A large grocery retailer client generated thousands of verbatim comments every week as a result of their in-store customer experience program. There was a lot of unstructured text requiring our client to both classify and quantify what customers were saying to prioritize awareness and action.
Using our proprietary in-house text analytics capability, we created a custom solution that both identified meaningful themes and their volume, with the ability to track over time. After identifying topline themes, we created common groups for key elements of the offer such as ‘Service’ or ‘Products.’ One of the unique benefits of our text analytics capability is the ability to completely customize the specific ‘defined rules’ that decide which groups or themes the comments are divided into. Once the defined rules were set up, every comment was coded in real-time. The defined rules are periodically reviewed and refined, to increase accuracy and enable relevant sub-groups to be identified.
The defined rules were further developed to ~30 categories, each based on a custom set of rules to generate optimum insight and action. Over a period of months, the accuracy was increased to 80% as the classified comments were verified against actual comments and the rules tweaked to further improve accuracy. This drove engagement with the outputs from all staff in the business who found the categories both relevant and believable and used them in day-to-day activities to drive improvement.