Spotlight: How insurers can stay ahead in an AI-driven world

Despite reports of AI innovation and adoption accelerating, and substantial investment by insurers, value realisation has been slower than expected. WTW’s Pardeep Bassi explains the potential of automated model monitoring and insight generation, and how insurers can leverage AI more effectively to enhance pricing, portfolio management, underwriting and claims and, ultimately, improve business performance.

The predictive edge

In the fiercely competitive insurance industry, gaining and maintaining a competitive edge is crucial for companies to thrive. But with many insurers looking the same at their core, supported by the same foundations, the same core processes and similar products, where can insurers differentiate themselves?

As markets and risk grow increasingly complex and the need to stay in control and on top is becoming increasingly important, leveraging AI is proving to be a game changer. Or, more specifically, an insurer’s ability to successfully harness AI to improve and evolve their portfolio management, pricing, claims and underwriting capabilities.

The latest AI-driven solutions can automatically track emerging experience on a monthly, weekly, daily and even hourly basis.

Pardeep Bassi, WTW

The latest AI-driven solutions can automatically track emerging experience on a monthly, weekly, daily and even hourly basis, as well as being designed to help insurers monitor model performance and identify previously unidentifiable trends.

This allows portfolio managers and underwriting, claims and pricing teams to automatically perform calculations and assessments using emerging data and bespoke proprietary AI algorithms, delivering critical actionable insights swiftly and efficiently. This enables risks to be neutralised before they become material, and maximisation of any advantage.

Meanwhile, the ability to identify, isolate and understand key trends and multi-dimensional segmental changes, using a range of underwriting, rating and claims features, helps to cut through complexity to drive actionable insights faster. Critically, insurers also need to monitor model health, identifying degrading predictive models and their commercial impact, to help make judgements quickly and confidently.

Barriers in the pursuit of value

Managing fast-expanding model real estates has become a source of frustration for senior executives while putting additional strain on analytical teams – an already scarce and valuable resource – who are having to spend considerable time and effort on ‘model administration’ instead of value-adding activities that drive business performance.

The sheer number of models is not the only concern. The complexity and variety of model forms have also increased dramatically in recent years. Machine learning models, in particular, tend to become outdated more quickly than more traditional statistical models, with increased use also leading to pricing instability for pricing teams.

While machine learning algorithms provide additional predictive power, this will often come at the cost of reduced longevity if not maintained. Companies may implement machine learning models and then not update them for months or even years. This may have worked in the past, but it won’t work in today’s rapidly evolving market.

As models become outdated faster, pricing and analytical teams will struggle to quickly identify which models are aligned with shifting market conditions. Reliance on obsolete and less stable models increases the risk of insurers making suboptimal and incorrect decisions. With AI expected to shape every insurance decision and impact every process, insurers will find themselves increasingly exposed to these risks.

Machine learning models, in particular, tend to become outdated more quickly than more traditional statistical models, with increased use also leading to pricing instability.

Helping motor insurers see the unseen

The recent decline in motor claim frequency in the UK market led some insurers to speculatively attribute this trend solely to weather patterns or to the reduced likelihood of drivers making claims due to higher excesses. By adopting a more granular and sophisticated approach, insurers would have been able to identify a greater number of individual underlying components driving this trend, comprised of a combination of weather, excesses, speed limit reductions and vehicle technology adoption.

This deeper analysis affords a more comprehensive understanding of the factors at play. By uncovering previously inaccessible insights, it becomes easier to predict future changes in claims frequency changes and set and recalibrate prices for the next 12 months with increased accuracy and confidence.

AI-driven monitoring for smarter and faster insights

In today’s market, effective performance already requires empowering pricing and analytical teams with capabilities that extend beyond simply ‘turning the handles’ of model development. Success in this AI and model-driven world is determined by those that understand the strengths and weaknesses of their models in line with changing and emerging trends.

Models need to be designed, built and fine-tuned to allow insurers to manage them with greater control and governance, supported by insurance-focused technology that provides a comprehensive monitoring system for the entire business.

This will enable more forensic insight into potential risks and facilitate the early identification and resolution of emerging issues. Insurers will also be in a stronger position to mitigate risks associated with underperforming segments, identify opportunities for growth and prioritise the most relevant actions based on the most desirable business outcomes.

Successful insurers differentiate themselves through better understanding and by acting on changing experience more quickly than their competitors.

The capability to consistently monitor the health and effectiveness of hundreds of models – overall and segmentally – using sophisticated machine learning algorithms designed for insurance monitoring, will give insurers the reassurance they need that their burgeoning model real estates remain fit for purpose.

Implementation of best-in-class model governance and risk management will enhance an insurer’s confidence in managing expansive, complex model real estates. This is achieved through the alignment of analytical activities across multiple functions with standardised processes, language and activity.

Insurers can further boost their analytics teams’ effectiveness by automating tasks, tracking models and key business metrics, and integrating this monitoring into their analytics lifecycle. This approach allows a more efficient allocation of costly analytical resources, enabling greater focus on value-added activities, while minimising unnecessary complexity.

Successful insurers differentiate themselves through better understanding and by acting on changing experience more quickly than their competitors. Such agility is challenging when you’re deploying and maintaining large and complex model real estates while dealing with rapidly changing and volatile market forces. Instead, leverage AI for automated model monitoring and insight generation to stay ahead.

Sponsored content