Interview: New Weightmans data analytics tool saves average of £285k per case 

We speak to Weightmans partner Will Quinn about Weightmans new tool PREDiCT, which is based on 10-years of data and will help insurers to settle cases faster.

After 10 years of collating and analysing data from its large loss claims, Weightmans has launched a tool that it says can save insurers an average of £285,000 a case, by reducing the reserves they have to commit and the time it takes to settle cases.  

PREDiCT has been developed by Weightmans’ in-house data scientists based on data from more than 1,000 individual large loss claims, which are typically six-figure serious injury claims. The data has been collated and tagged in Weightman’s case management system, Thomson Reuters MatterSphere. 

Weightmans partner Will Quinn specialises in helping insurers with large loss claims and has been a partner since 2000. Explaining the need for the tool and the problem it solves, he told Legal IT Insider: “Say a cyclist goes over a pothole and breaks their neck and ends up tetraplegic. We will often be instructed by an insurer within a week. They will want us to do an investigation and they will put on a reserve to cover their financial exposure. We are talking about multi-million-pound cases, and we were finding that there was no consistency in approach.” 

Estimating a reserve often happens in a vacuum, without sight of medical records, however by analysing data from previous cases, it’s possible to say more accurately what the pay-out is likely to be. Quinn said: “We found that in 41% of cases they over-reserved on damages, which equates to a sum of £288m. So the insurers in those cases had set aside £288m that was released when they reached settlement, locking up sums that impact on their profitability.” 

The data exercise started a decade ago – long before many firms had worked out the importance or value of their data – and Quinn said: “We deal with hundreds of cases a year and had access to lots of data, but we weren’t collating it in a meaningful way. So, we set off on a data journey 10 years ago and it’s taken that long to reach a size that we can use.” 

The firm has been very granular in the data it captures – there are over 240 fields for each case. These include key indicators such as how long the patient is in hospital and whether they were wearing a seatbelt if they are involved in a road traffic accident. If a claimant wasn’t wearing a seatbelt, they lose 25% of damages, but that wasn’t necessarily being accounted for in the early stages. 

Quinn says: “We worked with our data scientists to create a model and input key indicators into the model to generate a range of outputs.” Those indicators include likely damages; claimants costs; and the lifecycle of the claim. 

There are two ways that using this data improves the result for insurers. Firstly it helps them to improve the accuracy of their reserves. But secondly, it can accelerate the speed with which a claim is dealt, by enabling an insurer to make an offer at an earlier date. 

The data analysis exercise has been extensive, and Quinn says: “We recognised that for this to be meaningful analysis and benchmarking that we need to know that we are comparing like with like. It took us a lot of time to think about what the features are of large cases that will drive outcomes. So how long were they in hospital? How long were they on a ventilator?”  

This inevitably requires that case handlers capture a lot of data both at the beginning and during a matter and Quinn says: “It wasn’t that popular when we set off.” However, the knowledge that insurance is a competitive market and that this kind of intelligence will give the firm an advantage has been the ‘carrot’ that complements a fair amount of coercion and Quinn says: We have a good team and people understand the value of the tool and its importance to clients.” 

The tool is particularly strong in some areas and Quinn says: “In some areas we could do with more data but in areas such as brain injury, we have good data and are confident that the model performs well.” 

The data scientists began working on the tool around two years ago and it has gone through various iterations. There was a lot of engagement with clients at the outset around what the firm was trying to achieve and now its live, Weightmans is tracking PREDiCT’s performance. Quinn says: “Some firms are really engaged and want to drive performance and consistency. Others want to see how it performs. It’s a journey and we are at different places with different clients.” 

Quinn is keen to stress that the model is not a substitute for the firm’s lawyers or case handlers and says: “It’s about trying to support them with a tool that will validate their processes,” adding, “Insurers can use the model to increase their confidence and make an offer even when they don’t have all the evidence to hand.” 

Some firms worry that tools such as this will cannibalise their income by, for example, reducing the amount of work required. Quinn is candid that they are still working out the costs structure.  

One would hope that insurers – or any clients for that matter – will recognise that work completed far faster is worth more to them, not less.