Feedback Intelligence
AI pipeline that turns hundreds of learner responses into actionable insights
Role
AI Automation Specialist
Client
Training Company (Confidential)
Timeline
2 Weeks
Key Result
80% less manual review time
01
The Challenge
A training company collected hundreds of learner feedback responses per course via Google Forms — NPS scores, open-text feedback, and satisfaction ratings. Manual analysis took the L&D team days per course, and even then they were reading responses line-by-line rather than identifying patterns. Three courses had declining satisfaction scores for months, but nobody knew why.
02
The Approach
I built a fully automated feedback intelligence pipeline that collects, analyzes, and summarizes learner feedback — delivering weekly insight reports without any human intervention.
Data Architecture & Collection Setup
Standardized Google Forms across all 12 active courses with consistent fields. Make watches for new form submissions and routes data to a centralized Google Sheet with course tags and timestamps.
Sentiment Analysis Layer
A Make scenario (triggered via webhook) calls ChatGPT to perform sentiment classification (positive/neutral/negative) and urgency scoring on each open-text response.
Theme Extraction
Batch processing groups responses by course and runs theme extraction via ChatGPT — identifying recurring topics (navigation confusion, pacing issues, content gaps) across all responses.
Visual Summary Reports
Make generates a structured weekly report: sentiment breakdown, top 3 themes per course, trending issues, and NPS delta — formatted as a clean Google Doc sent via email.
Critical Issue Auto-Flagging
Responses with urgency score ≥ 8/10 or specific trigger phrases ("can't access", "broken", "error") are auto-flagged and sent immediately to the course creator via Slack and Gmail.
Tools Used
03
The Results
80%
less manual review time
3
recurring UX issues identified (previously missed)
+15%
course satisfaction after AI-suggested fixes
< 1 hr
weekly — down from 2 days of manual analysis
Next Project