ARTIFICIAL INTELLIGENCE IN INSURANCE

Transcription

1 ARTIFICIAL INTELLIGENCE IN INSURANCE Optimizing Relationships and Insurance Results A PEGA INSURANCE WHITEPAPER

2 Executive Summary Insurers are struggling to achieve profitable growth. There is constant pressure to launch new products and be responsive to the market as it develops; there is additional pressure to drive efficiency and reduce costs, without sacrificing quality. There s a renewed emphasis to improve the nature of the relationship with the client and agent, including supporting both across multiple channels seamlessly with a consistent experience. With the rise of the internet of things driving a fundamental shift in how insureds research and interact with businesses and individuals, insurers know they need to transform not just to keep pace with technology, but to leverage it as a real competitive differentiator. How do you manage all of this? Insurers need to act from a position of insight insight into both their clients needs and their agents needs. To do this successfully, insurers need a line-of-sight into how all channels are performing, how each product and line of business is contributing, and why their business strategies are succeeding or not. However, insight itself is not enough. Insurers need to be able to convert that insight into action, and require a mechanism to make this possible. Insight without action is useless just wasted potential and those actions must drive positive outcomes. Insurers today often rely upon a strategy of hope hoping that their organizations are generating the right insights, and in turn taking the right actions they rely upon raw data and basic reports to inform their staff, and hope that their people can generate the insight needed to propel the business forward. Insurers hope that the insight derived will result in positive action, and hope that those actions align with company s strategic goals. The fact is: people are inconsistent. They don t always have the time to sit down and analyze the information available, or fully evaluate the full business landscape. They re often missing the requisite skills, are pressed for time, or work from assumptions that don t align with the larger strategic vision. The human element will always be a critical component to business growth - but the velocity of change in modern business has created new requirements for success. Insurers must provide their people with the tools they need to adapt insurers need to provide a stronger, more flexible foundation one that can scale to the pace of change. This baseline, helping them convert insights into action, and action into value is artificial intelligence. This paper will highlight the various scenarios in which artificial intelligence can be leveraged optimally within insurance operations. It will cover core insurance processes and strategies such as the acquisition of new business, retention and account penetration, marketing and channel management, and claims. 2

3 New Business The goal of every interaction with a prospect or client is to strike a delicate balance: create value for both the client, and the insurer. Every interaction must be seen as a tipping point, where we either bring that client closer, or push them further away. We want to optimize that relationship. Optimizing a portfolio is not the same as maximizing the premium. Maximizing premiums may result in over-selling the insured, which often results in both a shortened client relationship, and negative client perceptions. Underselling the client can be just as bad. The insured may be saving a few dollars but if they don t have the coverage they need when a loss occurs, severe client dissatisfaction is sure to follow. An optimized portfolio will provide the correct balance of products to meet the client s real needs at a price point and terms that seems fair, while providing a desirable level of profit for the insurer. Over the long-term, this results in a mutually beneficial relationship for both parties allowing insurers to maximize that client s lifetime value to the insurer. When dealing with clients, or when supporting independent agents/broker interactions with clients, there is no single best way to sell a product. Every interaction is different, each client has different needs, and these can change quickly. Additionally, all channels do not perform equally in all scenarios what works in one context might not work in another. The industry is moving away from a one size fits all, cookie cutter approach to sales. Rather, the preference is to position a personalized offer to the client that best fits their needs based upon what we know about them, what they re likely to do, and how that will advance the relationship over time. Part of this process is to create an actionable environment where recommendations are distributed to the agent, client or Customer Service Representative (CSR) depending upon both the channel, and the nature of the interaction. These recommendations are delivered using a Next Best Action approach that optimizes performance across all channels. Product Recommendations & Suitability Part of every sales cycle is determining what products, limits, and terms are the best fit for each individual customer. This must take into account the client s unique circumstances (both financially and from a life-stage perspective) to adequately serve and protect that client. In most cases, insurers rely heavily upon the experience and the training of the agent or Customer Service Representative (CSR) to assess the situation, and make the correct product/coverage recommendations. In direct online sales, the majority of this responsibility falls upon the client. Results from this can vary widely, with varying degrees of success both in terms of appropriateness, and adequate coverage. By leveraging artificial intelligence in conjunction with recommendation strategies, all channels can benefit from a consistent, client-centric approach to product recommendations. The AI will develop several recommendations and selects the best recommendation for that client based on their propensity to purchase and potential value - leveraging data from the client s profile, past transactions, and previously accepted/ rejected offers from both the client and clients with similar profiles regardless of the channel in which those occurred. Those recommendations are pushed to the client/agent/csr in the context of the channel, exactly when they re needed always leveraging the most current information about the client and their circumstances. This process is designed to optimize the value of the transaction for both client and insurer, and reduce friction within the process. 3

