Identifying claims leakage is essential for insurers seeking to control costs. Sam Barrett investigates whether new analytics and data visualisation hold the key.
With insurers working on tight margins, controlling claims costs is essential. Just a small increase in claims leakage can quickly turn a profitable organisation into a loss-making one.
Among the latest tools available to help insurers stem claims leakage are data visualisation, which generates reports in a format that is much easier to interpret than the more traditional spreadsheet crammed with numbers. And 'third generation analytics' are offering insurers greater data-crunching capabilities.
The power of these analytical tools was highlighted by Steve Bell, financial services advisory partner at Ernst & Young, at Post's Claims Club meeting in February. He explains: "Data visualisation is a much easier way for the human brain to identify patterns. It can spot a trend much faster and use this to highlight outliers and other anomalies."
Within the insurance arena, harnessing the power of data could deliver significant benefits. For example, rather than relying on the labour-intensive process of a closed file review, where an experienced claims handler is taken off the frontline to sift through hundreds of claim files to generate an idea of trends, insurers could automate the process with these analytics. Then, through data visualisation, any trends could be instantly highlighted. But to what extent are UK insurers employing such tactics to tackle claims leakage?
Imran Ahmed, claims advisory leader at Ernst & Young, comments: "An insurer can look at this rich data, taking those with signs of claims leakage and identify any discernable patterns. This profile can then be applied across all its claims to show the extent of the problem. Additionally, with a better understanding of how leakage occurs, the insurer can consider how to change processes and products to minimise the problem." On top of this, because of the power and speed of today's available analytics, trends and patterns identified can also be applied to live claims to show where leakage occurred in the past. This means outliers can be spotted quickly.
Clearly, this carries cost implications, enabling insurers to potentially reduce their spend on these claims. "By spotting these claims early, an insurer can avoid taking action that will cost money," says Edwin van der Ouderaa, managing director for financial services analytics at Accenture. "If triggers are activated, the claim could be passed to a more experienced claims handler or different questions could be asked."
This also avoids one of the problems that can plague insurers' claims departments — the power of the large claim. Andy Williams, managing director at Isis Claims Solutions, explains: "In many claims departments, the focus for leakage is on the big claims. Smaller claims will be dealt with quickly, with some insurers setting limits as high as £3000 before a claim is examined more closely. But leakage can occur on these small claims and, unless an insurer is monitoring this, the numbers will quickly add up."
Putting a figure on the financial benefits these new tools could deliver is difficult. But taking just one part of the problem, Mr van der Ouderaa says that with fraud alone accounting for 5% of premiums, the savings on claims leakage could be huge: "These analytical techniques promise to reduce fraud by 20% to 30%. This is a huge savings on top of the ethical benefits tackling fraud brings."
Analytics can also help insurers stay ahead of changes in claims leakage. These could happen for a number of reasons, as Mr Ahmed explains: "Fraud is constantly evolving but you can also get increases in leakage as a result of events, such as a merger between insurers, consolidation in claims handling, or a change in operations."
While a manual, closed file review might be conducted every six months or so, allowing the claims leakage to drip through unchecked on a considerable number of claims, these tools can be run much more frequently. Furthermore, as they run behind the scenes there's no additional demand on resources. "If you were really on the case it could be a continuous process, enabling you to pick up any new trends as they happen," says Mr Ahmed.
Given the power of these tools, it is not surprising some sectors have already adopted them. In particular, they are widely used in retail, engineering and among mobile phone companies. For instance, in the retail sector, data from reward cards is used to enhance customer loyalty and build targeted marketing campaigns. Data visualisation helps to highlight the types of product that sell well in a certain geographical area or to a certain type of customer — and this information is then used to market similar products to these people.
But take-up of such tools apparently remains low among insurers. By way of an example, a new management information system is being introduced at Allianz Insurance but it doesn't incorporate data visualisation tools. Property claims manager, Harry Rule, explains: "Data visualisation is really a 'nice-to-have', rather than an essential. Our current system works well and we're able to do a lot with data and text-mining to counter fraud. Unfortunately, when it comes to claims leakage, you can't really get away from the slog of getting your best claims practitioners to review the claims."
Actually implementing these tools may also prove difficult. Thomas Townson, senior manager in risk and compliance at KPMG, explains: "The claims landscape is not in a state where these techniques can be easily applied. In theory it would be a step forward but, in practice, it's one that's difficult to realise."
There are plenty of obstacles for insurers. The technology they have in-house is the first of these, with so many insurers relying on a patchwork of legacy systems and old technology to process their claims. Paul Whiskin, senior insurance consultant at IT-Freedom, sees this as potentially the biggest barrier. "Claims are very much an afterthought and systems tend to be policy systems with some claim recording added on. Furthermore, if an insurer does manage to get data out of the systems, that requires a process change, which often needs to be hardcoded. That can take another six months," he says.
