Achieving pricing maturity in insurance - A digital transformation roadmap

A Mindtree White Paper  Achieving pricing maturity in insurance - A digital transformation roadmap

 Summary

 How leaders in pricing operate

 The foundation – Consistent use of ‘unconstrained' generalized linear models in underwriting processes

 Institutionalizing AI in pricing

 Dynamic pricing

 Product simplification

 Full-scale transformation

 Program bird’s eye view

 Benefits

 Conclusion Summary

Pricing optimization is the most significant driver of sales and underwriting profit, and an imperative in these unprecedented times as Insurers struggle to provide value. Traditionally, pricing has been a ‘walled tower’ exercise undertaken by underwriters and product manufacturers based on periodic input from the market. Leaders in pricing however, are doing much more to change this traditional approach as a recent study from McKinsey1 put forth. This whitepaper operationalizes these findings into a roadmap for pricing innovation for organizations.

How leaders in pricing operate

From creating events that help understand purchase causality to promoting a culture of pricing awareness, industry leaders in pricing are changing the rules of the game. This approach in turn is reaping rich dividends. In the era of low interest rates and commoditization of innovation, sharp pricing driving underwriting profit is the best driver of operating profit and return on investment. McKinsey outlines five levels of pricing maturity in their study, and here, we attempt to operationalize a pathway to achieving them.

1 The post-COVID-19 pricing imperative for P&C insurers – McKinsey Insurance Practice Insurance carriers can be classified by level of pricing innovation and transformation.

Levels of pricing sophistication

Consistent Use of AI-based or Product Implementation Full-scale pricing application of machine-learing 2 simplification for 1 of robo-pricing transformation GLMs pricing tool the right pricing Description Consistently Implements In addition to using Adjusts product Approaches pricing applies AI-based, GLMs and AI, structure and transformation “unconstrained” automated pricing understands eliminate highly comprehensively, GLMs to all risk model to leapfrog real-time market unprofitable tariff including current models and rate-making dynamics and uses cells and product capabilities, people product types process using robotics for modules; enables culture, GLMs; ongoing, automatic end-to-end product organizational alternatively, could pricing including value chain design and build on existing sales development optimization and structure, data and GLMs behavioral pricing tech strategy

1 Generalized linear models. 2 Unavailable in the United States.

Figure 1: Pricing innovation and transformation maturity

The foundation – consistent use of ‘unconstrained’ generalized linear models in underwriting

Underwriting used to be known as both an art and science. However, with the growth of predictive analytics and machine learning, the verdict is clear – science is where the future lies. Moving beyond simple linear rate cards, a solid foundation to move towards pricing maturity lies in using generalized linear models with unconstrained variables in the pricing equation. This is a starting point in pricing maturity and more than half of the P&C companies in the US and Europe have already established this as a foundation (and therefore are in good stead). For those that have not, the adoption of a GLM (Generalized Linear Model) software in underwriting is table stakes.

Institutionalizing AI in pricing

Introducing Artificial Intelligence (AI) into the pricing process allows companies to leapfrog into sophisticated pricing. Typically, historical data from policy admin and sales CRM system is used to train the model and drive pricing changes in annual filings. An acceptable step taken in today’s environment is to invest for future scale and build AI in the cloud from the get-go (as opposed to an offline solution). This can lay the foundation for real-time inputs and dynamic pricing in the near future. Machine Learning

Automation Neural network Supervised learning Natural Deep Learning Classification language Speech Recognition Unsupervised Recommendation Processing learning Personalization Linguistics Intent Regression Autonomy Data Science Sentiment Platforms Analytics Dialog Sarcasm Computer Vision Business Intelligence

Figure 2: The Mindtree AI practice - areas of expertise

PURCHASE UNDERWRITING SERVICE CLAIMS RENEWAL

BILLING

SALES CONTACT CENTER SERVICE CONTACT CENTER DIGITAL CX PLATFORM eSERVICE CLAIMS CX CRM

POLICY ADMIN SYSTEM PRICING A.I. ENTERPRISE DATA LAKE

Figure 3: Systems in play to integrate with a cloud-based AI solution for pricing Dynamic pricing

The evolution of pricing maturity leads to dynamic pricing. To enable an effective solution, it is essential to draw upon a wide range of additional data to drive faster and meaningful learning. Here, AI establishes pricing in real-time based on a continuous stream of relevant parameters from sales channels coupled with historical data on purchases, claims and renewals. Dynamic offers are made to prospects depending on their propensity to purchase as well as the risk they represent.

Examples of additional data may include demographics (age, marital status, gender) and web metrics (media source, time on site, frequency of visit) from the digital channel, service levels (call back time, time to answer) and engagement levels (call sentiment analysis, time on call) from the contact center and additional customer behavior data from third-party services.

