As with most industries within the financial services sector, insurance is facing the dual pressures of needing to keep costs down in a competitive environment while maintaining satisfaction levels—and this can be seen very vividly in the area of customer service.

 

The call center is typically the front line of customer support, and is often characterized by long waits and cumbersome menus. This leads to unsatisfactory experiences, and ultimately lost business. In addition, interactions with customer service staff can sometimes prove to be frustrating—and it only takes one bad call for a customer to leave, or even worse, voice their grievances on online forums or social media.

 

One of the most common complaints is the length of time taken to speak to a customer service representative. However, there are often additional problems once the customer has got through.

 

According to the Consumer Reports National Research Center, the top customer service irritants among consumers include:

• Seventy-five percent of consumers surveyed found customer service rude or condescending.
• When making a telephone call, 74 percent of consumers surveyed were disconnected and unable to reach a representative again.
• When speaking on the telephone to a customer service representative, 64 percent of consumers surveyed felt they were being ignored.

 

On top of this, customers are getting more resourceful when it comes to proactively seeking information about their insurance policies. Their experience when doing so largely hinges on the interactions they have with customer service representatives. As a result, it is clear providers must innovate in order to make the customer experience faster, simpler, and more enjoyable.

 

The Cost vs. Satisfaction Challenge
The core question for insurance providers is how can they simultaneously raise customer satisfaction levels, lower costs and enhance productivity?

 

This is a significant dilemma as improving satisfaction and productivity can be achieved by hiring more and better trained customer support staff, but at a considerable cost. Historically many organizations in a variety of industries have offshored their call centers, but this has caused issues with comprehension, and especially among higher age groups. In fact, many companies have started to ‘reshore’ their customer support centers.

 

British telecommunications services company, BT, for example, pledged this year that it would raise the proportion of calls answered by UK contact centers from 50 percent to 80 percent in response to customer complaints. In order to do this, they would have to fund hundreds of new positions in the UK. (Source: “BT promises to bring call centres back to the UK,” The Guardian, Sept. 18, 2015.)

 

One regularly deployed solution is “live chat” or “click to chat,” whereby customers can message customer service representatives online. However, while this works for some queries and tasks, it is unsuitable for many others—especially any requests that involve complexity. What is required is increased conversational ability.

 

Another option has been Intelligent (or Interactive) Voice Response Systems (IVRs). These are pre-recorded menus, equipped with speech recognition software that takes customers through a range of options until their query is resolved. However, the menus tend to be limited and only able to respond to yes/no or a very small range of answers.

 

IVRs can sometimes help solve very simple queries without the need for a conversation with a customer service representative, but they often become a source of frustration for customers who frantically mash buttons to get through to a person as soon as possible.

 

A New Solution: Artificial Intelligence
Artificial Intelligence provides an opportunity to solve these problems at a reduced cost, especially in the long term. Virtual agents mean that interactions with customers can be reciprocal, conversational and pleasant. (Editor’s Note: The author’s company, IPsoft, has developed avirtual agent known as Amelia—an artificial intelligence platform that can understand, learn and interact as a human would to solve problems.)

 

Virtual agents work by intelligently combining a “process ontology” with a “neural ontology.”

 

A “process ontology” is essentially knowledge of what a customer is likely to ask during the course of an interaction. This can be preprogramed into a virtual agent, but taken in isolation it is not necessarily that different to the information programmed into IVRs.

 

Where AI differs, is the incorporation of a “neural ontology.” This is effectively a “brain” that allows a virtual agent to leave the limits of the process ontology and bring in a human agent. Straight away this increases satisfaction—the customer gets to speak to a representative before they get a chance to get frustrated rather than having to work through a menu.

 

However, it is not just this minor improvement to a conversational dynamic that makes virtual agents so suitable. The neural ontology also gives them the ability to learn. Before and after transferring the customers call, the virtual agent can observe the question and answer—adding the human representative’s response to the process ontology. The response is learned and so human intervention will no longer be required the next time a question is asked.

 

Incrementally the virtual agent is improved, effectively becoming more human, so that eventually it doesn’t need any help for even the most complex questions.

 

The AI agent only needs to be trained once rather than every time a new hire is made.

 

The additions to the process ontology can be checked by a manager to ensure that it’s appropriately refined and works properly. Once this is finalized it is never forgotten by the virtual agent; unlike human support staff who may forget or sidestep important processes. Additionally the AI agent only needs to be “trained” once, rather than every time a new hire is made.

 

Assisting Front Line Humans
AI agents have a huge amount of flexibility. For example, human agents can also collaborate with their virtual counterparts in a different way. An insurance provider may want to keep people as the front line of their customer support or sales staff. If this is the case, a virtual agent can be deployed to act as an interactive assistant for the employees. The virtual agent will be able to quickly look up information while the employee is on the phone interacting with customers. This can be done automatically through the process ontology, with the virtual agent looking up the relevant information without having to be asked. This speeds up call resolution, avoids confusion and means that employees can more easily follow guidelines.

 

Virtual agents can also take the place of lengthy online forms for front-line sales, helping select exactly the right insurance package for a new customer or prospect. For example, a virtual agent can:

 

• Proactively ask what type of insurance a customer is looking for.
• Look up a product’s book value.
• Inquire and incorporate extras, such as ski or diving cover for travel insurance.
• Look up and help select variables and appropriate excess.
• Finally make recommendations, select the package and take the payment.

 

This is all done in the context of a personalized interaction rather than through a faceless form. The agent can respond to queries and even provide financial advice; all at a cost far less than employing teams of sales support staff.

 

Enabling the Transition
Put simply, the biggest obstacle to AI implementation in insurance is fear. This is nothing new. Throughout history, the greatest inventions have often been the most intimidating and most challenged. No one thought that the personal computer would catch on, and now most of us keep one within arm’s reach at all times.

 

New technologies, while sometimes scary, have proven to create more jobs, push people to work harder and smarter, and improve quality of life. Artificial Intelligence is this decade’s transformative technology.

 

The practical obstacles to implementation are far smaller. The flexibility and continuous improvements inherent in AI systems mean that they can be deployed very quickly. The first step is identifying which parts of the business that would benefit from AI support. Some insurance companies have over 300 individual processes. While not all of them are suitable, many of the clunky operational procedures that take up a great deal of employee time are perfect candidates.

 

To ease the transition, virtual agents can be deployed incrementally—starting with one or two processes within an operational part of the business, such as customer support, and then expanded to any other area where there are benefits.

 

Fears of mass redundancies are also misplaced. AI can mean workers within the business are freed from the most mundane tasks. With AI systems taking over this kind of repetitive work, employees are challenged and encouraged to take on more creative and revenue-generating roles. This should improve quality of life and enjoyment of work, therefore turn decreasing staff turnover.

 

With this in mind, early adopters and fast followers will see the greatest benefits from AI technologies. By the time their competitors catch up, they will have been long enjoying cost savings, greater efficiencies and customer satisfaction.