Background:

A large auto manufacturer was looking to enhance its feedback loop of both its product and customer dealer experience. They had an abundance of feedback in text form but no easy way to synthesize it into meaningful insights, trends or conclusions. They were looking for a new approach to make this data useful and had contracted with a vendor of natural language software but were not getting the support needed to make it useful.

Challenge:

How to classify and rank importance of free form text gathered based on dealer and product experience into useful insights, further summarized in meaningful reports.

The Approach:

AAARL developed the prototype models via machine learning to frame the performance and viability of using natural language processing. Significant testing processes during the development phase identified major processes and model changes necessary for the client to achieve specific KPI’s.

The manufacturer chose to implement the recommended approach to natural language processing and further relied on AAARL to validate the early results.

Results:

An automated approach to gather, refine, store, analyze and report on customer feedback that is provided in written text form. This information was then automatically sent to the “correct” department, given its sentiment (ie extreme positive or extreme negatives). This way, the relevant department can make timely mediation measures for that customer.

This feedback now feeds directly into the manufacturer’s KPI reporting on both their product and customer experience with their various dealers.