Insurers themselves also acknowledge problems exist. David Williams, claims director at Axa Insurance, says: "We have multiple systems and I have to say that, so far, we've concentrated on getting data out of the systems rather than how we present it," he says. "It would be good to delve into the detail, and we do run programmes to highlight the outliers, as well as running closed file reviews, but issues with systems mean this is not something we're about to invest in."
The patchwork doesn't stop with systems either. Many insurers outsource some of the claims process, so data can be held by third parties too. This can also raise issues on ownership of the data, introducing further hurdles for those looking to feed data through these tools. "Collecting complete and accurate data is vital," warns Katie Doyle, director of product marketing at Guidewire. "If the underlying data isn't accurate, then the findings can be meaningless." There are further data issues too. "Many claims files are paper-based," says Mr Townson. "Additionally, they can be located all over the place and, if the insurer outsources part of the process, there may be issues regarding accessibility and ownership of data."
The way many insurers collect claims data could also jeopardise the validity of this form of analytics. With much of the data collected by telephone, errors are possible. Unfortunately, a simple spelling mistake in someone's name or their address details could mean they are not picked up as a multiple claimant and potential fraudster. Because of this, it would be necessary to spend time cleansing the data first by going through it manually to pick up any problems. Damien Margetson, director of risk and compliance at KPMG, adds: "When we've implemented these tools, we've had to spend as much as two-thirds of the time getting the data into a state where we can use it."
Newer players might experience fewer problems. Without legacy systems and with less outsourcing across the claims book, it could be easier to implement analytics and data visualisation to counter claims leakage. But, whether an established or new player, cost remains an issue. "Insurers would need to invest heavily, not only in these systems but also in improving their data. In the current economic climate, this isn't very likely," says Ms Doyle.
In addition, because of the scale of the investment, management buy-in is essential. But, as Isis Claims Solutions' Mr Williams explains, this can be problematic. "Management at insurance companies tends to be fairly transient. One day, the board might support the investment, but the next day you've got a new board that doesn't want to make the investment," he says.
The very nature of claims leakage also creates an obstacle. Although figures are bandied around about the scale of the problem, no insurer truly knows how exposed it is to leakage. Making a business case for this unknown is, therefore, rather difficult. But Mr Margetson believes this will change. "Ultimately, and driven by regulatory pressure, it'll become a requirement. This has happened in other financial services sectors — for instance, occupational pension schemes must now validate their data to ensure the right benefits are paid to members — and I expect insurance will move in a similar way," he explains.
Certainly, the signs are already there. Under the imminent Solvency II regime, there is a greater reliance on data in the reporting requirements. This will push insurers to look at ways of improving the quality of their data, which will create a more viable environment for tools such as data visualisation and third generation analytics. Additionally, as the technology matures, it will become more widely accepted. "It's a fast-moving field and already the concepts and technology are becoming more democratic," says Mr van der Ouderaa. "This will make it easier for more insurers to access it. The business case is too large to ignore."
Data visualisation and third generation analytics in practice
These tools enable connections and trends to be easily identified across claims data while also creating patterns from past claims to highlight any anomalies in current and future data. Here are some examples of how they could be used in the insurance industry:
By bringing together data from different lines of business, an insurer could identify inconsistencies in the size of claims. For example, if someone makes a large claim on their home contents insurance this might be investigated if, upon cross referencing it with their motor insurance policy, the insurer finds they drive an old banger worth a couple of hundred pounds. Similarly, profile data collected on an application form might lead the insurer to question how someone on a low income came to have amassed a collection of the latest high spec audio and television kit.
A motor insurer could use data visualisation tools to identify any claims that have a common witness or address. Likewise the methodology will highlight any potential fraud rings that share a common garage or loss adjuster. This could be an indication of fraud.
Using data visualisation tools on claims values will highlight any clusters. Of particular interest will be clusters just below the authorisation level. This could indicate fraudulent activity, with fraudsters aware of the size of claim they can push below the radar.
The customer acquisition process can also benefit from data visualisation tools. By using data visualisation to gain a better understanding of existing clients and their values and needs, an insurer could target marketing messages to attract similar clients and build their customer base.
- Roundtable: Is a single customer view taking off in insurance?
- O’Connor replaces Fairchild at the helm of Broker Network
- Home insurance insurtech Buzzvault launches
- Stackhouse Poland makes fourth acquisition of the year
- Aviva promises to 'reinvent' insurance and end dual pricing
- Ed unveils CEO Hearn’s replacement and plots Bermuda office
- CBL Corporation expected to be placed in liquidation, sees further delays to watershed meeting