A study by IAG researchers found Artificial intelligence (AI) and automation techniques improve the customer experience during motor total loss claims as well as the insurer’s operational expenses in handling claims and metrics like churn rates.

Prototypes have now been built and are in the early stages of execution after around 5000 people successfully received text messages warning them of a potential total loss early in their claims process, allowing them to better prepare for that likely outcome.

“There have been no significant complaints made in relation to false positives, which was our main area of concern prior to launch,” the research paper said.

"The techniques used are relatively simple for technical professionals like actuaries to grasp. Tools are free and open source, and individuals can freely train themselves in their use. Any actuary should be able to construct a model of this nature, provided suitable data is available.”

IAG analysis revealed motor total losses was a “standout area” to improve as these losses tend to be more severe and the claims process more complex.

A major challenge was uncertainty over the total loss declaration, which on average took around 15 days from lodgement to finalisation, with a sizeable portion the time taken to assess the claim as a total loss.

The IAG team constructed a model exploring various interventions to give greater certainty, earlier. It ensured that as soon as an assessor determined a claim to be a total loss, the customer was notified immediately via an automated message and asked to provide required information online.

The customer then received a formal settlement offer and payment was processed automatically.

“Waiting for an assessment was simply taking too long. Hence our analytics team explored whether a model could be built to predict a total loss outcome early in the claims cycle,” an IAG research paper said.

“The resulting structured dataset can be used by traditional machine learning regression techniques to predict if a claim was a total loss or not.”

The researchers said their initial test achieved outcomes exceeded expectations, with average gains of around 10 points in customer loyalty measures and no material issues encountered.

The machine interventions “dramatically improved the settlement process for total losses, reducing the number of customer touch points and improving cycle times,” it said.

The paper will be presented to the Actuaries Institute Summit on Wednesday.