Insurance Chatbots: Current and Future Opportunities
By Mike Fitzgerald, Senior Analyst, Celent
Chatbots change transactions into conversations. These programs interpret spoken language, or read text messages, in order to interact with customers in real time. Depending on the complexity of a solution, a bot may use multiple techniques— natural language processing, statistical prediction, workflow control, big data management—to interpret conversation, provide an appropriate response, and move data accurately from one source to another. Benefits realized by adopters include lower service support expense, higher consistency of response, and increased availability.
In insurance, chatbots are an emerging technology which Celent expects will be table stakes in three to five years. Leading insurers will transform customer experience and personalize consumer and agent service using these technologies.
Agents view chatbots as an opportunity to decrease training costs and increase consistency of technical knowledge
A recent Celent research effort surveyed 15 chatbot vendors to understand their offerings and their depth of penetration into the insurance sector. The results clarify the current state of the technology:
• Use in insurance is relatively low. Most solutions have yet to be tested against insurance processes. This creates both opportunities and risks for first-mover insurers.
• The insurance implementations that have launched are typically targeted at specific areas such as IT service desk support, first notice of loss reporting, or frequently asked question responses. Such a narrow approach can lead to higher total cost of ownership because insurers have to implement multiple solutions. Enterprise wide platforms which coordinate across processes are not yet in evidence.
• Agents view chatbots as an opportunity to decrease training costs and increase consistency of technical knowledge across their customer service representatives.
• Most solutions deliver functionality regarding data input from text and voice and workflow control. Reasoning functions (such as predictive ability and machine learning algorithms) are more prevalent, while modeling, business process analysis, and image input or analysis is not widely available.
The risk of not adopting chatbot technology is higher than the risk of adopting it too early. Leading Insurers are developing use cases to test the technology now, in preparation for the inevitable time when these tools are a new customer communication channel which both improves customer experience and reduces costs.