Root Insurance is a leading US motor carrier using mobile telematics to provide accurate insurance quotes, underwriting, and pricing, based on real driver behavior. In the latest interview, Managing Director and SVP at Root Insurance, Isaac Espinoza, and Dr Andrew Johnston, Global Head of InsurTech at Gallagher Re, discuss how mobile telematics has changed the insurance landscape, current trends in InsurTech, and the AI boom.
Can you provide an overview of your current role at Root Insurance?
Isaac Espinoza: My role as Managing Director and SVP at Root covers several areas across reinsurance, corporate actuarial, strategy, and venture. I initially joined to build Root’s reinsurance program and structure it in line with the organization’s needs as a rapidly growing InsurTech, but I am now also involved with driving the business strategy of Root and helping with new initiatives.
Telematics-based risk underwriting and pricing
Root’s telematics model
Traditionally, motor insurance underwriters relied on proxies to assess risk. However, these factors will only every serve as proxies and real-time driving data enables a far more precise prediction of risk and provides a comprehensive understanding of driving habits.
Based on telematics data, trends in driver behavior include variations in driving styles that can be effectively analyzed using AI and machine learning for improved claims prediction. Risky behaviors such as speeding, hard braking, and nighttime driving are significant predictors of accidents.
Root has revolutionized motor insurance in the US by utilizing telematics technology. Root assesses individual risk levels based on actual driving behavior using proprietary telematics data. By leveraging GPS data from mobile phones and smart vehicles, Root gathers vast amounts of data, spanning billions of miles, to inform their risk underwriting and pricing strategies.
Root has identified that factors such as distracted driving (e.g., texting while driving) are significant predictors of loss, often more so than traditional risk identifiers. This shift underscores the importance of real-time driving data in modern underwriting practices, leading to improved pricing for customers. Safe drivers benefit from lower premiums, fostering customer loyalty in an era of rising insurance costs.
IE: My degree is in computer science, focusing on AI, machine learning, and data mining. The funny thing is, today, this would probably be one of the hottest degrees you could have, but at the time it wasn’t the best major following the dot com crash! This has given me a strong foundation for understanding the role of technology in insurance.
After qualifying as an actuary, I wanted to go where the action was. I ended up going to Cayman and joined a startup reinsurance company which allowed me to experience a lot of innovation, including helping to set up a venture capital unit.
After that, an opportunity came up to join Root and lead their reinsurance initiative, which allowed me to use my experience in insurance and venture capital and apply it to the InsurTech space.
Evolution of telematics with smartphones
Telematics has existed in various forms for decades, often involving external devices like dongles or "black boxes" plugged into a vehicle's port. However, this process was not user-friendly. The development of smartphones has allowed for more user-friendly and cost-effective data collection methods, integral to Root's business model.
Modern smartphones come equipped with sensors like GPS, gyroscope, accelerometer, and magnetometer, allowing for cost-efficient telematics. Customers can use their everyday smartphones instead of paying for an independent device. Mobile telematics streamlines the capture of billions of miles of driving data, building a comprehensive profile of driving behavior that can be compared to claims databases.
Figure 1: Telematics in an automobile1.
Root has made significant strides in the InsurTech space, particularly in revolutionizing motor insurance in the US with telematics technology. What level of interest have you seen for risk-based pricing centered around telematics?
IE: We’ve seen lots of demand for how Root provides motor insurance.
We utilize our proprietary telematics technology and telematics data to assess individuals' risk levels based on their actual driving behavior. Upon receiving appropriate consent from the consumer, Root leverages GPS data from mobile phones or other devices, including signals from smart vehicles, gathering vast amounts of data spanning billions of miles. This data informs our risk underwriting and pricing strategies.
Conducting the Root test drive and sharing data before receiving a quote can substantiate their safe driving behaviour.
A study into driving risk factors and telematics impacts
In 2018, the State Insurance Regulatory Authority (SIRA) in New South Wales, Australia, conducted the NSW Young Drivers Telematics Trial, at the time being one of the most extensive studies in the world investigating the effectiveness of telematics on driver behavior (SIRA, 2019)2.
The objectives of the study were threefold:
- Investigate whether, and to what extent, telematics can positively influence young driver behavior
- Identify opportunities or barriers to broadening the use of telematics to improve road safety
- Better understand young driver behaviors through collection of real-time driving data
The study focused on young drivers as drivers under the age of 25 were found to be most likely to be involved in motor vehicle collisions, according to NSW Centre for Road Safety data, being up to 4.5 times more likely to be involved in a collision and up to 5 times more likely to be involved in a collision resulting in death or serious injury (Tables 1 and 2).
During the trial, 717 drivers participated, with a control group of 355 individuals and a treatment group of 362 individuals.
Treatment group drivers were given a telematics device to provide real-time feedback over a six-month period, with harsh braking, rapid acceleration, harsh turning, and speeding all analyzed.
Additionally, speeding frequency; difference between real-time speed and posted speed limit; frequency and magnitude of deceleration; and average speed at which drivers choose to travel when unconstrained by road conditions, were all analyzed.
Key findingsRelative to the control group, the treatment group have:
- A lower frequency of speeding per 100 driving hours
- 42% lower rates of extreme harsh braking events
- 24.9% lower rates of very rapid acceleration
- 24.1% lower rates of harsh turning
In particular, results were most prominent among the young male cohort.
Overall, the average risk score for the treatment group was lower than that of the control group (60 compared to 63), suggesting that real-time feedback did have a positive influence on moderating driving behaviour.
What societal role do you envision this insurance type playing more broadly? Are there any concerns about exacerbating the insurance protection gap by rendering certain risks uninsurable?
IE: We consistently emphasize that Root is geared towards good drivers. However, if data indicates that a driver poses a higher risk, we may not be able to extend coverage or may need to offer it at a higher premium.
