Case Study · Retail
Defacto IofTheCustomer: an NLP system turning hundreds of thousands of customer reviews into sentiment, category and action
Every product review from defacto.com and the marketplaces passes through sentiment analysis within seconds; if negative, it is assigned one of nine categories and routed to the responsible team. Built in 2021, running without interruption for three years, Stevie award winner in 2023.
- 9
- Categories
- 700K+
- Reviews processed
- 3 years
- Uninterrupted
- Stevie 2023
- Award
Defect, fabric, supply, shipping…
By the end of the first year, reportable
In-house, on the DSO's server
Problem
Hundreds of reviews are written every day about Defacto’s products on sale — on its own site and on marketplaces. Inside them are actionable signals, from product quality to shipping, from sizing to warehouse errors. But reading hundreds of reviews a day and passing them to the right team is not feasible by hand; an unread review is an action not taken.
Solution
IofTheCustomer automates that flow with natural language processing. Each review first goes through sentiment analysis: positive, negative or neutral. If the review is negative, a second model finds why — one of nine categories such as defect, fabric, supply or shipping.
Example: “Nice product, but I waited five full days for it to arrive.” The system first labels it negative, then sets its category to “shipping” and passes it to logistics. At the same time, reviews about product quality go to product management and those about warehouse errors to warehouse operations.
What sets the project apart is that the whole process runs without human intervention: reviews are collected from the channels, processed within seconds, routed to the responsible teams and reported as trends in Power BI. By the end of the first year more than 700,000 reviews had been processed and were reportable.
System
The project was built entirely in-house and runs as a Python procedure on the DSO’s own virtual server; results are written to DSODB and monitored through the DSO’s log structure. Lumtify both developed the AI models and managed the project.
Outcome
The system has been running without problems for three years since 2021. Improvements in areas that directly affect customer satisfaction — product quality, logistics, warehouse management — are fed by this feedback flow. The project won a Stevie award in 2023.
The sentiment analysis + problem classification approach built here later became the foundation of Lumtify’s Comentio product.