How does AI affect insurance policies

KI: "Insurers have recognized potential in their extensive databases"

A dream for every company: on the basis of data analysis, it knows exactly what customers want, price expectations and the date of purchase are known. This vision should also find its way into the insurance industry through the use of modern technologies and artificial intelligence. This requires the right software and analysis experts: neither can be found at the push of a button, but requires an intelligent selection process.

After all, it is a small consolation for the insurance industry that could promise the prospect of better times. According to GDV, the year before the corona crisis was a very good one for insurers. In 2019, after years of extensive stagnation, Germans again spent more on insurance policies - an average of over 2,600 euros per year and insured person. In some sectors even record values ​​were achieved. This is the conclusion of the “Statistical Pocket Book of the Insurance Industry 2020” published by GdV at the beginning of September 2020. Even if Corona will probably miss a dent in this upswing, the figures suggest that sales will soon point upwards again.

Investments in data analysis are increasing

And for this, the financial service providers want to invest more in digitization, among other things. Data analysis is very popular with them, as the results of the “Trend Study Digitization 2019” by Bitkom Research show, but published before the Corona crisis. After that, six out of ten institutes want to invest more in software for data analysis. A trend that runs through all industries. A good indicator of this is, among other things, job postings. Data analysts have long been coveted and not easy to find experts. In 2019, every fifth insurance company or bank was looking for data scientists. A year earlier, such job advertisements were rather rare. For example, anyone looking for job offers for “data analysts” on the Stepstone online job platform received more than 4,500 hits across all industries at the beginning of September 2020, including insurance companies such as Zurich, R + V and Axa.

The Bitkom Research study also showed that companies now value knowledge of all aspects of data analysis for all employees. At 89 percent, data analysis is right at the top of the scale as the desired competence. So also for those who are not specifically responsible for data analysis. The evaluation and use of data collections has always been one of the core competencies of successful insurance companies. So far, however, they have mainly used data analysis for tariff calculations. That should change now.

There is potential in data for all departments

The insurers have recognized that there is a great deal of potential in their extensive databases for advising existing customers but also for acquiring new customers - and not after weeks of analysis, but in real time. From complaint management to future-oriented sales control, the decisions with big data analyzes and artificial intelligence, often guided by the gut feeling of brokers and insurance agents, could in future be based more on hard facts. Specific areas of application for AI should primarily be the analysis of customer behavior and customer loyalty, lead generation and lead management as well as the automation of manual activities (RPA). For this, however, the insurance companies not only have to find data analysts, they also have to select the right software, on the basis of which the large amounts of data automatically generate the answers to key questions about customers and general market developments.

The insurance sector is often seen as a pioneer in using AI and techniques to analyze large amounts of data. However, only eight percent of insurance companies and banks use artificial intelligence to automate such tasks, reveals the Bitkom study. Although data analytics solutions are also available to insurers for almost all areas of the company - even beyond customer service and sales. That's why 86 percent of companies plan to increase their investments in Artificial Intelligence (AI) by 2025. Around a quarter each expect a relevant influence in the development of new products and services, in the development of new markets or industries and as a trailblazer for innovations.

Software selection based on needs, not just based on technology

But the willingness to invest in data analytics and AI is only one side of the coin. A study by the Economist Group shows that companies see risks with the introduction: in terms of security (40 percent), costs (39 percent), inadequate infrastructure (29 percent) and poor data quality (28 percent). What data analytics and AI consulting projects also show is that insurance companies often only approach issues such as data analysis from the technology side. Often, however, it is crucial to collect the interests of all those involved and to test their feasibility, before long-term commitment to software. Here is an example of a use case for sales impulses: From a marketing perspective, these must be made available to every customer advisor quickly and up-to-date. From a sales perspective, the impulses must be clearly presented and sortable according to criteria. From the management's point of view, the processing status should also be recognizable and measurable in order to be able to assess successes. In general, software should also be contemporary (e.g. data protection, architecture, scalability and license model), intuitive to use and economically sensible.

But where is artificial intelligence used? The decisive added value results from the dynamic prioritization of sales impulses. This includes on the one hand the determination of purchase probabilities and the consideration of channel affinities (end customer focus), but on the other hand also the interests of the sales staff (e.g. target achievement, commission calculation, time criticality), the end customers themselves (e.g. protection requirements, life situation, life planning) and the strategic goals the management. This triad is best orchestrated by modern technologies and methods that can use intelligent rules to superimpose the respective requirements.

Practical example for technology selection

Here is a practical example: In a project of a large insurance company, all requirements were first collected and two software providers were selected with whom proof of concepts were planned. As an example, sales impulses were generated with these providers and user-friendliness was tested. Typical stumbling blocks could be identified and corresponding requirements added or dropped in an agile process. The selection of sales impulses could already be improved in the test runs.

After the prototypical implementation of various sales impulses with the preselected software tools, a management consultancy was commissioned to check further tools and to assess their applicability according to the dimensions mentioned (functionality and usability for all users, implementation effort and architecture, costs).

The resulting long list from CRM tool providers was filled in for all specific requirements and reduced to a shortlist. Interestingly, one of the two previously pre-selected “top dogs” fell victim to this selection. In contrast, two new CRM software tools were added to this selection process, which is actually open-ended.

During the selection process it was noticeable that some providers of the solutions had only insufficiently dealt with the requirements of the insurer and ultimately wanted to position their solution above all. A necessary look "behind the scenes" with expertise could, on the other hand, cause surprises in the short list, especially with regard to the economic framework conditions.

One of the most important questions in the selection process was the ability of the solutions to be integrated into the insurer's process and application landscape. Here, more recent solutions have shown themselves to be more flexible and modern. Above all, they did not claim to “master” the solution architecture, but saw themselves as an important part of an overall solution.

Conclusion: Pay attention to an open-ended approach

Successful tool selection for both new AI solutions and big data technology tools is based on two simple findings: First, a hasty focus on a certain technology or software often prevents synergy effects from being achieved. Such a focus on one or two providers is usually due to the fact that a single specialist area is natural his Demands and compromises are made too quickly when there are requirements in other areas. Second, it needs a real openness to results! This is the only way to resolve and overcome a deceptive certainty that one of the two preliminary redemptions will prevail through careful research of all criteria in a concrete and cooperative manner: Here, too, the following applies: It is not always top dog in there where top dog is on it.

Authors: Igor Schnakenburg, Managing Consultant and Data Scientist, and Nikolaos Vlachantonis, Partner, both management consultancy Detecon