4 Cross-Sell & Upsell The best position for an insurer is when the client has multiple policies; not only does this improve the premium position, but decreases the likelihood of an insured severing the relationship and moving their business to another carrier. They become, in effect, stickier. Opportunities to expand the relationship into multiple policies do not come often, and when they do, there is a narrow window of opportunity within which the insurer can identify the potential, take action, and expand that relationship. Regardless of the channel in which the client interacts, AI can be used to identify the opportunity, and based on the client s situation, make a well-timed and relevant recommendation to the insured and/or their agent. The system driving those recommendations must constantly be evaluating the situation (whether on the web, in the call-center, through an agent desktop / portal or any other channel), and asking questions such as: Are there additional products and coverages that the insured will benefit from? Which are they most likely to purchase? Is this the right time to recommend that policy? It must then make that recommendation in the context of the channel and only when appropriate. Recommendations for additional products should be purpose-fit to both the client and the sales situation. As the client interacts with the channel, recommendations (and the resulting responses) will often change based upon the context of the situation. By gathering data from every interaction, AI learns which offers are best-fit for which clients and situations and changes its approach accordingly. Sales Actions Every sales cycle has a rhythm and timing all its own. It will vary by channel, market, product, client demographics, and a host of other variables. Top-performing agents will be able to sense these requirements and adjust their approach to match the needs of the sale, and bind the coverage. Not all agents, though, are top performers and in the case of on-line sales, there is no agent at all. How do you optimize all of your interactions? Artificial intelligence can be leveraged to support each agent and channel, tracking each sales cycle and potential opportunity, learning from past interactions, then integrate recommendations into the workflow and associated tasks. Some of the areas that can be addressed are: Objection handling Enumerating the benefits of coverage/limits offered Product Recommendations Interaction Strategy when and how to reach out to that specific client 4

5 Retention and Account Penetration While retention will never be at 100%, all insurers are looking at what they can do to improve retention/ persistency across their client base. Retention is about more than just price or great customer service it s about understanding client needs and context, aligning insurance business processes to support those needs making the insurer easy to do business with. Part of a solid retention strategy is developing a close working relationship with the insured, to foster a pro-active relationship that the insured attaches value to. Attrition / Lapse Risk The hardest time to save a client is after they ve told you that they want to cancel a policy, or have already lapsed. Allowing this is happen is a reactive approach that doesn t resonate with modern consumers, who often expect more pro-active hand-holding. Instead, by analyzing patterns in customer interactions, household changes, policy roll-offs, CSR calls, content consumption, etc., insurers can now leverage digital insights to identify clients at high risk of lapse and develop pro-active strategies to prevent a cancellation/lapse before it becomes an issue. Insurers can prioritize their retention initiatives based on a combination of lapse risk and the insured s lifetime value, then begin positioning messages and product recommendations accordingly well before the situation becomes critical. Data Collection & Profile Attributes There are many sources of 3rd party available data about clients (e.g. credit reports, demographics, category purchase histories, psychographics, firmographics, lifestyle segments, in-market indicators, social media commentary, etc...) that can be collected, analyzed, and used (as data points for predictive analytics and marketing, not necessarily for underwriting) to supplement the data collected directly by the insurer directly from the insured or the agent. Part of monitoring any business relationship is gauging the growth and development of your client; this includes assessing where clients are in their development as individuals or organizations. For individuals, are they at a point where they might be starting a family or buying a home? Are they planning for college or have new drivers? Are they coming to retirement age and shedding a lot of encumbrances (like a family home)? For commercial and group lines, is the client growing? Is there additional regulation or new risk that they must deal with? How has their risk portfolio changed? When monitoring and managing a client s profile, insurers need to be able to take into account any data that is available, both qualitative and quantitative, that can be leveraged to identify when to cross sell and upsell to the client. Additionally, environmental factors that are external to the client can be taken into consideration. If the client is situated in a geography that is going through a boom cycle, the strategy for managing that account might be different for clients that are in a shrinking economy or are experiencing regional disruption. This data can be leveraged by AI to 1) more accurately predict which clients are likely to lapse, and 2) to determine what actions the insurer can take to achieve positive outcomes with that insured. It is the function of AI to determine which attributes are predictive, and then how to use those attributes to identify the best possible course of action. Machines can now do this quickly, with a high-degree of accuracy where the process was resource heavy and very expensive, previously. 5

