Healthcare Retail / E-commerce
Voice of Customer Analytics
Multilingual review data from e-commerce platforms worldwide, run through custom aspect-based sentiment models to find out why sales were declining — and what R&D should fix.
The problem
A global leader in medical vacuum women’s healthcare retail products was watching e-commerce sales decline and did not know why. Reviews existed, spread across marketplaces in different countries and different languages — but as prose, not as data. Nobody could answer “what specifically are customers unhappy about, and is it getting worse?”
They wanted two things: the cause of the decline, and the opportunities to reverse it.
What I built
Acquisition. Customer review data pulled from multiple e-commerce platforms worldwide, across several products and several languages.
Analysis. Custom aspect-based sentiment models, trained for this domain, to attribute sentiment to specific product characteristics rather than scoring reviews as a whole. Medical retail products have aspects that a generic model does not know about.
Comparison. The same treatment applied to competitors’ products, so the findings were relative rather than absolute. A weak score on an aspect means something different once you know whether competitors score the same way on it.
Delivery. Interactive dashboards, so the insight stayed explorable instead of arriving as a one-off report.
The outcome
- Revealed the product characteristics customers actually weigh when choosing — and where this brand fell short of competitors.
- Identified specific areas for improvement, which the client used to channel R&D effort into product development rather than guessing.
- Turned a declining sales number into a list of things to go and fix.