Addressing automotive safety issues with analytics
For decades, technology innovations have enabled automakers to steadily improve overall safety performance in their vehicles. While that effort has greatly improved overall vehicle safety, recalls have dramatically risen, culminating in 2014 being the most expensive year for automotive recalls. These recalls affected nearly 64 million vehicles. A recent IBM Institute of Business Value auto study, Auto 2025: Industry without borders, reports that over two-thirds of industry executives expect automated vehicle safety to be an ongoing differentiator with consumers over the next decade. This differentiator outranks items such as vehicle personalization, data privacy and access to zero-emission cars.
Two key reasons are attributed to the increase in recalls. First, while the vehicles are safer and more reliable today than in the past, their complexity has increased tremendously because of the use of sophisticated software algorithms and electronics resulting in the associated risks. For example, problems with the ignition switch may cause it to accidentally flip from the on position to the accessory or off position while the car is moving at speed. As a result, the vehicle may lose power to the engine, which impacts stability control and other electronic systems. Second, the industry is under increased scrutiny from the National Highway Traffic Safety Administration (NHTSA), as it is using deep analysis to find vehicle defects.
The first capability that can facilitate rapid resolution of safety issues is organizational. Many automakers unfortunately have had disjointed capabilities around safety analysis that results in their having to play catch-up to the NHTSA. Safety teams across automakers tend to operate in silos and have pockets of analytics capabilities spread across the enterprise.
Safety-focused organizations need to be more integrated, agile and analytics-driven than ever. This need aligns with automakers’ business strategy that enables them to carry out several actions:
- Engage in proactive problem identification and timely field action to minimize brand erosion
- Make accurate and speedy diagnoses of problem root causes
- Offer early feedback to engineering and product development to improve product quality
- Provide effective early warning to reduce the number of problems and mitigate extended costs
A second highly significant capability for resolving safety problems is focusing on rapid integration and using natural language processing (NLP) to explore unstructured data. Previously, analyzing unstructured data was code for people doing lots of reading that yielded very slow and subjective analysis. In my experience, exploration with NLP-based tools can be invaluable rapid analysis of large volumes of text that previously wouldn’t have even been considered data.
Safety teams have a foundational need for a single version of the truth. They need to be able to access and analyze data across multiple sources. This need includes the ability to analyze data in internal sources such as warranty, contact center, legal, product development and manufacturing, and external sources such as NHTSA, social media including Twitter and forums, and other third-party data.
A true transformational safety program requires substantial organizational and cultural changes as well as improved analytics approaches. Strategic alignment of people, processes and technology transformation efforts enables such a program to create business value. Operationalization of world-class safety capabilities requires a comprehensive, integrated plan based on a proven change management framework.
Organizational capability focuses on removing silos among service, manufacturing and product development through enhanced integration of personnel with streamlined processes to resolve quality issues in systems across the product lifecycle. Mature organizations are expected to establish centers of excellence for addressing safety issues and look to develop responsive systems that automate warnings and likely causes for safety problems before they’re otherwise identified. These efforts need to be coupled with a greatly expanded analytical initiative. Data that can aid in safety issue detection and diagnosis comes from many places. Much of this data may be transactional and exist in operational systems.
Automakers that effectively mature both their organizational and analytics capabilities can expect to see several benefits:
- Differentiated customer experiences from product improvements
- Rapid action and effective communication through compliance with regulations set by the NHTSA
- Significant savings in recall costs by early identification of potential safety problems
- Significant reduction in the time to analyze and gain insights for rapid identification of safety issues by using IBM Watson NLP capabilities for mining internal and external unstructured data