6 Contact Strategy Artificial intelligence can provide significant gains when it comes to determining a contact strategy for each client; with the capability to analyzing and model a large amount of data very quickly, it is currently used to perform tasks such as the following: Product Recommendations as described above, balancing individual product recommendations based on what is relevant for the client, and still beneficial to the insurer. These are surfaced in inbound, outbound, agent-facing, and 3rd party or paid channels. Targeting Selecting the best individual client or prospect targets for outbound campaigns, based on their propensity to purchase, or engage. Timing determining if the customer is in-market for a specific product, at the current time. To avoid spending budget on clients unlikely to take an action at the current moment. Channel Selection determining the preferred channel, and integrating that preference into insurer workflows, processes, and outreach. Treatment Selection determining the treatment (images, messages, packaging, etc.) for that specific client. Cadence Analysis determining how often to communicate with a given client. Response Analysis determining the effectiveness of a channel or campaign initiative, both as the initiative is occurring (real-time), and how it ultimately compares to other alternative. Agent Guidance providing scripting to CSR s or sales recommendations to agents, based on the context of a call or other interaction. Cross-Channel Orchestration insuring that each interaction is fully-informed with insights collected from across channels, so that the subsequent actions are relevant, timely, consistent and always based on the context of the customer situation. Marketing and Channel Management As products are released, become mature, and are ultimately adapted to evolving business environments, insurers need to be able to proactively manage the product ecosystem supporting sales. They need to actively manage product evolution, based upon insights and results from the market. Artificial intelligence should play a significant role in launching, monitoring, and adapting products throughout the product lifecycle. Pipeline For both existing and new products, insurers should be watching pipelines and closely monitoring the different stages of pipeline development (receipt to quote, quote to bind, bind to issue) along with other data including win/loss, timeliness, etc.. Insurers can determine which products are hitting their stride and which are struggling. Using artificial intelligence built on business strategy, insurers can set up scenario-based planning that will notify product owners, channel managers and underwriting when the product experience has changed, and actions may need to be taken. 6

7 For example: High appetite declination rates: do agents and other channels understand the target market for a product? Do appetite declinations point to a new market opportunity? Low quote acceptance rates: is pricing or underwriting terms non-competitive? Is the quote process taking too long and damaging win rates? Are there specific coverage characteristics that are impacting sales? Is anyone blocking the market? Regional variations and differentiation: Is there a competitive issue or an operational issue affecting underwriting cycles? Agent win rate: which agents are performing well? Which are struggling? Which are blocking the market? All of this should be leveraged to determine which agents to reward, which to invest in and which to terminate plus drive agency visit schedules. This is a great example of where many insurers are operating on a strategy of hope; they hope that their people are looking at the numbers and are able to interpret them correctly, by leveraging artificial intelligence, insurers can make sure there is analytical baseline and change process in place. New Product Launch When launching a new product or updating an existing one, how do insurers make agents aware of it? They send out material via or regular mail expecting the agent to read it. Insurers will hold training sessions, send out marketing reps, or hold events to launch products but have no sure-fire way to ensure agents get the message, understand it, and are ready to sell. While all of these functions provide value and are part of a standard rollout, insurers should be leveraging artificial intelligence in conjunction with analysis of existing books of business, to identify potential leads that can be identified, prioritized, and positioned for agents. For example, as the product is launched and introduced to clients and prospects, the system immediately begins collecting data from across touchpoints which agents are actively introducing it, what types of clients are interested, what types of content and materials are being consumed in what format, how often, etc. That data is constantly analyzed by the AI to determine which approaches are successful, and which are not. The system will adapt automatically as it gathers more information, constantly changing its perspective in the days and weeks after launch. It will begin to prioritize certain types of leads and scenarios because they are more likely to convert and will push those to agents, along with recommendations on which materials they should use throughout the sales cycle. Agents will quickly begin seeing higher quality leads, and as they are driven to engage, will become more successful in converting them because they become more comfortable and proficient, and are using materials that are more likely to result in success. But learning doesn t stop after the initial roll-out, because the system constantly adapts as the product matures, it will continue to evolve and make different recommendations, based on the results. 7

