Arina Man: Using AI to Prevent Losses in Motor Insurance

We had an interview with Arina Man, the CEO of kasko2go. Kasko2go is a Switzerland and Israel based insurtech company specialising in the risk assessment of policyholders. They currently have 35 people and are growing. Arina shares her journey from a pursuit of German teaching to leading a cutting-edge tech company revolutionising the insurance industry. In this interview, Arina provides insights into kasko2go's evolution from a telematics company to a tech-driven force in AI pricing and risk prediction for motor insurance.

Arina, please introduce yourself - what do you do at kasko2go and how did you come to founding this company?

I'm the Co-Founder and CEO of kasko2go. Today it is a group of companies. We started off as a telematics company. Then we moved to AI pricing and risk prediction for motor insurance.?

Our next direction is the cleaning of data for anyone interested in putting their data in order. With motor insurance it is not only correct pricing and risk prediction, but also about such vehicles for marketing and sales like lifetime value of a customer. We are pivoting to a technology company who provides services for a rather wide range of customers.

How did you end up in this industry?

By accident. After school I wanted to be a German teacher, so I started studying German. But then I realised, I don’t like children in those quantities as I have seen them in school. Then I thought about what I like to do - I like to read. I thought about what type of job involves a lot of reading, but also getting paid for. So I began my law studies. After graduation I got a job with KPMG in Zurich. Then I realised that law works together very well with finance. So I also did my postgraduate in corporate finance. And after that I thought that education was enough.?

I’m not a typical insurtech startupper. It’s usually people who come from insurance and see a special demand in some niche and understand they can do it better or people from funds or venture capital. I am none of them, but this makes my job even more interesting for me. And I have a really strong team of experts. I acknowledge that at least 90% of my team members are more competent than I am. That’s a good feeling.?

How have the past few years impacted the insurance industry?

A major impact is that they have new models now. I respect that they reacted fast. For example, most of the cars registered in Ukraine are now somewhere across Europe and insurers don’t really know where it is. So from an insurance perspective, the fact that the car insured by you is not in the intended country, but somewhere else in Europe, requires new models - new calculations of risks and tariffs.?

What are the main origins of loss in motor insurance (for insurance companies)?

The main origins of loss in motor insurance for insurance companies include the traditional factors of mispricing of risk, crime, and competition. In broad terms, we agree with this assessment. However, we believe that a correct assessment of risk is crucial in addressing each of these factors effectively. By accurately assessing risk, insurers can ensure they are pricing policies correctly and avoid discounting high-risk customers unfairly. Additionally, a thorough understanding of areas with significant risks of theft and vandalism is essential. Our approach involves taking a different perspective on risk, allowing us to find innovative ways to better segment problematic drivers and mitigate potential losses in the motor insurance sector.

There are also new risks and events that have emerged in this industry. We try to react to this as fast as possible. One impactful risk is for example inflation. After covid there was a lot of helicopter money and now it is inflation and we have to deal with this. Another issue after covid is that the logistics chain of supply for car spare parts is broken, so it lifted the prices and also additional costs. In my view, insurance companies need to respond quickly to these events.?

Many new risks also bring numerous opportunities. I strongly believe that new challenges require new ways of thinking and new solutions. You have to be proactive. Like for example the historical tables like GLM - they’re stable, but they’re not efficient enough, because they don’t come together with the dynamics of the market.?

Are there specific trends or emerging issues that you believe are contributing to increased losses in recent years?

Yes, there are specific trends and emerging issues that we believe have contributed to increased losses in recent years in the motor insurance industry. The COVID-19 pandemic, particularly its impact on logistics and supply chains, has been a significant factor. The resulting interruption in logistics has affected the availability and pricing of motor parts, leading to mispricing of policies and higher costs for replacement parts.

Furthermore, the pandemic relief funds and the subsequent inflation have had a lasting impact on policyholder behaviour. The economic effects, including inflation, have influenced the overall pricing dynamics of motor insurance policies. This, in turn, has contributed to increased losses as insurers grapple with adapting to the evolving economic landscape.

Moreover, climate change has emerged as another noteworthy factor. The widespread and severe impacts of climate change have led to large-scale damage to cars, creating additional challenges for the motor insurance industry. The changing climate patterns have resulted in?

more frequent and intense weather events, causing damage to vehicles on a grand scale.

How do you think these losses can be mitigated?

To mitigate future losses in the motor insurance industry, adopting advanced technologies, particularly Artificial Intelligence (AI), is crucial. AI can play a pivotal role in navigating the nuanced and evolving risks in the present landscape. By leveraging more advanced modelling techniques, AI can provide a deeper understanding of the complex risk factors associated with motor insurance. AI’s analytical capabilities can help in making sense of the intricate relationships between various variables affecting the industry, such as economic changes, supply chain disruptions, and climate-related risks. This enhanced understanding will enable insurers to develop more accurate risk assessments, allowing for better pricing strategies and risk management.?

In addition, AI can contribute to real-time monitoring and analysis of changes in the insurance portfolio. This proactive approach helps insurers identify and respond quickly to shifts in the market, policyholder behaviour, or external factors, minimising the potential damage caused by these changes. The ability to adapt swiftly to evolving circumstances is a key advantage in mitigating losses.

How can AI help to prevent these losses?

AI can play a pivotal role in loss prevention within the motor insurance industry by leveraging several key capabilities:

Deeper Segmentation for Risk Reduction: AI enables a more sophisticated analysis of data, allowing for deeper segmentation of potential policyholders. By identifying specific characteristics and behaviours associated with lower risk, insurers can selectively target and attract the best possible customers. This not only reduces the overall risk in the market but also enhances the profitability of the insurer’s portfolio.

