Challenge
In digital vehicle sales, users leave thousands of comments every month - praise, criticism, ideas. This cannot be analysed systematically by hand; valuable feedback went unused.
Solution
We built an infrastructure that processes comments automatically:
- Processing of more than 15,000 comments with several AI methods, including sentence embeddings.
- Categorisation into business-relevant topics such as UI/UX, performance, product and price.
- Sentiment analysis - positive or negative, per topic.
- Visualisation of the results in management dashboards.
We apply the same pattern outside automotive: for a healthcare provider, a GenAI-based crawler extracts KPIs from online reviews and makes service quality measurable.
Result
- User criticism is sorted by topic and can be prioritised.
- The digital sales experience is monitored continuously.
- Product and digital teams see which problems and suggestions users actually raise.
* Project: we owned the delivery