8 Line of Business Trends A quote from an insurer: we just installed a new policy administration system but all it provided was the ability to write bad business faster. Insurers need to be able to analyze all of the factors driving product sales success, but they also need to analyze whether it s selling profitably. Leveraging artificial intelligence, insurers can take sales data, claims data, and client trends and craft strategies focused on optimizing profit margins, with built-in alerts if a product is trending in the wrong direction. Underwriting The discipline of underwriting has come a long way since the times when each individual risk had to be reviewed and underwritten independently. Many lines have become commoditized, where underwriting has become more of a pricing exercise. Personal lines and many individual life products shouldn t require a human underwriter to review the individual applications using the data collected with underwriting criteria programmed into underwriting or new business systems, the risk can be underwritten automatically with pricing set at the appropriate level. Even in commercial lines and group lines, the industry is seeing some commoditization with both small business policies and small group policies. In property casualty, there is an upswing in program business, where like-risks can be grouped together to set better pricing and profitability models, and to streamline underwriting processes. Similar trends are emerging in the small group business. Not all lines of business are subject to commoditization. Wherever terms are negotiable, factors are based upon professional judgement and risk level there will need to be both a flexible underwriting process and a need for latitude. This does not mean, though, that an interactive or manual underwriting experience can t be improved and supported. There are several areas where artificial intelligence can be leveraged to support the individual underwriter and underwriting team. Pricing Whenever pricing is negotiable, underwriters are leveraging underwriting guidelines, risk levels, and their own experience around what the market will bear when quoting premium and negotiating with agents and clients. With artificial intelligence, the system can review existing books of business, identify risks similar to the one being underwritten, and make a recommendation around pricing; in effect, to win the business without leaving money on the table. It can also identify a floor beyond which the underwriter should not exceed the threshold at which the policy would likely be unprofitable. These strategies can take into account the full account of products owned or being positioned to the client with the understanding that the insurer may be taking on a less-qualified risk to win the entire account of an insured whom might be highly desirable. All of this would be driven by strategies tailored to each line of business, with the understanding that the business characteristics of each line may be different. 8

9 Underwriting Tests and Steps Keeping underwriting cycles short and reducing overall costs is a balancing act when striving for overall underwriting profitability. By monitoring which data and tests provide the most relevant results, taking into account the characteristics of each submission and overall account, which underwriting tests, data to be collected and steps that need to be followed can be recommended to the underwriter as part of a next best action strategy. Risk Management The role of risk management not only improves an insurers overall performance, it can be positioned as a value add to the insured. This includes both the underwriting process and during the lifecycle of the policy itself. Recommending risk management steps to the underwriting team during the underwriting process, underwriters can immediately lower the chance of a claim in the future, possibly enabling a lower premium, and demonstrating to the insured and their agents that the insurer wants to work in a partnering fashion. During the policy itself, monitoring conditions and offering risk management advice can help to reduce not only the clients overall losses, but create a positive experience enhancing the relationship between the client, the agent and the insurer. Risk management practices and strategies will vary dramatically by line of business and market segment. How risk management applies and the services offered will also vary based upon the state of the relationship with the client and the agent. To be the most effective and optimize the risk management spend, AI can be leveraged to embed risk management processes into the lifecycle of the client and their overall account. The goal being to reduce overall claims cost, keep risk management costs reasonable and optimize the customer/agent relationship. Book Management While automated underwriting is enabling straight through processing on many products as lines become commoditized, that does not eliminate the need for underwriting. Rather, underwriting should be managed at a book level where entire books are analyzed for profitability, for underwriting criteria, for appetite and other aspects of profitability. While human underwriters may only need to be involved with this process on a cyclical or timed basis, artificial intelligence underwriting strategies should be constantly monitoring a book s performance and sending alerts to underwriting management whenever certain criteria or thresholds are exceeded. With artificial intelligence, insurers can not only set performance boundaries but incorporate recommendations on how to proceed. This would include when results are both positive (loosen underwriting, expand into new markets, launch new marketing campaigns) or negative (tighten underwriting, issue a moratorium, increase loss reserves), all based upon the type of results that are emerging. 9

