The Human Touch: The Untapped Opportunity for Making AI Work for Insurance

The opportunity for AI is in targeting the end-to-end processing of documents for human-machine collaboration as an essential part of the customer journey.

(Image credit: Tumisu.)

Customer acquisition and retention remain the essential drivers of insurance revenue and profit. However, knowing how to create experiences to acquire and keep customers happy remains challenging. A new breed of competitor—InsureTechs—target smart technologies at the customer experience and are disrupting the status quo for traditional insurers. What Fintechs did in consumer banking, InsureTechs are doing to transform insurance customer experiences through smart mobile apps, integrating the phone’s camera as an input channel, and collecting data to target personalized offerings.

Artificial intelligence (AI) technologies show a lot of promise in transforming customer acquisition and retention, yet it is all too easy to assume that better technologies can, in themselves, transform customer experience. What’s missing in these discussions is the human factor in the use of AI technologies: how AI tools can effect change in customer experience and operational excellence by focusing on the interactions of people, process and content. Competing digitally means targeting AI to remove friction both from the customer experience and processes that directly support it.

In insurance, AI is being used to remove friction from the processing of documents and unstructured content critical in customer journeys from recruitment to onboarding, through underwriting, claims and settlement. When approached from a human experience, rather than strictly an automation perspective, AI in the forms of OCR, natural language processing, robotic process automation (RPA), and machine learning can deliver digital intelligence—the understanding of content, processes and the people who rely on them to make better business decisions.

AI for People, Process and Content

Content: Insurers have looked at AI primarily to automate the processing of documents in targeted customer interactions, calling this automation Optical Character Recognition (OCR), a widely-adopted simple AI technology. OCR tools, when used with good processes and image quality, are directed data extraction tools for incoming content. However, extracting data from documents using AI-infused OCR tools is just a small fraction of the value and opportunity of AI’s potential. Infused with machine learning, insurers can have content automatically extracted, classified and taught to go into upstream processes.

Process: Adoption of RPA (robotic process automation) automated insurers’ repetitive, back-office tasks supporting processes. RPA drew new attention to processes as a locus for AI, but, like OCR, its value comes from its being a tool in directed automation. Now that RPA has reached wider adoption, insurers are realizing the bigger opportunity is to use these digital workers at the front line of customer engagement, where intelligent understanding and responding to customer needs pays off exponentially in revenue and profit. Targeting digital workers and AI directly at removing friction from engagement may soon eclipse the back office as the focus of AI investment.

People: Processes and content are essential parts of insurance customer experience and productivity. Insurance being a document and process-centric industry, offerings such as policies, claims, adjudication and compliance, rely on process efficiency and content. Even with infusing AI technologies, they are implemented today with the goal of automation, and the human experience is an afterthought.

The emerging opportunity is to think of AI-based automation from the perspective of customer journeys.

AI in Document-Centric Customer Journeys

No matter what technologies are used, insurers interact with their customers through the exchange of documents, messages and unstructured content. As customer journeys and engagements get more attention for AI and automation, the handling of documents within those processes remains an afterthought, pushed to the end of the process where users open documents from email and key data into their systems of engagement. Transforming documents in real-time into process-ready data is critical to ensuring the right level of automation and decisions are available to the customer-facing process.

Document handling needs to be at the front of automation, but is overlooked because this process is not fully discovered, understood and mapped as a function of customer engagement. While both OCR and RPA provide automation within these processes, they are often deployed incrementally in projects to optimize existing processes.

Furthermore, insurance documents pose problems for automation. These documents are complex, highly variable and often come in high volumes. Notification of loss, reports, estimates, invoices and supporting documents are full of free-form text, nested tables, variable information in unpredictable forms from variable sources; they present a formidable processing challenge when handled individually, but become exponentially worse in any kind of volume. Here, understanding the context, customer need and complexities of the documents, and addressing these issues in real-time represent a significant opportunity for AI.

Now that scanning and service bureaus have been replaced by mobile cameras, email and other digital channels, it carries the impression that these functions have been digitized. However, problems with inputting content from these documents into customer processes remains largely unaddressed, even if the input channels are digitized.

Document Processes versus Processing

Two fundamental problems with document-handling still need to be understood to automate customer experiences around them. These are process and processing—two distinct functions that are assumed to be the same.  Process, the flow of events and actions, is where customer experience and engagement live, and automating these processes of engagement is the key to speed, quality and satisfaction. Customer engagement is a process during which information from documents must be inputted directly at certain points in it.

Processing is how documents are located, prepared, identified, classified, read and data extracted, validated and released into the processes, systems and stakeholders for decision-making. While insurers have largely digitized document input via smartphones and PDF attached to email, the actual processing of their content occurs at the end of the process, instead of the point of engagement. As a result, insurers are losing the “now moment” with customers by deferring the processing of the most valuable content their customers provide—claims, supporting documents, statements, reports, invoices—to the end of the process. This branch process, rather than fully integrated document processing, causes additional friction, delay and frustration for customers, with less good data available in the now for next steps in decision-making and follow-up.

Having digital intelligence deployed throughout document processing can address these challenges with content and process, but all too often they are over-simplified with plug-and-play automation and OCR projects. Or worse yet, focused only on one piece of the process, rather than its whole. Understanding the context, relationships, entities and intent of the extracted data is what drives the actual benefits of automation in customer journeys. AI has made major advancements in all of these areas. The opportunity for AI is in targeting the end-to-end processing of documents for human-machine collaboration as an essential part of the customer journey.

Using AI to Do Insurance Customer Care Right

Reginald Twigg, Ph.D. // Reginald J. Twigg, Ph.D. is Director of Product Marketing, Data Capture at ABBYY. He is an expert at helping global organizations leverage AI technologies to gain better insight into their content and improve customer needs.

Leave a Comment