Insurance chatbots need to walk before they run

Chatbots enhanced with computational linguistics, machine learning and other forms of artificial intelligence (AI) have the potential to transform human interaction with insurance. These technologies can be overwhelming to insurers, begging the question: Where to begin?

By starting with middle and back office operations, companies can reduce risk, enable a better chatbot strategy, and provide significant value by enhancing the human staff’s abilities.

Related: The rise of the insurance chatbot

Outside of simple requests, insurers trying to connect chatbots with consumers have run into issues including:

• Customers don’t understand the kinds of requests a chatbot can handle (nor can the insurer train the bots to understand); we seem to repeatedly forget that people don’t see insurance transactions the way they see Siri or Alexa.
• The variations in intents (the things customers want to know) are extraordinary in insurance and even the best chatbots struggle to keep pace.
• The ultimate objectives of a chatbot program are not always thought out ahead of time meaning that some “strategies” are just to get the bot out there.

Difficulties aside, all is not lost with chatbots. Why? Because an “insurance assistant” can significantly improve the customer experience indirectly with chatbots by enhancing the employee experience by:

• Facilitating learning and development — For question/response scenarios and standard operating procedure (SOP)-related intents, a chatbot can be invaluable for learning and development. Experienced staff teach the chatbot appropriate questions and responses (and maintain this through machine learning), and inexperienced staff get to benefit by having a resource for answers.
• Providing the right system information at the right time — Chatbots can learn across and outside of the carrier. By providing information to the staff at just the right moment, the chatbot helps the customer service representative offer real-time information. For example, in a first notice of loss (FNOL) call, the chatbot can send representatives repair service information as they gather incident information, not as a separate interaction. Similarly, the chatbot can surface sales and marketing information during quotes, endorsements, and renewals – especially for more complex lines of business.

To deliver this information, the chatbot must integrate with systems of record — including policy, claims, billing — and external services such as towing and repair. For a chatbot, this means not only data integration but also converting data to conversation.

• Creating value in the middle and back office — One current truth about AI is that the combination of computer plus human is still more powerful than either alone. For this reason, the chatbot creates value by maximizing the value of the humans. Chatbots do this by “remembering” things humans forget; bots can prompt staff to ask questions, provide information that the human doesn’t know exists or cannot readily access, and facilitate a consistent process.

Chatbots also add value by:

• Always being present — Immediacy is a critical component of client satisfaction and when the human can have the right information at exactly the right time, insureds and claimants receive better service.
• Acting as a single source of knowledge for SOPs and FAQs — Chatbots add value by consolidating institutional knowledge.
• Allowing the insurers’ chatbot strategy to evolve with less reputational risk — Better risk management is indirect value of starting with the middle and back offices.

Perhaps the most important use cases involve claims. During the entire claims lifecycle, the chatbot can provide value to claimants, staff, and insurers. The chatbot can surface critical policy and repair service information during FNOL. At the same time, the chatbot can use analytics to spot fraud, triggering special investigative unit activities. At every point in a claim, the chatbot can help the human provide more relevant and informed information on a timely basis.

Helping sales and service

In underwriting, chatbots can help cross-sell and up-sell by prompting call center staff. The chatbot can consider a wide range of data, including information from marketing, policy administration, and potentially public sources. This scenario is equally applicable regardless of the underwriting transaction — including new business, endorsements, and renewals.

Agents, who need simplicity, access, and speed to best do their jobs, also can benefit from chatbots. Easy access (e.g., through text messaging) can facilitate work with straightforward transactions such as certificates of insurance, claim status, submission status, and risk appetite.

Renewals deserve a closer look since a call center conversation may not happen. Chatbots don’t need to live chat with a human; the bot can generate renewal letters that humans can alter and send. In this way, the chatbot/human interaction is asynchronous but equally valuable.

Billing is another area where the middle and back office chatbot can be valuable. As the call center staff is speaking with the insured, employees also can receive information from the chatbot back to the billing system. For simple transactions, the chatbot can save significant time without costing a lot of money.

Driving internal efficiencies

Internal services also are a key area for chatbots, including IT-help, desk inquiries, and human resources. Some good examples for IT include how to reset passwords, information on major system outages, and initiating user provisioning/de-provisioning. For HR, the scenarios include any form-related inquiry (e.g., where to get W2s), and basic queries such as holiday lists, education resources, and common HR policies.

Chatbots also can be very valuable to finance and accounting (F&A) because of the regulatory environment. If chatbots can enforce F&A processes, the bots also can help insurers comply with current regulations.

A brief word on the future

While the middle and back office chatbot are the here-and-now, the future of ambient chatbots is very real and exciting. The future chatbot will be listening to voice (and even text) and will react based on conversation’s context – surfacing information and enforcing processes. Ultimately, chatbots will allow human staff to focus entirely on enhancing customer experience versus some of today’s administrative minutiae.

Frank Neugebauer has dual roles at Genpact as the Digital CTO for Insurance and the Digital Consulting lead for the Americas. He has over 19 years of property and casualty insurance experience in standard lines, excess and surplus, and reinsurance. He can be contacted via LinkedIn.