Customer satisfaction is highly competitive in the insurance industry especially when it comes to making a claim. Consumers expect quick and simple interactions with their insurer and a speedy settlement of the claim. However, insurers have to strike that delicate balance of faster claims resolution, without increasing the amount of exposure from fraudulent claims.
Insurers can of course resolve this by hiring more investigators, but that means extra costs, and squeezed profit margins or having to pass the extra costs onto customers in terms of premiums. Technology that can help validate claims is crucial to speeding up the process and delivering efficiencies; however, with numerous technology solutions in the market, it can be a problem to know and understand which one is best for you. In this article, we pick apart the core elements of a successful claims investigation so you can better evaluate a provider’s offering.
Automating Enhanced Investigations
When a technology provider says that they can help automate the claims process it is important to understand what aspect of the process they mean. Validation and fraud detection is an investigative process that necessitates the need to aggregate data across multiple data sources, filter it, organise and analyse it, then prepare reports that justify your decisions. If a platform only helps automate the claims’ submission process for applicants – for example, the extraction of data from the claims’ form, and the adding into a workflow for next steps but – leaves you having to do all – or most – of the data gathering, filtering and finding hidden links and connections in the results manually, it means it is only a case management system, and is just one part of the puzzle. You still need an investigation solution.
Multiple Data Sources
Claims fraud detection investigations, like any kind of investigation, are complex. Different data sources provide different insights into individuals or businesses that can be used to evaluate and discover hidden risk. For example, an insurance claim might appear legitimate but then a search of Companies House might show that they recently had several failed businesses or an association with a company that is in fact the supplier of services in the claim. Besides the incredible number of commercial data sources available, there are consented sources, consumer, and open source.
As multiple data sources are essential for any investigation, any solution provider should have the capability to offer automation that utilises them, rather than just focus on the patterns that can be identified from internal data – no matter how large your pool of data.
Fraudsters are getting more sophisticated in their approach, meaning search and risk assessments have to be smarter, inevitably needing to utilise data from across more diverse datasets. In most cases, decision making relies on data from upwards of 7 or 8 different sources in order to achieve the correct level of due diligence. Therefore, you should not only look for a solution provider who can automate over multiple external datasets, you should go a little further and ask if they have a data agnostic approach.
A data agnostic approach means that they do not limit themselves to a fixed number of data suppliers but will work with you to ensure all the data you need for each, and every kind of claims fraud risk assessment is aggregated and then can be automated. This way you maximise the amount of automation in your investigation and operational efficiency.
If the solution provider only aggregates their own proprietary data, you may find that you can indeed have seamless automation across their datasets, but then have the time-consuming headache of trying to integrate crucial information from other sources manually.
Cross Data Source Analysis and Collation
Multiple data sources are key to any claims’ fraud investigation because each of them reveals different information about a name, phone number or email address. But searching different datasets is just one part of a risk assessment – aggregation and analysis of the findings to detect links between entities across the different results, is equally important.
For example, if you are investigating an insurance fraud and searching for information around the claimant, third party, and the repairer ecosystem, you may find the car had been owned by them both before, that they live on the same street, that they attended the same school, or that they have similar directorships.
It is imperative, therefore, that, rather than just allowing you to set up workflows on each data source as a silo – in effect acting as just a portal – any automation solution is able to carry out automated aggregation and data collation across all your data sources. Visualising the associations between the entities allows you to join the dots and more easily discover hidden links. In other words, give you a unified, holistic view of your subject and eliminate the need to spend the time to manually read through each set of results to identify the links above.
Risk assessments are key to identifying and fast-tracking genuine claimants to improve customer satisfaction. In addition, reduce the burden on claims teams by allowing them to focus on more complex claims’ cases.
Any solution provider should therefore offer the ability to set up sophisticated risk assessment rules. But for a rule to successfully assess complex risk, it has to collate across multiple data sources concurrently, as each will have its own impact upon the investigation. If not, your risk assessment workflows will only work on each data source individually, which means operational efficiencies.
When risk assessments have identified potential or likely fraudulent claims, the next steps are likely that your team will need to investigate the individual further to build a case or dismiss it as a false positive. It is probable, you will be required to present a report with evidence to support your decisions.