10 Claims Even a modest improvement in either claims expenses or loss experience can have a dramatic impact on an insured s bottom line. Similar to underwriting outside of standardized guidelines, most adjusters make decisions based on their training and experience, and try to leverage what is considered best practices within their group. Adjusters are not able to take advantage of all the data within their organization because they either 1) don t have access to the data or 2) don t have the time or option of examining this data to determine what the optimal course should be. With artificial intelligence however, the engine itself can pull in all associated data, and leverage best practice strategies to help drive the optimal claims process. Examples include: Settlement Options The critical balance insurers strive to achieve is settling each claim for a reasonable amount. On a single claim where the claimant believes the settlement isn t adequate, the insurer risks poor customer satisfaction, a lawsuit or a regulatory complaint. When it applies to a book or a large scale event like a catastrophe, inadequate settlements can seriously impact an insurer s reputation and erode public confidence or maybe worse, result in a class-action lawsuit or significant regulatory fines. If an insurer goes in the other direction and overpays claims, this can severely damage the loss ratio, and compromise the overall ability to achieve a profit. How can insurers address this issue and optimize the settlement process? Today, it s generally through guidelines, training and experience of the staff. This experience however, is again relative to the claims adjuster s personal experience, their training and the support of supervisors and possibly co-workers support. Leveraging artificial intelligence, insurers can compare individual claims to entire books of claims to identify what the optimal claims settlement value should be based on a variety of factors. This could include the amount of settlement, type of settlement (cash vs replacement vs services), likelihood of litigation, estimated overall expense and other variables that can affect the length, outcome and cost of a claim. Leveraging artificial intelligence in this way would be a guide for adjusters to make an informed decision. When leveraging self-learning models, all new outcomes are automatically included in the analysis, to continually improve both predicted outcomes and recommendations to staff. Adjusting Steps In addition to the settlement itself, artificial intelligence can be leveraged to recommend next best actions on how to proceed with a settlement. Can it be a straight-through process, or does it need to be manually adjusted with a full investigation? When should the insurer settle vs proceed to litigation? What are the optimal third party service parties to support a claim? All of these issues can be addressed with claims strategies and decisioning that s institutionalized within an artificial intelligence framework, and deployed within claims processing. Financial Planning In particular for life claims, disbursement of a benefit check can be a stressful process for beneficiaries. As part of the settlement process, insurers can offer financial planning options to beneficiaries to help them manage the benefit in accordance with their long-term financial needs. Similar to the new business process, based upon known demographics and suitability strategies embedded in artificial intelligence, life insurers can make a difficult time a little less so. 10

