Woolworths uses LLMs in customer feedback analysis

Woolworths is increasingly using aggregated customer feedback to power its customer and employee experience (CX/EX) decision-making, courtesy of a "feedback analytics" platform made by Thematic.

The retailer told iTnews that Thematic uses "innovative and secure large language models (LLMs) to support Woolworths' customer and employee insights."

The models are understood to include GPT in Azure.

LLMs have become an area of intense experimentation for Australian enterprises through the first part of 2023.

Data from Woolworths' Thematic dashboard, a snapshot of which was published last month, showed 'customer service' and 'range and stock' as key themes of customer commentary, with the latter including whether items were in or out of stock when visiting a supermarket.

Woolworths is using both 'comment analyzer' and 'theme summarizer' features of Thematic's platform.

Comment analyzer is intended to quantify and visualise the volume and sentiment levels of feedback data; theme summarizer generates descriptive overviews of ‘base themes’ and ‘sub themes’ in aggregated feedback. 

For example, Thematic’s ‘theme summarizer’ identified that a recurrent grievance in Woolworths’ online shoppers’ comments about the ‘range and stock’ ‘base theme’ was “having to purchase more expensive items due to the unavailability of cheaper options.”

Woolworths is also using a ‘sentence cluster’ feature to be able to review the underlying source data - the actual customer opinions, as they were written.