You should therefore ask yourself if a solution provider has the capability to automate report preparation in a standardised fashion with all activity logged to the appropriate standard. If an investigation might result in legal proceedings before the civil or criminal courts, you will need to have increased levels of auditing capability governing the data sources interrogated, the person conducting the search and on many occasions the grounds for processing the data.
Whether for further investigation efficiency, measurement or continual learning and data integrity, it is necessary to understand whether the risk assessments and their methodology are the proprietary IP of the software provider or transparent to you as the customer. In the event you simply have to trust the software provider that the flagged cases are fraud, it has far-reaching implications in many areas:
Case Management – Without the sources of evidence why a particular case was flagged, it means you will effectively have to start again in your investigation to build out the case, spending much more time.
Operational Efficiency – Being able to see cases with evidence, enables someone to quickly evaluate and use their decision making to filter again and move the process forward faster to identify false positives.
Measuring Fraud Detection Success – Without access to the algorithms being used by the solution provider, it is hard to see if they are accurately discovering fraudulent activity.
Process Improvement – Without access to why cases are being flagged, there is no way to get the data that will enable you to learn and improve processes and efficiencies.
Underutilisation of expertise – Your fraud team are experts in identifying fraudulent behaviour, if they are unable to have input across your risk assessments then it is not efficient.
The process of investigating claims fraud is not limited to the initial risk assessment stage but includes the process of further investigation. While some solutions in the market may address those initial risk assessments based on internal data, the latter cannot be achieved without multiple data sources.
If your provider cannot offer automation for validation, fraud detection and then for the investigation stage, then you will have to do the enhanced investigation manually or look for two solutions and then have the added inefficiencies and costs from syncing data between the two solutions.
At best, it is a method of narrowing down to the field the more likely cases to do further intervention which requires manual investigation.
AI and machine learning have incredible potential, but it comes with several considerations. While AI is used to detect additional patterns inside the data, if the focus is simply on internal data, can it make up for the shortcoming compared to using a more holistic approach with multiple datasets?
Inevitably, when you have flagged certain claims and decided to reject them, if it then goes to a dispute, will the evidence from the AI be available or inadmissible? If not, you will still find you have to build out the case again using a traditional data gathering strategy over multiple data sources. In which case, the AI is limited to the initial risk assessment filtering stage and even then, you will have many of the problems associated with a lack of transparency in methodology.
In more complex investigations, it is likely that you will need to follow up on possible links between entities and build out a trail. Ideally your technology provider’s solution will be responsive and scalable enough to allow you to automate that process, offering the tools to update your results seamlessly – in real-time – and get visualisations of the impact of your new variables on your overall investigation. This may be the ability to conduct follow on searches linked to the earlier findings, or it might be the need to monitor ongoing changes for that business or person.
Investigations are exceptionally time-consuming and a burden for your investigative team. Any solution provider will tell you their solution is best, but how much time can they give your analysts back to focus on decision making? This is the key measurement.
If you’re in doubt as to how much efficiency a provider might bring, challenge them to provide you with a free pilot or ask to speak to their references.
While the real savings and efficiencies lie in their ability to offer a genuine aggregated automation experience across all your chosen data sources, there are other questions you might want to ask:
Knowing how to choose the right technology to help create operational efficiencies and gain better insight for your fraud investigations can be complex and confusing, especially when there are many providers in the market. However, you should ask yourself the question: Does this help me complete an investigation with as few, or ideally without, any manual processes? If you don’t get a satisfactory answer, you should be looking elsewhere.
Synalogik’s software platform, Scout®, is a one-of-a-kind automation solution for EDD checks and investigations. Scout is data agnostic, integrating internal, open source and out-of-the-box most 3rd party data providers, allowing you to seamlessly automate search and reporting across all the datasets you use, not just the ones included from your existing solution provider. Our 3rd party integrations include CRIF, Equifax, W2 Global, LexisNexis, Creditsafe, TransUnion, GBG and many more.
As every customer has their own investigation requirements, it inevitably means that they end up using additional datasets outside of one single solution provider and any automation isn’t seamless. Our open approach means it is possible to have more complete automation across all your datasets, delivering greater efficiency and insight.