Stop missing user feedback scattered across channels. This workflow collects feedback from Intercom conversations, Twitter mentions, and support emails, uses AI to identify themes and sentiment, and delivers a weekly insights report to Notion so you always know what users are saying.
Collect feedback from Intercom conversations
The workflow queries Intercom for all customer conversations from the past week. It extracts customer messages, complaints, feature requests, and compliments, filtering out purely transactional exchanges to focus on substantive feedback.
Fetch brand mentions from X/Twitter
From X/Twitter, the workflow collects all mentions of your brand, product, and relevant keywords. It captures both direct mentions and indirect references, including replies and quote tweets that contain user opinions.
Extract feedback from support emails
The workflow scans your support inbox for emails containing feedback, complaints, or suggestions. It extracts the core feedback from email threads, filtering out automated responses and focusing on genuine user input.
Analyze themes and sentiment with OpenAI
Using OpenAI, all collected feedback is analyzed to identify recurring themes, sentiment trends, and priority issues. The AI categorizes feedback into buckets like bugs, feature requests, UX issues, and praise. It highlights emerging problems and opportunities.
Generate weekly insights report in Notion
A comprehensive feedback report is created in Notion with theme breakdown, sentiment analysis, top feature requests, critical issues to address, and positive highlights. The report includes example quotes and links to original feedback for context.
Why aggregate user feedback with AI?
User feedback is scattered across Intercom, Twitter, emails, and app reviews. Without aggregation, you miss patterns and make decisions based on whoever was loudest most recently. AI-powered aggregation reveals what users actually care about.
See the complete picture of user sentiment
That bug report in Intercom might seem isolated until you see five similar Twitter complaints and three emails about the same issue. Aggregation reveals patterns that individual channels hide.
Identify trends before they become crises
When the same complaint appears across multiple channels in one week, something is wrong. Early detection lets you fix issues before they drive churn or damage your reputation.
Make product decisions based on data, not noise
The loudest feature request isn't always the most important. AI analysis shows you what the majority of users actually want, not just what the most vocal users demand.
How to set up feedback aggregation
Setting up this multi-channel feedback workflow takes about 20 minutes. You'll connect your feedback sources and configure categorization preferences.
What you need to get started
- Intercom account for chat feedback
- X/Twitter API for social monitoring
- Gmail or support inbox access
- OpenAI API key for analysis
- Notion workspace for reports
Configuring feedback sources
- Connect Intercom and select relevant inboxes
- Set up Twitter search terms (brand name, product, common misspellings)
- Configure email filters for feedback-containing messages
- Add any additional sources (app store reviews, G2, etc.)
Customizing theme categories
- Define your product areas for categorization
- Set up feature request vs. bug vs. UX issue classification
- Configure sentiment thresholds (what counts as negative?)
- Add custom categories relevant to your product
Frequently asked questions about feedback aggregation
How do you handle duplicate feedback from the same user?
The AI identifies when the same user provides feedback across multiple channels and deduplicates while noting the repetition. Repeated feedback from one user gets counted once but flagged as high-intensity.
Can I add more feedback sources?
Yes, you can extend the workflow to include app store reviews, G2/Capterra feedback, community forums, or any source with API access. The AI analysis works with feedback from any text source.
How accurate is the theme categorization?
AI categorization is highly accurate for clear feedback. Ambiguous items are flagged for review. You can improve accuracy by providing examples of each category and refining over time.
What if we get too much feedback to process?
The workflow handles high volumes by focusing on representative examples rather than every single piece of feedback. The themes and sentiment are accurate even when sampling from large datasets.