In this latest piece for Insurance Edge, Mark Rayner, Head of Financial Services Consulting at BAE Systems, looks at how companies can get the best value from data scientists, when it comes to identifying and fighting fraud.
Data scientists are viewed as hot property within the insurance industry right now, which is understandable given the extra insights they can offer. But to be truly effective data scientists need to be embedded within a counter fraud team, so that the maximum benefit can be gained. There’s little point in discovering patterns of potentially fraudulent activity, fake claims, data breaches or multiple attempts at by-passing security, unless those insights are acted upon immediately – and managed effectively.
Let’s take a look at three ways we can get the most from data scientists;
1. An end-to-end approach, so that we are following the fraud from the first red flag to the claim resolution.
2. Proven detection techniques, that can be fine tuned and adapted to an insurers particular portfolio.
3. Access to industry expertise worldwide, maintaining a constant learning programme and implementing best practice.
There’s a bit of hype about data science and of course, it isn’t a cure for all fraud-related ills. The key thing is understand the data and take the correct actions and often this means getting data scientists to truly work as part of a team. It is easy for people to slip into a kind of silo system at work, where their own everyday targets and goals are what matter, not the bigger picture.
One way around this is to establish an advanced fraud analytics platform within your company, so that your data scientists are getting the benefits of more case studies, raw data, previous fraud patterns and input from third party specialists within the counter-fraud sector. Nobody can argue with your company’s deep knowledge of a particular niche within insurance for example, but there is definitely a value in having a wider view, and support, so that your scientists can deliver a real impact day-to-day.
MULTI-FUNCTIONS, WRAPAROUND SUPPORT
Your data scientists may be required to assist the marketing, or product development divisions within an insurance company or larger MGA/broker. Having support when it comes to fraud reduces the chances of something vital being missed. Following a fraud case to its conclusion may well involve staff from claims, compliance and other departments, so again, having the back-up from other parts of your fraud analytics platform, just when you need it, will help you get the right result.
In short, a good data science platform will enable your company to join the dots: Identify risks, dubious claims or data breaches and then triage various cases. You can build a holistic case management system, so that all actions can be noted and everyone in the loop knows that their input made a difference to the eventual outcome.
KNOWLEDGE IS POWER
Getting a third party vendor involved also means utilising a steady stream of knowledge and insights from around the world. You aren’t replacing your data scientists, but adding something to their expert analysis, as you filter through the latest news on fraud attempts in other countries, receive alerts and understand the latest hacking or ransomware threats etc.
The biggest advantage gained from working with an experienced specialist company is that your data scientists are getting an edited, curated flow of news and insights, that is truly actionable. By applying leading edge techniques and skills from around the world, your company is likely to save a great deal of time and money, because having the right data and counter fraud support 24/7 means your data scientists have that extra resource to call upon.
By building a platform that has clear objectives and a complete suite of tools and knowledge, you install a powerful defence against insurance fraudsters. Data scientists can work wonders, but not alone, so pooling everyone’s knowledge is always going to help produce the right result that bit quicker. In an era of rising cyber attacks and fraud attempts online, that has to be good practice.