In 2014, UK insurers uncovered 130,000 fraudulent claims worth £1.32 billion across all insurance products. This danger shows no sign of slowing either, with professional and opportunistic insurance fraud numbers still on the rise. So, how can insurers effectively analyse data help to combat insurance fraud?
A major insurance company has proactively combatted fraud before a fraudulent claim is filed or before they would even offer the customer a policy. It deployed an integrated data platform which can instantly access and analyse data to assess if the claim or request was likely to be fraudulent. With an integrated platform, insurers will be able to scrape and analyse social media information against other sources. Sites like Facebook and Twitter can prove very valuable in helping to spot fraud, giving insurers a more rounded view of their customers.
Telematics: putting insurers in the driving seat
In addition, telematics-based solutions are helping insurers reduce fraud and manage risk effectively using big data technologies. Organisations such as Octo Telematics have transformed how insurers assess risk, deliver crash and claim services and detect fraud. With a sophisticated Internet of Things (IoT) strategy, Octo can now analyse 11 billion data points daily from five million connected cars. Machine learning is helping the company to make more accurate predictions and risk models. It works by fitting a black box to a customer’s car, which constantly records information such as GPS location, driving speed, distance and time of drive, monitors rapid or smooth acceleration style and braking and cornering habits.
Take for example, the common road injury of whiplash. Claims are 50% higher than a decade ago, despite the UK having some of the safest roads in Europe and a decrease in the number of accidents. This has been fuelled by a predatory claims industry that encourages minor, exaggerated and fraudulent claims, driving up the costs of insurance premiums for ordinary motorists.
Telematics can provide crash reconstruction dossiers that can show, with a forensic level of detail, where the vehicle has been damaged as well as other factors such as the speed and angle of impact. From the reconstruction it’s possible to conclude that vehicles were not travelling at sufficient speed to generate a whiplash injury claim. This means millions of motorists can now enjoy reduced car insurance premiums and responsible motorists will no longer have to pick up the tab.
Machine learning: a force to be reckoned with
Advanced analytics, and in particular machine learning, is key to delivering the efficiencies that insurers need. Machine learning is creating many new opportunities for insurers in a number of areas.
One of these is improving on the performance of traditional analysis or human judgement through ‘submission prioritisation,’ which helps insurers to predict premiums, conversion and how much a policy is likely to cost them more accurately. The ability to identify high-risk policies early on and spot more valuable business opportunities will save insurers time and money dealing with expensive and complex policies down the line.
Another area in which machine learning is helping to power smarter decisions is when a claim has been made and a settlement needs to be reached. Insurers can save millions in this area by reaching settlements faster, carrying out smarter, more targeted investigations and managing cases more efficiently. Being able to spot these complex claims early ensures insurers can get on top of the situation and reach a less expensive and resource-intensive resolution faster.
To make the most of machine learning, insurers need to build or reshape their business model around advanced data analytics and ensure everyone within the company is working towards this goal. Integration, both of providers’ internal teams and processes and of the data itself, is key to success. An integrated data solution will enable insurers to automate decisions and mine valuable insights safely, and this will be key to gaining a competitive advantage in this sector.
Case in point; a leading UK-based general insurance company used machine learning to significantly improve the performance of its insurer hosted rating hub. It had been processing up to 25 million quotes a day and its systems were struggling under the pressure, leading to increasing costs and making it a labour-intensive process.
Today, it instantly analyses data from sources including policy and claim systems, adjuster notes, social feeds, weather data, traffic patterns and third-party data. This has resulted in a 120% increase in quote to policy conversion over 18 months.
The big data era has already had a dramatic effect but now, with the innovation and automation on offer through machine learning, advanced analytics and telematics, insurers are able to get more accurate insights into human behaviour that drive better business outcomes in critical areas such as fraud detection and prevention.
Machine learning will become an essential part of insurers’ business models, helping them continually improve the efficiency of decision making and to keep pace with sophisticated fraudsters. Fraud has become a machine-scale problem, and to fight it will require a machine-scale solution.