We aspire for this approach to promote safer driving habits among all individuals and contribute to enhancing driving behavior overall. By offering competitive pricing based on driving behavior, we not only incentivize customers but also underscore the value of safe and efficient driving practices. Ultimately, we aim to reduce fatalities and accidents by prioritizing accident prevention over solely addressing the aftermath through insurance coverage.
From its origins in GPS technology to the advent of real-time data analytics, the timeline below offers a glimpse into key milestones that have shaped modern telematics solutions.
In today's landscape, defining 'InsurTech' can be nuanced. Some argue that the label is becoming obsolete due to technology now being integrated by all insurers. Given this, how would you define 'InsurTech' currently?
IE: I agree that the label 'InsurTech' is evolving, as all insurers now use technology to varying degrees. I see InsurTech as a spectrum, with companies ranging from traditionally focused to highly tech-oriented.
Typically, InsurTech refers to companies founded in the last decade, often backed by venture capital and driven by technology, usually established by people with tech backgrounds rather than insurance.
However, this definition is limiting. Many established companies, labeled as traditional, have heavily invested in technology and innovation, aligning them closely with InsurTech principles. Therefore, a more nuanced understanding of the label is needed.
Andrew Johnston: I've considered this for several years and would also agree with Isaac's perspective.
InsurTech, as a label, is becoming obsolete due to its broad cultural association with technology integration in the industry. Historically, linking "tech" to any financial service has increased its valuation, which explains the widespread use of the label.
However, we're starting to prioritize technology as the product, rather than as a tool to enhance business models. For example, Root's strength is its business model—underwriting, pricing, and risk management strategies developed by insurance professionals. Technology and telemetry are just tools to implement these strategies and are insignificant on their own.
Therefore, while technology offers superior risk assessment capabilities, it still relies on human underwriting hypotheses, making the label 'InsurTech' increasingly outdated.
What do you anticipate as the prominent trends or technologies in the near future for InsurTech?
IE: We can't ignore AI, which will undoubtedly become increasingly integral to the insurance sector. However, as Andrew mentioned, the focus shouldn't solely be on AI itself.
The most significant trend in InsurTech will be applying new technologies to existing processes across the entire value chain. This includes enhancing risk selection, increasing operational efficiency, optimizing distribution and sales channels, and streamlining claims handling.
At Root, for example, we use AI to help claims adjusters process claims more efficiently. Customers can file claims through an app, making the process much more efficient than traditional methods. If insurers adopt such technologies in various areas, the cumulative benefits will lead to lower insurance costs and higher margins, allowing for greater reinvestment in new technologies and products.
AJ: One significant trend we're observing at Gallagher Re is the increased willingness to leverage the outcomes of appropriate technology. Just a few years ago, many viewed the current technological wave as a fad. Today, that skepticism is rare.
AI will be crucial across the value chain, as Isaac mentioned. However, the foundation of AI is data utilization.
Another trend is the growing recognition that data is meaningless without context. There was a time when quantity was valued over quality, but now there's a clear shift towards emphasizing data quality and management. This shift highlights the indispensable human element needed to interpret data effectively.
AI use cases in InsurTech
Data holds immense significance in insurance overall, especially within InsurTech. However, are there risks of introducing unintentional bias by insurers using AI for their data?
AJ: Yes, this is definitely a concern.
A model is only as good as the data it uses, and unfortunately, that data often carries inherent biases from the developers. Relying solely on this data can reinforce these biases.
While human interpretation of data also carries the risk of bias, it allows for discernment and the opportunity to challenge existing biases. However, it’s currently difficult to train AI models continuously based on human feedback. When a human challenges a model's output, providing a clear explanation for rejecting certain data is challenging, which limits the model's ongoing improvement.
This means that, while individual biases may be addressed, the overall model might still reflect a narrow perspective. Therefore, while AI offers significant opportunities, we must prioritize transparency and address biases in the modeling process. Although there are industry initiatives to monitor and regulate this, there's still a long way to go.
IE: This is true. We must ensure that we can effectively articulate our decisions during underwriting and pricing to mitigate the risk of overfitting trends and perpetuating bias stemming from the input data.
Ensuring that data does not autonomously drive decision-making is crucial. It should be part of a spectrum of decisions that includes human input.
We've seen recent EU legislation regulating AI use in European markets, especially regarding data privacy. Do you anticipate similar global regulations or broader compliance measures for technology use in insurance?
IE: I'll focus on the US market since that's where I primarily operate. Regulators are highly concerned about how AI is employed, particularly regarding privacy and bias, as discussed previously.
In discussions with regulators, most have shown openness to the idea that innovation is necessary for new products and technology. They encourage exploring new methods of conducting insurance.
However, balancing innovation and ensuring customers' best interests remain paramount from a regulatory standpoint.
One approach we're adopting to foster greater collaboration with regulators, ensuring best practices and compliance, is through a newly established non-profit trade organization called the InsurTech Coalition.6 Although formally established in late 2023, its development had been underway for some time.
The group's primary objectives include advocating for technology within the insurance sector, collaborating with regulators, influencing public policy, and addressing concerns regarding the use of AI and technology in insurance operations. However, it is important to point out that while there is a lot of focus on the negatives of AI and technology, there are also many positives, and this group also helps promote these. For example, AI and telematics can help improve road safety, reduce insurance costs, and bring fairer insurance rates.
The Coalition has been well received, and member companies have sent representatives to attend the National Association of Insurance Commissioners (NAIC) meetings to contribute to policy and regulatory discussions, ensuring collaboration on common issues.
How can we help?
Please contact our Gallagher Re InsurTech team to learn more about how our team can help you assess and manage your InsurTech requirements.
