While more P&C insurers are embracing data, artificial intelligence (AI) and machine learning as part of their digital strategy, we often hear them voice frustration about generating business outcomes from these investments. Many find it difficult to tie their AI initiatives to key performance indicators (KPI) and profitable growth.

As they gear up to implement AI in the underwriting and claims processes, here are three changes P&C insurers need to make to get the business results they’re looking for.

  1. Think big, and dismantle business siloes

    Insurers typically take a piecemeal approach to AI. We’ve seen clients pursue interesting automation applications, such as leveraging predictive analytics and intelligent algorithms to streamline the quote delivery process from five days to two. That’s a decent ROI, but AI can go far beyond incremental improvements and fundamentally change the way insurers do business.

    For example, more data than ever will soon be available to P&C insurers through Internet of Things (IoT) sensors embedded in everything from vehicles and machinery, to safety devices like helmets and belts. By applying AI to the data in real-time, including machine learning algorithms, insurers can offer usage-based insurance products.  

    We’re already seeing pay-as-you-go pricing in auto insurance. While initial products target low-mileage drivers, continuous data feeds can lead to real-time usage-based policies, such as automatically beefing up coverage for drivers as they encounter bad weather, for example, or find themselves in bumper-to-bumper congestion.

    In the P&C arena, we’re partnering with insurance clients to explore how they can apply AI-based insights to asset use to create products that adjust coverage based on the number of employees in an office or customers in a store. When a store experiences higher footfall, for example, liability coverage would increase. When employee occupancy rises, or the nature of work is more injury-prone or hazardous, workers’ compensation coverage can track upwards.

    To create policies that flex in real-time, insurers need to bridge the longstanding gulf between business units. Breaking down these silos, however, requires dramatically different thinking. It means preparing an organization to succeed in the future, when customer engagement will no longer be focused on claims or billing-focused interactions but on “experiences” built around individual customers and businesses.

    With algorithms embedded into everything, and data freed from siloes and disconnected systems, insurers’ underwriting units will know, for example, that a small-business owner covered by the company’s commercial line is also a personal-line customer who maintains primary and secondary residences – and create a tailored policy.

  1. Be patient about seeing results

    It’s common for intelligent automation initiatives to compete for budget dollars with more conventional automation projects that provide 200% savings in a year. AI and machine learning, however, typically don’t provide next-quarter or even short-term returns. Because AI’s game-changing role is to produce business value rather than efficiency, its success is measured by more sophisticated metrics that can take longer to materialize.

    Take the contact center, for example, where AI can predict the questions callers will ask or escalate calls based on the questions asked or the caller’s tone of voice. While these complex systems take time to train and implement, the payoff is as important as cost reductions. Since our team implemented the IBM Watson cognitive platform for a global P&C insurer, the carrier now drives more than 50% of its high-volume, low-value inquiries (such as policy expiration dates) through a self-service chatbot. Use of advanced natural language processing (NLP) techniques and real-time feedback enable the service team to be more empathetic and engaging with customers, improving net promoter scores significantly.

    One way to demonstrate returns and validate AI initiatives is to break projects into segments, each with its own tangible results. For example, a KPI for the first phase of a project might measure a two-point lift in customer satisfaction. This approach requires the willingness to continually adjust as the project advances, recalibrating at the end of a feature release or an MVP (minimum viable product). 

  1.  Gather the data AI needs

    AI insights are only as good as the data fed into the system. AI algorithms depend on large datasets to reveal patterns. Insurers have lots of data on hand, but most of it is unstructured, from scanned documents to notes from physicians, risk engineers, underwriters and claims adjusters. Digitizing this historical data is essential to laying the foundation for AI.

    We see insurers that are serious about AI converting written text into digital data stored in data warehouses and data lakes. Because studying patterns is especially important for injury and liability claims that can result in significant payouts, the greatest progress is being made in workers comp and liability, where AI insights can help carriers better understand which claims to escalate.

    Equally critical for insurers is creating a data ecosystem. The idea of pooling data through partnerships with manufacturers and other carriers is a challenge for insurers, which traditionally prefer to own everything. But change is happening. For example, Tesla has launched an insurance service to share data on its electric vehicles with carriers. Modernizing assets to take advantage of AI means constant renewal, and we’re seeing clients begin to break from tradition as they explore how to better leverage third-party data with IoT and telematics data.

AI has the potential to be so much more to P&C insurers than just an efficiency boost. Meeting business KPIs is possible – it just takes new ways of thinking to attain them.

Agil Francis

Agil Francis

Agil Francis leads Cognizant’s Property and Casualty (P&C) insurance advisory practice, focused on business and digital transformation. He has helped develop leading... Read more