AI Builder credits
• Credits are allocated per environment, based on license type
• Ensure Dataverse and Power Automate usage within limits