Innovative Pricing Strategies: AI empowers insurers to implement innovative and unexpected pricing strategies. By analysing extensive datasets and customer behaviours, AI can identify unique pricing models that competitors may find challenging to replicate. This strategic advantage allows insurers to differentiate themselves in the market, potentially attracting a more favourable customer base.

Optimised Marketing Strategies: AI can provide valuable insights into strategic locations where marketing efforts are most effective and pose the least amount of risk. By analysing data on customer behaviour, preferences, and regional risk factors, insurers can tailor their marketing campaigns to target areas with high potential and minimal risk.

Lead Qualification and Customer Lifetime Value Prediction: AI can enhance lead qualification processes by analysing a variety of data points to identify high-value prospects. Additionally, AI can anticipate the Lifetime Value (LTV) of a customer by considering various factors such as historical data, behaviour patterns, and engagement levels. This enables insurers to prioritise leads that are likely to result in long-term, profitable relationships.

Do you use AI in your everyday work??

Yes, absolutely! I like Midjourney a lot. And ofcourse I use ChatGPT as well. With AI, people tend to be a little scared that they will lose their jobs, which I understand - big changes can be frightening. I think everyone has a choice here. You can use the AI to your advantage or not.?

How does your company use data and analytics to identify patterns and potential sources of loss in motor insurance?

Our company places a strong emphasis on utilising data and AI analytics to identify patterns and potential sources of loss in motor insurance. The process begins with the continuous collection of a wide range of relevant data, covering aspects such as geographical situation, market dynamics, economic factors, and external influences like climate patterns. We view data collection as an ongoing process, ensuring that our insights remain current and reflective of the evolving motor insurance landscape.

Data quality is of utmost importance to us, and we dedicate significant resources to thorough data cleaning. This meticulous process eliminates inaccuracies, inconsistencies, and outliers, ensuring that our analytics are built on reliable and accurate information.?

Recognizing that different types of data require specific modelling approaches, we use modelling tools tailored to the characteristics of each dataset. Whether it’s customer demographics, historical claims data, or market trends, our AI analyses are precise and account for the unique aspects of each data type.

Our goal is to achieve a holistic understanding of risk in the motor insurance sector by diligently collecting, cleaning, and modelling diverse datasets. This comprehensive view enables us to identify patterns, anticipate potential sources of loss, and react proactively to changes in the market.

With a solid foundation in data and AI analytics, our company is better equipped to react quickly to the changing state of the market. Whether it’s shifts in customer behaviour, emerging risks, or economic fluctuations, our ability to analyse data empowers us to make informed decisions and adapt our strategies promptly.

How do you get good quality data?

Nothing good comes easy, to be honest. Before we receive data from the customer, we collect the data ourselves. We have a huge data lake, where we have over 800 parameters (like weather, events, traffic, criminal situation, crash test of the car and so on). We have a whole department called Data Search, who works on this data and cleans it. It is also important to acknowledge the relevance and importance of the data. For example, statistics of events is not as important as the crash test of the model.?

Cleaning the data is also a challenge. Businesses usually learn how to collect the data and more or less learn how to store it, but not clean it. Approximately 80% of our data engineers' time is dedicated to cleaning the data. Data is the new gold, as they say, so we want it clean. We want that the outlier data we receive is as minimal as possible (the data where we don’t know what’s going on; for example, the driver is 180 years old and lives on a street that doesn’t exist). It’s because the data forms are usually completed by people. So it’s important to clean the data and harmonise it, so it makes sense.?

Then we take the insurance company’s data and enrich it with our data. Then our neural networks come in, that try to combine our parameters with their data to understand what is the most optimal model for them. We don’t have a universal model for everyone - we create separate models for every new customer. We respect their preferences and unique situations.?

Have you measured the impact of using your AI tech in your clients’ businesses?

We have calculated that our return on investment is 1500%. So for every euro a client pays us, they will save at least 1500 euros. When we do our predictions and make our models, we can also show what it means in financials: how much money they can save and how much more money they will gain.

What is the impact of your AI on insurance companies’ clients?

From the statistics of our customers, we see that very often, independently from the company or the country, the very good and safe drivers pay much higher policy price than the risky drivers. We have seen it very often. So good drivers pay for bad drivers in some way. We can’t break this trend. Otherwise the insurance companies will lose customers and market share, and we don’t want it.?

We recommend being aware of the situation and working with it: customise the prices and charge the risky drivers more and give discounts to good drivers. AI assists in providing personalised offers and pricing. It is possible to give personalised offers to each customer, but it is up to the insurance company how much they want to granulate their pricing models.?

Looking ahead, what do you foresee as potential future challenges in the motor insurance sector, and how is your company preparing to adapt and minimise losses?

Looking ahead, a potential future challenge in the motor insurance sector is the shift towards electric vehicles (EVs). The transition to EVs presents a substantial unknown in the market, with implications for risk assessment, pricing models, and overall market dynamics. As a service provider collaborating with multiple companies in the motor insurance sector, we recognize the importance of staying ahead of these challenges.?

Our company operates with a commitment to data privacy and integrity. While we would never mix datasets of our customers or utilise models of one customer for the benefit of another, our unique position allows us to learn about market situations more quickly than other players. Our quick comprehension of emerging risks allows us to help customers adjust their models promptly, ensuring they can adapt to new situations ahead of others in the market.