The Automotive Industry And The Data Driven Approach

Though often overlooked, cars serve as a rich data source. Millions of transportation vehicles whizz past us on a regular basis, each of which generate swaths of useful information that automakers are now figuring out how to monetize. Some of the biggest passenger car automakers have more than 10 million vehicles’ worth of data sitting in their data repositories. Failure to tap into these vast data stores amounts to lost value-added for customers, lost safety opportunities and lost revenue and business intelligence. 

According to a McKinsey Report, “The overall revenue pool from car data monetization at a global scale might add up to USD 450 - 750 billion by 2030.” In addition, according to a market analysis report on the Automotive Cyber Security Market, “The global automotive cyber security market size was valued at USD 1.44 billion in 2018 and is expected to grow at a compound annual growth rate (CAGR) of 21.4% from 2019 to 2025.” This trend has prompted a number of data-driven companies to appear, each of which are geared towards enabling new services for the automotive industry. One such company is Viaduct, a data-driven, machine learning startup that just raised $11 million from Innovation Endeavors to enable automakers to gain insights and extract value from vehicle intelligence. 

Creating business value from connected vehicle data is a non-trivial technical challenge, as any successful machine learning approach needs to contend with leveraging asynchronous and highly heterogeneous time-series data at scale in order to generate predictive insights. As an example of the technical challenge described above, accurately estimating which vehicle on the road will experience a failure in the next month is in itself a highly unbalanced problem (Few failures are observed out of over 10 million and counting potential vehicle candidates), and requires processing the full historical set of data on all vehicles in the fleet, including hundreds of individual sensor streams, historical failure information (which includes unstructured text information from dealers and  Service centers), as well as integrating a multitude of build and usage pattern information.

Indeed, intelligence gathering from automotive data is no walk in the park. However, a company like Viaduct can leverage artificial intelligence technology in order to unlock automotive insights by way of machine learning approaches that have only recentlybeen developed in academia. To paint the picture of how difficult this would be for Original Equipment Manufacturers (OEMs) to execute themselves, without partners like Viaduct, they would have to build their own custom AI software for data extraction. This means that OEMs would have to go out of their way to find, recruit and hire world-class machine learning engineers––a challenge for even Silicon Valley startups. It is no wonder, then, that OEMs and players in the automotive industry are looking towards companies like Viaduct for smart solutions to the vehicle data analytics problem.

One connected data use case that can serve as a major driver of cost savings in the automotive industry is “smart campaigning.” When automakers find a systemic issue in their fleets (e.g.