11 Fraud Finding and dealing with fraud (preferably before a claims check has been cut) is of paramount importance for insurers today. Leveraging artificial intelligence, insurers can take known data about the claim and individual, incorporate third party data (such as ISO claims search), and past data around known-fraud to asses current fraud levels throughout the claims process. Once detected, systems need to be able to launch notifications to the Special Investigations Unit and other concerned parties with suggested steps to investigate and manage the fraud going forward. Any improvement to claims immediately impacts an insurer s bottom line. Artificial intelligence with real-time analytics and next best actions can have an immediate impact to insurers overall performance. Big Data Insurance has always been a data-intensive environment but, as an industry, insurance has been slow to pick up on the use of big data, and leverage it for business benefit. Two streams of large data that are still relatively new to the industry are social media interactions, and data from connected devices. This data can be leveraged in conjunction with artificial intelligence to trigger multiple processes, and drive more productive and efficient interactions with the client. Some of the most common use-cases are: Usage Based Pricing - The most public example is the use of telematics and auto insurance, but versions of this are being looked at across multiple products. A use-case is also emerging in life, where life insurance pricing can be discounted by activities such as achieving quotas on fit-bits and other devices. This data can not only be used for pricing but to introduce risk management influenced behaviors and activities into the client s user experience - what s known as gamification. Risk Management - Based upon data that is received from connected devices, insurers can launch a series of risk management protocols to decrease the chance of loss. This includes machine-monitoring for service (preventing breakdown and other loss), medical monitoring with associated recommendations to improve health and longevity, characteristic behavior (such as aggressive driving through a telematics device) resulting in driving recommendations or remedial training. Artificial intelligence is used to determine what the risk-predictors are, and is also used to analyze and deploy the findings in real-time, deploying them wherever they will have the most impact. Day to Day Experience - Insurers, at all levels and across all lines of business, are looking to incorporate themselves into the day-to-day experience of their insureds. By leveraging data from social networks, web analytics, newsfeeds (in the case of corporate clients), financial and credit data, and connected devices, insurers can create an ongoing dialogue with insureds that ensures that the insurer remains relevant and considered a valued contributor. Pulling all of this data together is not a small task and is usually done on a channel-by-channel basis. Artificial intelligence solutions that support predictive analytics and machine-learning models (coupled with contact strategies) can consume that data with the speed and efficiency required to keep pace with the customer. 11

12 Pega Customer Decision Hub (CDH) Pega Customer Decision Hub is the only product in the market today that consolidates data from across the enterprise, enriching customer interactions with business strategies designed to be leveraged across sales, marketing, service, risk, and operational functions. The Customer Decision Hub provides predictive and adaptive analytics that recommend actionable activities and can set in motion workflow, tasks alerts and reports to help insurers drive the business. CDH is an independent product that can function not only in a Pega environment but as an independent engine integrated with third party software, custom-built solutions, and legacy systems. It s designed to utilize information from multiple sources (both internal and externa), and leverage all the data available to make intelligent decisions, and provide actionable recommendations to relevant parties. CDH can also consume analytical models developed in other systems (such as SAS) and incorporate those models into CDH processing. As a core part of the CDH solution, Next Best Actions provide recommendations to systems, channels, or users; it also launches specific processes that support business strategies. As a learning engine, CDH analyzes an organization s data collected over time, and works to optimize the potential outcomes. The self-learning engine can be leveraged with little to no data (if data is unavailable) and enables insurers to start from scratch when entering into either a new market or experimenting with new strategies. As the system makes recommendations and records the responses, it learns from the results and the results improve rapidly, in a small amount of time. CDH enables insurers to provide the insight driven support needed today to underwriters, front office staff, and product managers but most importantly, to the agent and the insured. As a resource, CDH will make the insurer a value-added partner that both agents and insureds will want to continue doing business with over the long term, optimizing the relationship over time for everyone. To learn more about AI and the Customer Decision Hub, you can download the following ebook, The Customer Decision Hub: What s Inside the Always-On Brain You can also reach out to your Account Executive, or use the following link to contact us: 12

13 ABOUT PEGASYSTEMS Pegasystems develops strategic applications for sales, marketing, service and operations. Pega s applications streamline critical business operations, connect enterprises to their customers seamlessly in real-time across channels, and adapt to meet rapidly changing requirements. Pega s Global 500 customers include the world s largest and most sophisticated enterprises. Pega s applications, available in the cloud or on-premises, are built on its unified Pega 7 platform, which uses visual tools to easily extend and change applications to meet clients strategic business needs. Pega s clients report that Pega gives them the fastest time to value, extremely rapid deployment, efficient re-use and global scale. For more information, please visit us at 2017 Pegasystems, Inc. All rights reserved. All trademarks are the property of their respective owners.