How data-driven automakers are getting faster, smarter, and smoother

The most advanced automakers are using sophisticated analytics and data-driven processes to enhance the customer experience and dramatically lower costs.

Physically recalling large numbers of vehicles to fix faults is a costly exercise for automakers, both reputationally and financially. But with the advent of connected software-defined vehicles, these costs can be cut dramatically through the smart use of sophisticated analytics and data-driven processes. By avoiding mass recalls, the most advanced players in the sector, such as Tesla, Rivian, and Lucid, are saving billions of dollars.

Culturally and organizationally driven by data, this new breed of automakers uses sophisticated analytics and technological tools to take a highly targeted approach to fixing vehicle faults. By being selective, they may only need to recall less than 5 percent of the fleet, rather than every vehicle in that category.

Their ability to capture and analyze data in a timely fashion allows the new generation of OEMs to rapidly identify and address emerging issues and, in doing so, proactively mitigate the risks that in turn prevents further incidents. Our benchmarking suggests they can move five times as fast as automakers that rely on a traditional approach. The result: a major reduction in the number of issues in the field and the time vehicles spend off road. Swift and efficient problem resolution maintains customer loyalty and trust.

The financial benefits can be enormous. We calculate that a typical automaker could save an extraordinary $2.5 billion a year by applying an advanced data analytics system to a fleet of five million vehicles.

Comparing a challenger’s troubleshooting to that of a traditional automaker

When a long-standing automaker in the United States suffered a vehicle fault in 2023, it took 12 months to fix the problem, and the company had to recall more than 50,000 vehicles. By contrast, the data-driven OEM approach enabled it to resolve a vehicle fault globally in just 2.5 months and limit the number of vehicles recalled to 2,000.

For the traditional automaker, the problem first came to light when the R&D team received 10 notifications over a four-week period. As service engineers were directly resetting the cars for customers, it took the automaker three months and more than 100 vehicle breakdowns to identify that the cause was a fault in a specialized engine control unit (ECU).

In the challenger’s case, the R&D team moved into analysis mode after being notified of less than 10 faults over a two-week period. Their service tagged every breakdown in the system, and once a threshold is passed, the issue is escalated to the R&D team. By applying data analytics to the error codes and signals, engineers identified that the cause was a kinked cooling hose. It took just four weeks to identify the fault.

Whereas the traditional automaker issued a service guideline to tow cars and reset batteries, the challenger used incoming vehicle data to identify the affected cars, which turned out to be less than 5 percent of the fleet. It offered these customers a replacement car until the issue was fixed and used a temporal software-based solution: a front axle relieving torque distribution. The R&D team recommended changing the hose for approximately 1,000 vehicles through a service. Although the company fixed 2 percent of the fleet via immediate service call-ins, another 3 percent could be fixed with the next regular service. An additional 10 percent of the fleet was identified as in need of observation, while the remaining 85 percent was considered not to be at risk.

By contrast, the traditional automaker took six months to implement a permanent fix. Having identified the root cause (an ECU stuck in boot-up), the company had to redesign the software and used a federated approval process for release. It then recalled all the possibly affected derivative models—more than 50,000 vehicles—for a software flash of the faulty ECU.

Inside a data-driven automaker

Let’s take a closer look at how a data-driven automaker, such as Tesla, Rivian, and Lucid, is able to resolve issues much faster than a traditional player. First, the most advanced players tightly integrate their service and R&D operations so that the engineers receive direct and fast feedback via an interactive service diagnostics app (containing detailed descriptions, pictures, videos, or audio recording of faults) and a customer app used by an active community along with alerts from a sophisticated analytics platform. They also tend to make use of joint platforms for bug and issue tracking and agile project management.

When a fault becomes apparent, the response is coordinated through regular exchanges between a service board (responsible for quick fixes) and a vehicle board (which develops R&D solutions, managed via Jira-based prioritization and assignment). Drawing from multiple centers of excellence, standing teams are pieced together from interdisciplinary functions with the end-to-end capabilities to best solve a given task. Both service and R&D team members bring a cross-section of industry experience and are encouraged to communicate actively and frequently with one another to expedite solutions to arising issues.

The integration of the service and R&D functions is underpinned by a digital twin—a centralized, transparent, and consistent model that makes information from multiple databases accessible via a single user interface (see figure 1). Within the digital twin, containing vehicle configuration data, in-field live insights as well as R&D data are supplemented by customer profiles (such as information on location or driving style) based on customer consent settings. Through the data digital twin, service org and R&D staff can analyze individual vehicles or entire fleet to pinpoint potential issues.

The most advanced automakers also have world-class analytics capabilities in terms of both tools and human resources. A talented data science team with broad vehicle know-how is deeply embedded in the R&D function. This team can systematically model error alerts and perform root-cause analysis with the aid of standardized tooling built on commercial off-the-shelf analytics software as well as data infrastructure. This ensures quick integration when expanding the data team by making onboarding simple and familiar.

Vehicle and customer data analytics allow for a risk-based triage of the fleet, which minimizes the impact on customers and costs (see figure 2). This analytics draws on individual customer profiles, such as their geographic location and driving style, to evaluate whether they are likely to be affected by a specific vehicle issue and individual vehicle profiles (such as production and parameters) to monitor the vehicle’s vulnerability to the issue. These filters allow the automaker to pursue a vehicle-specific service strategy, meaning only a small fraction of vehicles will need to be called in for a service. Most vehicles can remain on the road with no need for physical work or a parts exchange, limiting the impact on the customer experience. Once a subset of vehicles has been identified for monitoring a specifically developed in-vehicle service alert can be implemented on this fleet segment. This ad hoc alert model can be developed and verified using the digital twin fleet data. The alert is only monitored remotely by R&D so as not to unsettle the customer.

At the same time, it is vital to keep customers in the loop. Tesla, Rivian, and Lucid, for example, makes extensive use of two-way and near-real-time communications using in-vehicle channels, such as notifications in the infotainment system, over-the-air (OTA) updates and data collectors, and apps and other customer-based interaction channels. The goal is to create fast data and feedback loops, driving product innovation and ultimately customer satisfaction. In cases where a vehicle is deemed to be at high-risk of a fault, quick fixes can be made OTA, supplemented by customer service alignment and notification via the communications channels and apps.

By using data to optimize the performance of the vehicle, the automaker can deliver a compelling customer experience. OTA updates to the vehicle software support that mission by enabling the automaker to adapt the experience as required. As they have put the customer at the heart of their product design, Tesla, Rivian, and Lucid and other advanced automakers regard a constant bidirectional exchange of data as crucial for the success of their businesses.

Data analytics keep cars on the road and customers happy

The automotive industry is moving on. The days of every vehicle fault being diagnosed and repaired by mechanics in service garages are numbered. Automakers can no longer afford to indulge in mass vehicle recalls that cause disruption for their customers and majorly impact their brand reputation. Instead, they need to focus on using connectivity and software to proactively identify and fix faults without taking vehicles off the road. To do that, automakers need to be driven by data.

BinaryCore, together with Kearney, is helping automakers learn from Tesla, Rivian, and Lucid and other high-tech challengers that organize themselves around data. To help our clients realize the full potential of software-defined vehicles, we typically provide an upfront modular assessment, which is then used to implement a proof-of-concept and create a transformation road map. As you embark on this journey, the financial and reputational benefits will become very apparent. Don’t get left behind.