The US insurance market today is extremely competitive. It is fairly easy for consumers to switch their existing carrier and move to the one offering them a better deal. As the customer base gets saturated, insurance carriers run many programs for customers to switch from their existing carriers. With low barriers to switching, customers become indifferent and shop around for a better deal each time they need to renew their insurance policies.
I recently realized how deeply entrenched this consumer behavior is when a close friend casually mentioned to me a supposedly fantastic deal he was getting from an insurance carrier. Intrigued, I asked him if it wasn’t cumbersome for him to submit personal details and set up a new policy again with a new company who had no clue about his needs. He shot back saying that neither did his current one! He added that there’s nothing unique about the level of service he’s receiving currently and the cost is the primary driving factor for him. His exact words were, “I don’t see how my service level will be any different, so why not save some money?”
My friend’s story is not unique. The industry today more than ever is facing this tug-of-war between retaining customers and acquiring new ones. However, it is not practical for companies to deploy an extensive customer service program that monitors every customer’s communication and measures their satisfaction level in real-time using traditional methods of analysis.
The traditional approach has been to rely on mass surveys, electronic, telephonic communications, and agents to identify points of dissatisfaction. Mostly, if the customer does not initiate communication, the carrier will have a limited view of the customer’s viewpoint. Couple this with the way data volume is increasing (especially unstructured data), and one realizes that it is nearly impossible to understand customer behavior in any meaningful way using traditional approaches.
Deep learning and AI methods in the recent years have gained significant prominence because of the outcomes generated by companies like DeepMind. The question that naturally comes to mind is, “Can Deep Learning methods help decipher and predict customer behavior and thereby help retain customer better?” I genuinely believe that deep learning can help insurance carriers bring differentiating value in the way they serve their customers; here are some of the ways:
- Predict Customer Behavior: With deep learning capability, it is possible to create customer profiles and continuously analyze the interactions. A simple neural-network based model can be created based on historical data and patterns to predict churn. Then the model can be used for the current set of customers by analyzing their emails, calls to customer care, and online activity to determine their propensity to stay or leave.
- Determining Causality: The machine learning capability of a solution can learn from historical data and identify typical patterns that lead to a customer getting dissatisfied and not renewing their policy. The model can flag top contributing factors and causes, which can then be analyzed by retention department for long-term mitigating strategies such as customized service, custom offers, and rewards.
- Get actionable insights of the competitive landscape: By using a solution rooted in deep learning, carriers can analyze the customer base on the basis of geography, demography, and customer demand, and factors in local markets to develop and launch customized products and programs quickly to stay ahead of the competition. The models can predict how likely is the competition to have an edge for each existing individual policyholder, this information can then be used to put a customized customer retention program focused on proactive competitive offerings.
- Offer tailored experience to customers: There are two ways in which carriers can tailor the customer experience:
- First, by building enriched customer profiles, predicting behavior and predicting lifetime events, targeted communications can then be designed which are specific to the customer situation resulting in enhanced customer engagement.
- Second, to identify the preferred channel of communication-based on customer behavior (paper, telephone, tablet, smartphone, email, instant messaging, or any other channel) and use the most effective channel to deliver communications and offers.
We all realize the importance of retaining customers and lowering the churn rate. After all, retaining customers is far more cost-effective than acquiring new ones. There are numerous benefits to the approach of adding deep learning based insights into any customer retention program:
- Speed and accuracy: The insights are in near real-time and latency is minimal. Quick, proactive actions can help with policy renewals.
- Automated responses: Models can generate red flags, and each red flag can be designed to trigger an action to automate the customer retention process further.
- Context awareness: Any customer service that is context-aware, can save a lot of time and resources. With deep learning, it is possible to not only identify issues proactively, but also get recommendations on possible solutions.
- Better and tailored customer experience: With consumers becoming more self-service oriented and favoring mobility-based channels, an AI-based solution with deep learning capabilities is the way to manage and deliver desired customer experience. It can keep track of customer preferences and use it to identify patterns and predict customer behavior.
As we know, deep learning based systems are self-learning in nature, and with time the accuracy of prediction will only get better, delivering more value and improved benefits as a result. I do believe that the insurance industry is primed for a technology revolution, which is centered around knowing its customer better and offering them what they need even before they realize they need it.