Sentiment Spotlight: Crisis In The Automotive Industry

By: Dheepan Ramanan

September 24, 2015

Tags:
Sentiment Analysis

The Volkswagen emissions test scandal is just the most recent crisis to hit the auto industry. Over the past five years, car recalls have been in the headlines more than ever before, led by malfunctioning Toyota accelerators, exploding Takata airbags, and faulty GM ignition switches.

Stories like these can destroy customer loyalty.  To avoid another year of record-breaking recalls, it is critical that automakers identify and quickly correct product defects.  An omni-source approach to data analytics gives car companies the best chance to find and fix emerging issues before they become the latest trending topic.

Using Clarabridge’s NLP and sentiment analysis technologies I examined social data, safety records from the National Highway Traffic Safety Administration, and dealership logs to find trends and insights that could be used to monitor competitors, find early recalls, and spot product defects.

Social Media
Social insights can unlock how customers feel about your brand, your products, and your competitors.  Monitoring and analyzing social is critical to understand how consumers perceive your brand.

It is not enough to use social media just to identify high-level trends, however. Automakers need to conduct deeper analytics that involve sentiment analysis and deep text parsing.

I examined social media data from Facebook across even major auto brands to find competitive intelligence. I also uncovered a chief source of negativity behind one company’s sentiment scores.

  • We found that recalls can effect brand sentiment unequally. For example, the Toyota recall that occurred recently dropped that particular brand’s sentiment; however, Chevrolet suffered a far steeper drop in sentiment during GM’s recall the previous year.
  • Using competitive analysis, we found that Mazda and Hyundai featured the highest overall brand sentiment.
  • Analyzing themes, we quickly found an issue with rusting frames that were driving low sentiment in Toyota.

Brand Sentiment Trended

Toyota
Chevrolet
Honda

  • During the GM recall in 2014, Chevy’s brand sentiment dropped to negative -.27
  • Even though Toyota’s brand reputation may have taken a hit from recalls in the past 5 years, this most recent recall has not dramatically dropped sentiment (now at -.1)

National Highway Traffic Safety Administration (NHTSA)
Automakers are required by law to report defects to NHTSA within five days of identifying the vehicle flaw. By analyzing complaint notes made to the NHTSA, it is possible to spot trends as they begin to accelerate.

  • Both the Takata airbag recall event and the Toyota acceleration recall are evident in the NHTSA and feature massive spikes in volume
  • In the case of the the acceleration recall, spikes in binding & sticking comments in the category of engine rose 126% in the beginning of 2009, almost 10 months before the recall began
  • Issues with airbag deployments could be seen up to 4 years before the Takata airbag recalls

Sticking Accelerators Trends

INOP / Inaccurate
Binding / Sticking
Performance Issue
Improper Function
Intermittent Issue
  • Although engine throttle comments jumped 63% from 2008 to 2009, binding and sticking comments in the same topic increased 126%
  • This would have given an early warning indicator that there was a manufacturing defect

word cloud
Dealer Notes
Dealer notes can be a particularly valuable source of feedback to discover potential issues with new vehicles as they happen. New vehicle launches can be compared to older launches to identify problems. Tracing these issues back to the manufacturer allows companies to fix issues quickly.  One recent automaker used Clarabridge to analyze dealer notes for a visual refresh of one of their top selling hatchbacks.

  • Comparing this launch to a previous visual refresh they found increased volume and negative sentiment around front bumpers and dashboards.
  • Using this insight, they were able to trace back an issue with vertical stacking of bumpers at the manufacturing plant. This stacking caused dents in some of the cars.
  • Additionally, since this vehicle was coming from a new plant, practices to wipe down the cars for transport had not been implemented, creating dusty dashboards
  • Fixing these issues at the manufacturer eliminated these comments, boosting overall customer sentiment.

Front Bumper Comparison


As the fallout from the VW scandal floods our social timelines, it is clear that brand perception is more important that ever. Using omni-source analytics is the way top brands will not only win hearts, but keep them.

 


Dheepan Ramanan is a data scientist at Clarabridge. Follow him on Twitter @DheepanRamanan.
2016-1486_clarabridge_demo_request_cta