COVID-19 has had a dramatic and significant impact on the auto industry globally, with new car production dropping sharply and manufacturers being forced to hold lower inventories. Current data suggests a total global auto sales of around 64 million units in 2020, which is a large drop from the 80 million units sold in 2017.
McKinsey estimates that the 20 largest manufacturers in the industry will see their profits decline by about $ 100 billion in 2020, down about six percentage points from just two years ago and stresses that it will take probably several years to recover from this drop in profitability.
In parallel, and quite separately from the coronavirus crisis, the impact of another crisis began to be felt: that of the climate. The industry faces significant challenges in how its customers want to use its products, whether it’s automating part of the drive, the carpool craze, or the shift to electric vehicles.
As auto companies continue to operate in a rapidly changing market, brands will also need to innovate to stay ahead of the competition. However, this will now have to be done in a context with fewer resources and over a shorter period of time.
The right talent must be hired and retained, especially in R&D, data science and data engineering, in order to effectively use the available data. These new employees will also have to automate a large part of the activity, and quickly, to ensure that the information generated is used as quickly as possible.
True trust in AI must therefore be established in order to increase the adoption of data-driven technologies that are useful in many business processes. Machine learning models should be able to be clearly explained to decision makers, with the logic behind the predictions identified. Businesses that don’t use AI to scale quickly and automate many processes will continue to be exposed to the turbulence of this challenging environment.
Why ? Because as we enter a post-COVID era, the auto industry cannot afford to be ostrich. As McKinsey also reminds us, “Even if manufacturers now focus on the essentials, failure to explore other possibilities could hurt them in the long run. “
Clearly, the battleground of the future is data. Today’s vehicles of all categories already generate a wealth of data, with hundreds of sensors monitoring everything from engine temperature to speed, acceleration and braking processes. Also according to McKinsey, 25 gigabytes of data are created every hour by the majority of cars, and as we move to even more autonomous vehicles, up to 3,600 gigabytes of daily data could easily go untapped.
In addition, data collected throughout the life cycle of the automotive industry such as product design and development, supply chain and manufacturing management, as well as marketing, sales as well as service and maintenance operations are also available.
It’s time to collect and harness this in order to make the driving experience more enjoyable, but also to improve our own manufacturing efficiency and make our processes more profitable. However, how can organizations in the automotive sector harness this wealth of information in a timely manner and transform themselves into 21st century businesses?
Artificial intelligence (AI) and machine learning (ML) are essential. They will help the industry harness this incredible wealth of operational data to help us cope with uncertainty as the emergence of AI-driven transformation projects promise to cut costs and boost productivity.
However, in many ways this journey has only just begun. Automotive IT leaders seem to be struggling to effectively integrate AI into applications, wasting time and money on AI projects that never go into production.
While not limited to the automotive industry, Gartner warns that CIOs struggle to develop AI projects because they lack the tools to build and manage a quality AI pipeline in production. According to IDC, about 28% of all AI and ML initiatives end in failures.
Does this rather grim record suggest that AI is not useful? Far from there. The problems are much more structural: the lack of staff with the necessary expertise, the lack of data available to fuel machine learning models and the lack of an integrated development environment are the main obstacles.
AI can make a difference
If we find that access to data is not the problem of the automotive industry, working effectively with that data is, and it becomes essential that companies give their employees the right tools and the right technological environment to operate efficiently.
We have reason to be optimistic here, as the automotive industry has a long and impressive track record of harnessing the latest technology to deliver reliable, innovative and safe vehicles, while continuously striving to reduce manufacturing costs.
We see examples everyday of AI being integrated into many processes in order to achieve real success. Applications that we know directly include:
In the United States, unscheduled stops cost $ 1.3 million per hour in the U.S. auto market alone (and therefore $ 50 billion per year). But by using the data generated during the manufacturing process, we are able to predict equipment failure that allows for “at the right time” service interventions, reducing machine downtime and maximizing the output. machines can provide.
This type of analysis can also be applied to use data generated by driving a car to predict the need for service. By using predictive maintenance, customers can be assured of an overall reduction in maintenance costs of 30%, the number of breakdowns up to 70% and scheduled repairs of 12%.
Better optimization of stocks and parts
H2O.ai helps the automotive industry manage inventory costs by generating more accurate demand forecasts for hundreds of thousands of parts at a time. This helps reduce costs as manufacturers can now have minimal inventory while maintaining service levels and ensuring product availability remains high.
Improved sales and marketing activity
By using the data in large machine learning models to understand the buying behavior of customers, the marketing department can target a potential customer with a personalized message at the right time and through the right communication channels. AI is also applied to improve the sales forecasting process and help with vehicle setup.
So we know the auto industry can leverage data and make artificial intelligence solve its problems. Yes, we are facing challenges: the shock of 2020 on profit margins means that companies in the sector have had to contract high levels of debt; sales levels are not expected to reach pre-vaccination levels until 2022 at the earliest.
It is therefore important that the industry finds the means to optimize operations and expenses for the next half-decade (if not longer).
Rising customer expectations will continue to push the auto industry into a time of transformation through AI. We have access to great data, and companies that can quickly derive actionable intelligence from it will emerge from the COVID-19 era ahead of the rest.
The technology, talent, and processes surrounding AI now exist. It is therefore the actors who apply it now, and who do it correctly, who will see their costs decrease and the satisfaction and loyalty of the drivers of their vehicles increase dramatically.