Understand automotive customer sentiment at scale
BUSINESS CHALLENGE
User feedback on automotive reviews is scattered across thousands of YouTube videos. The lack of centralized analysis, high volume of comments, and the need for qualitative insights (beyond views or likes) made it difficult to understand real user sentiment, detect recurrent issues, or identify market preferences.
SOLUTION IMPLEMENTED
Built a secure, scalable application to analyze YouTube reviews using AI-driven sentiment analysis and topic extraction. The solution automatically retrieves, processes, and categorizes user opinions (e.g., on fuel efficiency, design, comfort), turning raw feedback into actionable insights.
RESULTS & ROI
INSIGHTS
Surfaced patterns in user feedback across thousands of videos, helping manufacturers understand user perception per model or brand.
DECISION
Enabled marketing and product teams to base decisions on real customer voice, leading to improved targeting and product positioning.
SCALABILITY
Automated processing pipeline supports any number of new video lists, unlocking reuse across models, languages, and regions.
TECHNOLOGY
DATA PLATFORM
Used Azure Functions and Azure OpenAI to process comments with NLP for sentiment analysis, topic detection, and opinion mining.
SERVING
Developed a web app with Next.js hosted on Azure Container Apps to collect and manage YouTube video lists and associated metadata.
AI
Queue management with Azure Storage and secure login via Microsoft Entra ID to ensure authenticated usage and scalable workloads.