The situation
A large online home improvement retailer sold a huge range of products, and almost every product collected reviews and customer questions. The category teams who looked after each range were responsible for making it better.
The problem
The most useful answers to "what should we fix?" were sitting in free-text comments. But nobody can read thousands of comments, so each team read a sample. Decisions ended up leaning on whichever comments happened to be seen, and recurring problems could go unnoticed for a long time.
What we did
Instead of asking people to read everything, the team had the comments sorted automatically by what they were about. Think of themes like "hard to assemble" or "arrived damaged". People who knew each category then checked the groups and gave them plain names, so the themes meant something to the business.
On top of that went a simple report. Analysts could pick a product range and see which themes customers raised most, how that was changing over time, and what they were saying in their own words.
The outcome
Reading comments by hand stopped being part of the job, which freed up more than 2,700 hours. More importantly, teams could now see the whole picture, not a sample, and point to specific areas to improve. Customer satisfaction measures rose by 4% as those improvements landed.
What it meant for the business
The work moved from reading to fixing. Analysts spent their time on what to do about a problem, and the numbers gave them evidence to point to.
Could this work for you?
Any business sitting on reviews, support tickets, survey answers or call notes has the same opportunity. If you have more feedback than anyone can read, this is a good first project.
Related services: AI solutions and automation, Data visualization and reporting
This project was led by our founder, Achyutanand Jha, in an earlier role. The client is not named, and figures are those reported at the time.