Fraudsters love digital


Anyone who believes that digitization has made the issue of insurance fraud go away is unfortunately wrong. One in ten reported damages and claims are still invented or manipulated.

The cost: at least €5 billion per year in Germany alone for property and casualty insurance. In other words, not even including life or health insurance. There is no such thing as a typical fraudster, with the scale ranging from small-scale individual cases through commercial fraud in the grand style to the financing of terrorism through insurance fraud. The industry is starting to realize the truth: fraudsters love digital.

The double-edged sword of streamlining

The standardization, streamlining and acceleration of processes that is necessary for digitization has created new potential threats. These are particularly common in connection with customer-friendly measures, such as the introduction of reporting apps or the removal of original documentation.

The reasons for this are obvious on closer inspection. First, these changes have reduced personal contact between clerks and customers, reducing the potential for the clerk to recognize suspicious behavior from the start. Digital anonymity is particularly important for international fraud. Second, professional fraudsters know that automated fraud detection continues to work with rigid rulesets. Fraudsters test different approaches and know exactly where the weak points lie. According to a recent study on the degree of analytics maturity of the insurance sector, around one-third of companies are still exclusively using rule-based systems.

In order to stay abreast of fraud, especially in the digital age, insurance companies need to stop thinking rigidly, or in terms in rules. Instead, they have to recognize suspicious patterns and anomalies, even if these patterns have not been suspicious before. Only amateurs among insurance fraudsters still make mundane errors like using the same address multiple times (on the other hand, there are still plenty of "amateurs" and they must not be overlooked). Usually, the anomalies that reveal organized fraud rings are far less obvious and very cleverly concealed, however.

Is digitalisation a menace or opportunity for fraud detection? Tune in to this webinar to learn how Advanced Analytics and Machine Learning techniques can help banks and payment organisations to detect and prevent fraud faster and more accurately than ever.

Help is at hand

It is therefore important to look beyond individual cases. Analytical methods look across all cases and find patterns, so they can easily identify statistical outliers. In one quarter, for example, what if there were suddenly far more similar claims, all coming in at the same amount of just under €1,000? A rule-based system with €1,000 set as a limit for relevant damage would not detect any anomalies, and this would not be investigated further. However, an experienced investigator, or an analytical fraud detection system, would be more likely to want to look again. An analytical platform will form a complex score from the parameters of the loss and the customer and highlight issues that might benefit from manual follow-up.

A further step towards explorative investigation is social network analysis and visualization. Here, analytical systems use complete data sets to visualize patterns and cross-connections that cannot be detected from individual data. This includes, for example, multidimensional relationships between people, places of residence, payment links, and claims.

For social network analysis in particular, it is increasingly important not to rely only on structured data (i.e. numbers and tables), but to integrate unstructured information as well. This includes, for example, correspondence with the claims helpline, recording of conversations and even pictures. Today, the possibilities for simple text analysis go far beyond this. Through sentiment analysis, for example, it is possible to analyze the tone of voice used to discuss a particular topic.

What have you seen?

Insurance companies and their investigators need funding to provide simple and visually-supported questions to the system and get answers in real-time. Despite all the automation and non-personalization of digitized insurance companies, fraudsters do not necessarily work in standard ways.

I’m curious about your experience - whether you are part of the insurance community or a customer. Are we striking the right balance between efficiency and enabling fraud?

This view was first published in German, in Mehr Wissen


About Author

Michael Rabin

Michael Rabin assists insurance companys on their way towards digitalization, big data analytics and IoT. Prior to this, he worked as a technical account manager in this segment. He began his career with a German insurer (including businesscontrolling of property / casualty insurance and strategic bankassurance) and is therefore familiar in the field of classic insurance. Michael Rabin unterstützt Sie als SAS Account Advisor für Versicherungsunternehmen auf ihrem Weg in Richtung Digitalisierung, Big Data Analytics und künstlicher Intelligenz. Zuvor war er als Technical Account Manager in diesem Segment tätig. Seine Laufbahn begann er bei einem deutschen Versicherer (u.a. in der Geschäftssteuerung Schaden-/Unfallversicherung und im strategischen Bankenvertrieb) und ist daher auch im Bereich der klassischen Versicherungsthemen zuhause.

Leave A Reply

Back to Top