
91%
sentiment accuracy
The story
From problem to proof.
Trendora, a regional retail chain with 40+ stores, was making merchandising decisions on gut feel and month-old sales reports. Meanwhile, thousands of customer reviews and social mentions sat unread across a dozen platforms.
We built a unified sentiment intelligence platform that ingests every review, mention and survey response — and turns it into a live signal their buyers actually use every morning.
What we brought
The challenge
Thousands of voices, zero signal
Customer feedback was scattered across review sites, social channels and in-store surveys — in two languages. Merchandisers had no way to know which products were quietly failing until the quarterly numbers arrived.
- 12+ disconnected feedback sources
- Bilingual reviews (English and French) broke generic sentiment tools
- Buyers needed answers in seconds, not analyst tickets
What we did
Four moves that made it work.
Unified the pipes
One ingestion layer pulling reviews, social mentions and surveys into a single governed stream — deduplicated, cleaned and timestamped.
Tuned the sentiment engine
A bilingual sentiment model fine-tuned on retail language, reaching 91% accuracy on Trendora's own labelled review set.
Built for buyers, not analysts
A decision-grade dashboard: what's trending, what's failing, and which stores feel it first — no SQL required.
Closed the loop
Weekly digest automation pushed the three most actionable signals straight to the merchandising team every Monday.
The product
See what we shipped.


What we achieved
The numbers that followed.
Merchandising meetings changed from arguing about anecdotes to acting on evidence. Slow-moving products are now caught in days instead of quarters.
91%
Sentiment classification accuracy
3.2x
Faster merchandising decisions
+17%
Repeat purchase rate
12+
Feedback sources unified
“For the first time in our company's history, the head office knows what customers feel before the store managers do. That inversion changes everything.”
Danielle Roy
Chief Merchandising Officer, Trendora
Someone's case study like this is being written right now. It could be yours.
Every engagement above started with one free, no-obligation conversation.