Additionally, the growth in telematics /IoT based insurance and pay-as-you-go product development has led to further proliferation of once unknown data points that can monitor risk in real-time and feed a dynamic pricing AI.

Driving History rd IOT 3 PARTY DATA Social Profile Usage Life Events Geolocation Online Behavior t-if” an ha aly “W sis

DIGITAL CONTACT PRICING A.I. CLAIM RENEWAL CENTER

Demograohics Engagement # of Incidents Rate Appetite

Web Metrics Call Back Time Average Value Cross-sell

Media Source Qualifiers Risk Category Channel

rd 3 Party Data Coverage Deductible CLTV

Figure 4: Real-time feedback and potential data points that can feed the dynamic pricing AI model About dynamic pricing Our interpretation of dynamic pricing, given the US’s regulatory laws, is more of a multi-faceted product combination covering product versions, riders and coverage levels. These are facets of prior filed products that are stitched together by AI, factoring in optimal uptake price and maximized underwriting profitability.

Coverage

Version Riders

Figure 5: Configurable parameters that impact end-customer pricing

PURCHASE UNDERWRITING SERVICE CLAIMS RENEWAL

BILLING

SALES CONTACT CENTER SERVICE CONTACT CENTER DIGITAL CX PLATFORM eSERVICE CLAIMS CX CRM

POLICY ADMIN SYSTEM PRICING A.I. ENTERPRISE DATA LAKE

IoT PLATFORMS 3RD PARTY DATA AGGERGATORS

Figure 6: Systems in play to integrate for an AI led dynamic pricing model Mindtree case study

Real-time pricing platform for a leading European Airline

The ask: How to leverage AI at scale to optimize pricing in real time at a ‘per-seat’ level?

The solution: Mindtree launched an AI platform using distributed scaling architecture in the cloud, integrated it with multiple real-time data sources and scaled the model to 17000+ onward and destination markets, driving increases in revenue and yield.

Product simplification

As pricing maturity evolves, the focus on profitable cells and products that meet emerging needs benefit from richer and real-time inputs. Organizations with simplified portfolios benefit from focused targeting and lower operational costs, driving further ROI. Mature organizations institutionalize the feedback loop between pricing intelligence and new product development so that they can continuously optimize their product portfolios. A further necessary step is to manage these products on modern policy administration systems where feature development, policy issuance, billing, commission accounting and filing with state entities is carried out using a ‘no-code’2 approach.

Mindtree case study

Product rationalization and simplification when moving to a Policy Admin System (PAS) for a large P&C carrier.

The ask: Simplify and migrate a portfolio of over 150 different commercial P&C products to a modern Policy Admin System (PAS).

The solution: Once the carrier rationalized the portfolio after a profitability review exercise, Mindtree designed a template-based variance approach and launched a factory model to move these products to the PAS. A configuration-based approach and Mindtree’s expert knowledge of the domain and technology ensured product migration in three months vs. the earlier model of nine months.

2'no code': Platforms that allow developers and advanced skill non-technical users to develop custom implementations using a “code-less” configuration console. Full-scale transformation

Full-scale transformation into a mature pricing-led organization involves further cultural changes and an organization-wide governance structure that drives everything from a periodic review of market insights, to pricing changes, to championing investment in technology. All of these drive competitive advantage in pricing.

Program bird’s eye view

Mindtree envisions a multi-year program where infrastructure and technology are rolled out and institutionalized across the organization. Maturity is built layer by layer and a long-term strategy is embraced in order to extract benefits down the road.

Phase 1: Adopting unconstrained generalized linear models in pricing

 More than 50% of P&C insurers have achieved this level of sophistication and various sophisticated models are provided through the likes of Willis Towers Watson, Milliman, Prophet and other actuarial software providers for those that need to adopt this level. These are historically on-prem products with newer versions presented either stand-alone or through packaged solutions in the cloud. Phase 2: Using AI for pricing

 Again, actuarial software providers have gained ground in this, given their historical role. For example, “Emblem” by Willis Towers Watson comes packaged with machine learning models for companies with large datasets in customer and claims data.

 The option exists of course, to build custom models where internal 'tribal' knowledge and data science can work together to build models fine-tuned to the company’s DNA and risk appetite. Tensor Flow, Watson Analytics and Sage Maker can be deployed in the cloud and integrated with a corporate data-lake by an IT services vendor to continuously train your pricing AIs.

 A corporate-wide data lake strategy is also a necessary step either before or when you reach this stage. This sets the stage for consolidating data into one central area and simplifies programming models for the AI.

Phase 3: Dynamic pricing

 A move to dynamic pricing models provides (and needs) insurance companies to significantly expand the amount of data they track. The creation and capture of additional marketing events is well supported by a sophisticated marketing operations program, which in turn benefits from back-end input. Mindtree defines a mature markops program along the dimensions of campaign management, operational execution and marketing analytics and supports the implementation of platforms like AEM and Salesforce Marketing Cloud.

