We know that 2020 saw the pandemic accelerate widespread adoption of digital channels by businesses and individuals alike. But there is a dark side to our societies’ greater and deeper digital dependency; a parallel rise in fraud and online crime. Cyber security experts have reported on a ‘gold rush’ of COVID-19-themed cyber attacks globally, and insurers are encountering a similar rise in assumed fraud rates.

A study from insurance fraud specialists, FRISS shows that industry professionals report an increase in claims containing fraud from 10% to an average of 18% of all claims. The sharpest rise in 2020 was fraud related to staged incidents, vehicle theft, procedure billing and phantom services. The new economic and social norms are ideal props for business frauds based on home working or how people might be using their personal vehicles for business purposes, like a restaurant that has pivoted to takeaways.

As the economy unlocks, we must expect a surge in business insurance fraud fuelled by the way in which the pandemic caused such economic damage – as was the experience during other periods of economic upheaval. High up on the list must be how some may make fraudulent claims about lost or damaged stock and equipment or by fabricating their pre-lockdown financial circumstances.

In their hunting fraudulent behaviour, insurers must tread carefully, especially after the fallout from the mishandling of business interruption claims. Fraud investigations can risk delays in claims processing and upset honest customers.

Indeed, a dominant characteristic, evident in the increased use by organisations and individuals of digital channels, is their intense impatience with slow response on a quote or claim approval.

Speed and efficiency have long been highly rated by customers and ahead of human interaction, personalisation and use of up-to-date technology. So even the slightest pause in the process risks causing tension and a breakdown in trust.

Despite this, too many insurers are still relying on their staff to spot suspicious behaviours and manual interventions to track and investigate fraudulent activity. Indeed, despite one in two insurers believing the use of automated fraud detection tools will improve their fraud-fighting, more than 64% currently use the experience of staff to detect fraud. And, of course, COVID-19 restrictions have affected fraud inspections too, due to the impact of remote working on insurance employees and related professionals.

Prioritising technologies that be both nimble on service and forensic on fraud eradication must be part of the answer. And that means more insurers adopting AI, machine learning, and behavioural analytics to sense and judge fraud risk while simultaneously reducing friction for trusted, but increasingly fickle, customers.

Yet it is not about simply throwing more tools at the problem in an attempt to fix it. What is essential is how AI and machine learning become integral to straight through claims processing for honest claims, by automating fraud scoring in real time, and supporting the detection and investigation of fraudulent claims.

Talking to fraud experts, it is clear that AI is a powerful tool but you need to understand how it benefits the fraud detection process specifically. AI is very effective at detecting known fraud types and some unsupervised use cases directly. When there is not enough proven fraud data, AI can be automatically set to retrain not just from daily feedback, but also from other data sources like business rules, third party data and network analytics. This enables the AI to develop new models to detect frauds. Essentially what you want to create is a closed loop for fraud data analytics applied at all points in the insurance process.

When properly designed and managed, integrated solutions are claimed by some vendors to have hit rates of 75% or higher (Source: Shift) when processing claims because of their rich, accurate, real-time fraud scoring. This smart use of smart technologies will be central to disrupting fraud while delivering fast, friendly customer experiences at every digital touchpoint.

How insurers get to this endpoint does need to start with how an organisation formalises its AI resources, perhaps setting up an AI centre of excellence, to drive change. It is really essential for insurers to understand the value of AI to their businesses. When there is pressure to move fast on this issue, insurers should be able to access technology partners who can help inject AI into key processes. It also helps if the insurer has modernised their core technology around cloud and cloud-native technologies to give the computing capacity and scalability for AI and machine learning to digest and analyse their data.

Using AI and machine learning should be a part of the everyday processes involved in serving customers – built in, and not an annex of an insurer’s technology architecture.

One year on, and as we start to emerge from the pandemic, most businesses, regardless of size, are more digitally dependent, technology-driven organisations than before. To understand them and their risks overall, insurers must understand an array of much more varied and complex data sets and take a more advanced data-driven underwriting and claims management techniques.

Predictive analytics and especially AI can crunch and interpret a new ocean of risk indicators like demographics, crime rates and sentiment, and generate much more meaningful risk profiles to price policies and accept or question claims.

AI is no silver bullet or cure to all insurers’ ills, but when properly applied and integrated with other systems improvements, it should enable insurers to push back and control fraud growth.

Share Story: