Artificial Intelligence and Machine Learning: These digitalization buzzwords are hot topics in almost every industry. Insurance is no exception: But instead of using it only to reduce costs and make processes more efficient, insurance companies should use it to improve their customer centricity.
If you google the keywords Artificial Intelligence (AI) and insurance, search results amount to more than 60 million hits: The industry is in turmoil; and digitalization plays a huge part in that. Mostly lagging behind in this regard, insurance companies know they have to do something. But do they know what exactly? And are they doing it successfully?
Machine Learning enables efficient processes
Some pioneers already use AI and Machine Learning (ML) for insurance purposes. One company, for example, offers its customers an app with which they can document damages on their cars. Automatically, these pictures are compared with similar cases within a huge databank. Consequently, more than 80 percent of damage claims can be handled without human interaction in a very short timeframe. Another insurance company replaced their customer service hotline with an AI chat bot. A third example is the Bavarian insurance chamber that developed a solution to manage their reams and reams of correspondence. An algorithm detects keywords within, sorts the correspondence according to topics, determines urgencies, distributes it into different departments, and even suggests next best actions. One thing have all of these examples in common: AI and ML are used to optimize processes, reduce costs, and increase efficiency.
Digitalization doesn’t pay off
These three examples use Machine Learning in a very profitable way. But what about the rest of the industry? In this year’s Global Pricing Study, we found out that more than 70 percent of insurance companies have invested in digitalization initiatives in the last three years. However, less than a third of all insurers surveyed were able to see a clear impact of their digitalization projects on their top line. That’s not very effective! So, what’s going wrong? From our project experience in the industry, we were able to detect a couple of widespread problems with Artificial Intelligence and Machine Learning:
- Low data quality: In accordance with the guideline “Garbage in, garbage out”, insurance companies struggle with the quality of input data. As a result, models can’t be built to give targeted recommendations. For example, an algorithm that is supposed to combine contract data with information from CRM systems in real-time is really unreliable if both data banks aren’t maintained meticulously. And what is the use of developing an algorithm if you can’t let it run without constant checks?
- Bad project design: It really affects insurers’ yield potential that many get pro-active prematurely. The order of developing a Machine Learning solution first and then start thinking about how to monetize it quite often doesn’t work. Without focusing on customers’ needs and finding out what they are willing to pay for, AI projects are doomed to failure.
- Security concerns: We’ve seen insurance companies that had to withdraw their ML-based apps from the market because missing security protocols mad it a suitable target for hackers. Risk mitigation in this regards concerns many insurers when it comes to AI and ML projects.
3 levers to make ML/AI projects successful
How can insurance companies tackle these challenges and concerns? We identified three levers:
- Insurers aren’t short on ideas on ideas for Machine Learning projects. The problem is to combine countless initiatives to a holistic solution that’s useful to customers. Usually, apps are developed hastily and adoption rates are incredibly low. Efforts have to be channeled to aim for an overall goal that is aligned with customer needs and requirements.
- Insurers tend to think large – but small measures are usually much more effective. For example: If a customer’s life situation changes in the near future, many insurance companies do not follow this lead consistently. An automated process (based on ML) could send out a slightly personalized advertising brochure as well as measure the success rate of such measurements.
- Behavioral economics are worth the while: Insurance companies can use a ML-based gamification approach. We have conducted banking projects operating on the sticker album principle, where customers are led to collect as many offering elements as possible. The more products they use, the more stickers they get, which provides a nice, motivating visual. This can be additionally supported by incentives like a special status or given discounts.
Learning from advanced players
Where Artificial Intelligence and Machine Learning are used in a very beneficial way is the entertainment industry. For instance, Netflix, provider of TV streaming services, uses an ML algorithm for its personalization purposes. Netflix’ business model depends on giving its customers suitable recommendations from the start to keep their loyalty. Basis is user information on what shows are watched, how often, which genre, etc. By means of this data, Netflix identifies its users’ value drivers and decision-making criteria. The result are targeted recommendations as well as personalized preview pictures in accordance with viewers’ tastes. What can the insurance industry learn from Netflix? Most importantly, to focus on customers’ needs and wishes in product and service development. Apart from that, Netflix shows the value of properly executed personalization and recommendations. In addition, the Netflix algorithm is very simple and based on only a few criteria as well as simple market research.
Insurance: becoming more personalized and customer-centric
How might this learning look like put into practice in the insurance industry? Right now, insurers often use the scattergun approach instead of personalizing their offering. But hoping to meet your target group’s requirement is no longer enough – they need a more dynamic approach with customer differentiation and dynamic pricing along every step of their sales process. Artificial Intelligence and Machine Learning are the tools to use to enable this. Moving away from using digitalization only to reduce costs and optimize process to generating revenue is the name of the game. If insurers start small, generate high quality data, and think of obvious solutions to increase customer centricity, this will be beneficial to them.