 IoT is rapidly becoming a necessary component in any P&C offering. Regardless of whether companies currently offer behavioral pricing based products to customers (which will rapidly emerge post the COVID-19 experience), launching safe driving or home monitoring pilots is a must have to build the dynamic pricing products of the future.

Phase 4: Portfolio optimization

 We don’t see this necessarily as a phased event, but rather a continuous process companies should embrace to optimize profitability. Modern policy administration platforms like Duck Creek are essential tools in ensuring that desired changes are quickly rolled out. Inheritance models and default templates in Duck Creek significantly reduce turnaround times for filing new products. Level of pricing sophistication

Consistent Use of AI-based or Product Implementation Full-scale pricing application of machine-learing 2 simplification for 1 of robo-pricing transformation GLMs pricing tool the right pricing Description Consistently Implements In addition to using Adjusts product Approaches pricing applies AI-based, GLMs and AI, structure and transformation “unconstrained” automated pricing understands eliminate highly comprehensively, GLMs to all risk model to leapfrog real-time market unprofitable tariff including current models and rate-making dynamics and uses cells and product capabilities, people product types process using robotics for modules; enables culture, GLMs; ongoing, automatic end-to-end product organizational alternatively, could pricing including value chain design and build on existing sales development optimization and structure, data and GLMs behavioral pricing tech strategy

Implementation of GLMs Integration of 3rd party data

Building / Augmenting Data Lake Implementing Pricing A.I.

Create and capture measurable events Real-time Pricing A.I. A.I. in the cloud

Feedback led portfolio rationalization Faster Portfolio Optimization in PAS

Build organization wide pricing focused culture Optimize Pricing Governance

Figure 7: A representative view of a multi-year technology program to achieve pricing excellence

Benefits

Adoption of a pricing maturity approach yields results at every stage of the maturity continuum. There is an immediate impact on loss ratio improvement by moving to GLMS and AI-based pricing, and real-time pricing drives higher overall profitability. Full scale transformation has a significant impact on the combined ratio, probably because companies embrace a pricing focus at every level of the organization and view investment in training, technology and talent through this lens. Additional benefits in premium increase, retention and anti-selection (the lack thereof) are also seen. Level of pricing sophistication

Consistent Use of AI-based or Product Implementation Full-scale pricing application of machine-learing 2 simplification for 1 of robo-pricing transformation GLMs pricing tool the right pricing Impact Improved loss ratio Improved loss ratio Higher new Double-digit 3-6pp observed of new business by of new business by business growth rates of of combined ratio 3 0.8-1.5 pp 2.1-4.2 pp profitability by new business year improvement 2-4 pp of over year, while Improved loss ratio Improved loss ratio combined ratio loss and cost ratio 3-4% 4 of renewal of renewal are improved by Additional GWP business by business by Higher new growth business 1 pp 0.8-1.5 pp 0.6-1.3 pp premiums of Reduces severe cross-subsidization 10-15% of more than Improved 20% retention by and therefore 10-12% anti-selection for 10-15% of the portfolio

Figure 8: Quantification of benefits by McKinsey Inc.

Conclusion

Technology is an essential ingredient in moving an insurer to pricing maturity. Companies should consider that a point solution alone will not solve the problem. Instead this calls for an investment in a technology roadmap that delivers an infrastructure capable of executing this at scale. The good news is that investment in policy admin systems, AI solutions and marketing operation platforms (for starters) – all considerably important to any CIO – yield immediate benefits for pricing maturity (provided this is contemplated in the roadmap). If anything, this should cement purchase and implementation decisions in these focus areas to prepare insurers for adapting to changing times.

Riddhish Trivedi

General Manager

Riddhish has a deep background in Digital, Insurance and Direct-to-Consumer marketing. His career experience spans across various roles, including Global Head of Broad Market, Head of Digital Experience and VP of eCommerce. He has wide-spread experience in building consulting practices, re-launching brands in the market and collaborating with key stakeholders across geographies. Riddhish is an MBA alumnus of SP Jain in Mumbai and has a degree in Chemical Engineering from the University of Pune. About Mindtree

Mindtree [NSE: MINDTREE] is a global technology consulting and services company, helping enterprises marry scale with agility to achieve competitive advantage. “Born digital,” in 1999 and now a Larsen & Toubro Group Company, Mindtree applies its deep domain knowledge to 280+ enterprise client engagements to break down silos, make sense of digital complexity and bring new initiatives to market faster. We enable IT to move at the speed of business, leveraging emerging technologies and the efficiencies of Continuous Delivery to spur business innovation. Operating in more than 15 countries across the world, we’re consistently regarded as one of the best places to work, embodied every day by our winning culture made up of over 21,800 entrepreneurial, collaborative and dedicated “Mindtree Minds”. www.mindtree.com ©Mindtree 2020