Innovation begins with finding problems to solve. There are no shortage of these that already exist for insurers and other organisations operating within the insurance sector, and that is even before one considers the new problems that will be coming down the line through such diverse factors as:
- New or amended legislation
- New or amended regulation
- Behavioural changes in opponents encountered in claims
- Your own evolving operational pressures and risks
Innovation is enabled by technology and technology is powered by data. We have been experimenting in using Generative AI to extract key data points from various types of claims documents. Early results have been promising in terms of the accuracy of the outputs, the speed with which the outputs are produced, and the range of different document types with which it can cope.
As the designer of many algorithms which operate on key data points available in claims documents, we were also interested if Generative AI might be able to 'convert' a piece of information it was asked to find within the claims document to one from a strict list of options that we gave it. From our testing it is clear this can be done and with good accuracy.
Of course, the challenge is being able to find a way of directing the data extracted from documents into data warehouses to be used for data-driven solutions. A winning feature when it comes to innovation – the ability to scale a solution.
Scale:
- To do something at high volume
- To do something in an automated fashion
- To easily transfer a solution to different classes of claim
- To easily transfer a solution to different jurisdictions
The last two are highly specific to insurance claims.
Let us now bring 'scale' and 'volume' together. One of the challenges for insurers lies within the volume claims environment, whichever class or classes of claim they deal with. Claims are notified as low value, high volume – but which claims will stay that way, and which will turn out to be something else altogether is difficult to anticipate.
If we take personal injury claims as an example, which in the RTA, EL and PL claim classes, start life in a low value claims Portal. These low value Portals are designed for low friction settlement, but plenty of those claims are not suitable for this, for a variety of reasons – for example, fraud, claims farming and claims which deteriorate to high value. Insurers need to identify these claims as early as possible to get them to the right teams for the appropriate strategies to be applied. Human claim handlers must make these decisions, but all too often, these decisions require new information to be disclosed by the opponent claimant solicitors. And if those solicitors don’t want to disclose that information for any reason – they won't.
How then do insurers identify various different types of claim risk – the risk of the claim being something altogether different to what it has been presented as – at scale, and as early as possible in each claim's life? The claims data has the answer, and DWF have designed an algorithm that can be applied at scale to an entire portfolio of claims to produce risk scores as outputs, allowing insurers to manage their resource more effectively by prioritising review of those claims with high risk scores. We have successfully deployed this solution and believe the methodology is transferrable to various different classes of claim, creating significant measurable value for insurers in different ways.
Our new innovative tool will ultimately assist clients predict more certain outcomes, streamline processes and enhance efficiency.
If you wish to discuss this in further detail please get in touch with our team.
