By Hamzah Chaudhary1
1Director of Product Management, Cytora
Insurance is a product that has remained unchanged almost since its inception. This filters through to the way in which it is transacted today. The majority of commercial insurance is still done in person or over the phone. Although a lot of personal lines insurance (e.g. car insurance, life insurance) is now sold online in a fast and automated way, the same changes have not been adopted in commercial insurance.
There are a number of factors that play into why this should be, but the major one is that commercial risks are often seen as far more heterogeneous than personal lines, even at the smallest business level. Insurers therefore require human underwriters to be involved in understanding the true nature of the business, and to price accordingly.
However, as customer demands change, this system is no longer acceptable to the end user. As individuals, we expect fast and frictionless processes, and this is something that is currently not supported by the insurance infrastructure in place for commercial underwriting. Consumers expect all transactions to be simple and automated, regardless of whether they are buying a car insurance policy or a policy for their business. This means that commercial insurers must change how they do business to meet customer demands.
Also, underwriting decisions today are often made using incomplete and outdated information. Paired with legacy systems that make it difficult to automate processes, it is easy to see how the commercial insurance workflow today is far below the expected standard. It also means that the frictional cost of operating in these markets is extremely high for insurers, and often the cost of change can be even higher.
Like many sectors, insurance has also seen some companies already attempt to incorporate artificial intelligence applications into parts of their business. This article walks through a few specific use cases that AI can enable to transform the customer experience and drive productivity growth. The focus is on applications that exist today, rather than future possibilities dependent on technological developments.
At a high level, there are three near-term use cases powered by AI that are within the grasp of implementation by insurers today, and that impact different areas of the underwriting value chain:
Using AI to structure incoming submissions from brokers and customers to help teams prioritize their work. The majority of commercially underwritten policies are intermediated transactions, often with a broker between the customer and the insurer. Typically, brokers will fill out a submission form containing relevant information which is then sent via email to an underwriter. Underwriters must go back to the broker if they have questions on the submission or require any extra information. This can often mean it may take days to get a price back to the customer. It is also an extremely unoptimized process, as underwriting teams must go through each submission manually to understand if it is within their business appetite.
Using computer vision and natural language processing (NLP) technology, information can be automatically extracted and structured from broker submissions and put into a queue for underwriters. Underwriting managers can then easily prioritize these submissions around the kind of business they want to write. This means that brokers get faster responses, and underwriters focus their time on business that is most relevant to them.
Use advanced matching algorithms to link external data as well as advanced modelling techniques to gain a more holistic risk profile of a customer. Using advanced matching algorithms to link disparate data sources helps leverage the power of alternative data to gain a more holistic risk profile of a customer. Leveraging external data sources not traditionally used also means data can be accessed that the customer may not even know about themselves. As an example, when insuring properties, fire can be a major hazard, so looking at the distance to the nearest fire station may be beneficial. Pairing these new data sources with advanced machine learning modelling techniques can provide a much more accurate price for an individual customer.
This helps create a fairer risk price, using comprehensive, contextualized data, avoiding biases and discrimination. This also means a future where the user journey is more frictionless than today, while still ensuring customers are properly understood by the insurer.
Leverage data collected from connected devices within machine learning models to provide advanced and tailored risk management to customers. Risk management is about realizing changes that a customer can make to their insurable property and advising them on the best way to rectify this. Often, this can result in a lower premium for customers due to the reduced probability of a claim. However, this process is mostly manually carried out by risk engineers who have to conduct on-site assessments, and it just isn’t economically viable for most insurance companies to carry out these assessments on smaller customers.
However, with the data collected from new connected devices, the power of advanced risk management can be available to all customers. Take the example of connected smoke alarms: if a customer’s smoke alarm goes off a few times in a week, the insurer can send them a notification with tips on how to reduce these false alarm incidents to avoid a fire.
Collecting data directly from customers like this means that insurers have information that is directly relevant to the assets they are insuring. Over time, they can train machine learning models on data from connected devices that allow them to serve up relevant and real-time risk management information to all customers. This can help avoid claims and save money for customers.
There is a sliding scale of how AI can help commercial insurers through different classes of risks. In the smaller end of the market – known as SME businesses – it can help to achieve full automation of the underwriting process, and as risks get larger and more complex, it becomes a tool to augment the capabilities of an underwriting team.
Leveraging the power of AI, insurers can offer their customers effortless access to fairly priced insurance, whenever they need it. AI can be used systematically along all parts of the underwriting value chain to achieve this. The future of insurance has to be one where underwriting teams are empowered by AI-driven tools.