Saving money and improving lives: BLM’s head of analytics discusses the new LSE partnership

BLM in mid-February signed a two-year research and development partnership with the London School of Economics and Political Science (LSE), in which the UK top 40 law firm’s in-house data analytics team will work with three LSE professors to develop new litigation risk management offerings for BLM customers.

Legal IT Insider caught up with BLM’s head of analytics, Andrew Dunkley, to find out more about the new arrangement.

How did your arrangement with LSE come about and what are you trying to achieve?

I joined at the end of 2016 and over the course of the last working year, my team has continued to expand and the work we’ve been doing has become more sophisticated, but we’re finding that a lot of the gains in this area come from being able to put together different skillsets and perspectives. People often ask, ‘what piece of technology have you bought’ and we tell them ‘we’ve decided to invest in people, not technology.’

There are lots of brilliant bits of technology but what we’re trying to do with LSE is bring real expertise and brain power as to how you value and manage litigation risk. All the technology that is trailed in this part of the market is trying to help answer that question. The logic behind this partnership and the collection of skills in it is very much to say ‘we’re not going to focus on buying or building a snazzy bit of technology.’ Yes, AI will be a big component, but we think we can do more by blending skills to help manage litigation risks in the two strands we work on – volume litigation and lower volume, higher value claims.

Are you risking reinventing the wheel?

No, one because we make sure we keep abreast of what is out there and if we do come across a use case already solved, ask if we build it ourselves or procure it.

A lot of the AI solutions on the market are about data extraction – people get a bit confused about what AI is and the use cases. Fundamentally most solutions are predicting classification. Once you’ve got to that metadata you still need to build a predictive model off that, and that sort of model isn’t something that’s productised in the UK.

So who is involved from LSE?

Professor Henry Wynn will be the lead contact and have the majority of the input: he is a statistician with a strong background in decision science, so asking ‘how do you best make a decision in an uncertain environment where you don’t have all the information you’d like and there may be a cost in obtaining that further information’, which is a pretty good analogy with litigation. He’ll be the lead point of contact and we will also draw on Professor Pauline Barrieu, head of LSE’s statistics department, who has extensive experience of the insurance sector (a traditional area of strength for BLM); and Professor Milan Vojnovic, who is chair in data science and an expert in machine learning.
LSE is appointing a dedicated post-doctorate research officer who is 100% dedicated to the process and will spend a lot of time at our office working with us. It’s very much a hand in glove partnership and a serious commitment.

Ultimately are you hoping to create a product?

We’re hoping to create an offering. In relation to the volume claims business, that is likely to be based on one or more predictive models and we’re whittling those down. How that then gets implemented in a commercial sense is likely to be driven by where the research ends up – it’s likely to be an offering wrapped into what BLM currently provides.
There are various different ways of getting value out of that kind of model. The risk in all of this is that you set out saying ‘we’re going to build a computer program that solves the litigation problems of the world’ and there are a few players in the market that are setting out to be that ambitious and maybe it’s possible. But I do know that if you say you’re going to do that, it might be quite some time before you start extracting value, so we’re going to develop the model and find ways of implementing it so we don’t have to wait five years for an all singing, all dancing computer system. We’re trying to be pragmatic in delivering value out of the research.

What is the best outcome you can envisage from this partnership?

I’d like to take the opportunity to put all of this in a human context, which is often forgotten. We’ve got a situation in any dispute where you have a claimant who frequently has suffered some kind of loss – in our world that loss is often life changing – and an insurer who will have to compensate that victim to the appropriate amount and, looking at the broader picture, an insurer has to charge policy holders enough to pay on its claims. To sort that out we have a process that is painful and expensive for everyone involved. What I see this collaboration as opportunity for is 1) to help us triage and value claims as early as possible in their life to help our clients realise cost savings but also to help everyone in the dispute get to the right answer as early as possible.

It’s about making sure people with legitimate claims get what they’re entitled to. It’s about catching people who are trying to defraud the system and making sure people with a valid claim get paid the right amount. If we get it right, our clients will save money and everyone will have a better life.

This article first appeared in the February Legal IT Insider/Orange Rag newsletter – to see the issue or sign up for your free monthly copy click here: