Expertise In Action - Big data: Thinking big


‘Big data’ has long been hailed as the next big thing in insurance – so why have firms been so slow to harness its undoubted potential?

The commercial world has developed an obsession with data, and the general insurance industry is no exception. Rather like the song from Grease, firms constantly plead ‘tell me more, tell me more’ – but they aren’t always sure what to do when they get more information. This is just the latest of many challenges the industry has faced in its struggle to unlock the potential of data – or ‘big data’, as it is now fashionably labelled.

zurich-expertiseThere are two main drivers behind the insurance industry’s desire to unlock the potential of data. The first – and so far the most significant – is the fight against fraud, while the second is the potential for better risk profiling and all the implications this has for a more sophisticated relationship with customers.

Increasingly, these two drivers are interwoven, and no more so than with the Insurance Industry Access to Driver Data project that will see My Licence launched in June. It also highlights some of the barriers that still lie in the path of better data usage.

This project is a collaboration between the Driver and Vehicle Licensing Agency, the Motor Insurers’ Bureau, the Association of British Insurers and the Department for Transport. It is intended to give insurers access to accurate data from the DVLA, including the type of licence held, convictions and penalty points. Brokers and insurers will be able to access this information through an MIB‑hosted data hub when policyholders provide their driving licence number during insurance applications.


My Licence – data provided by DVLA

Conviction codes 
Penalty points 
Conviction dates 
Type of licence held 
Length of time licence held 
Entitlement to drive (i.e. manual or automatic, motorbike, HGV and so on)


The industry will see several benefits from this, not least a reduction in application fraud and less potential for disputes when errors are uncovered at the claims stage. The guarantee of accurate driver information can also benefit customers, says Jeremy Haynes, insurance analytics partner at Deloitte: “My Licence has some significant customer benefits – consumers won’t have to provide so much information and the information that is used will be accurate and up to date. Some people don’t realise when their penalty points have expired, for instance.”

It has its limitations, however – for example, not all of the market will be using it to start with. “This initiative is a positive step forward in providing fairer premiums, while tackling fraudulent behaviour,” said Ageas recently in its Talkshop blog. “Companies that do not adopt this initiative early on will not have access to the data and will be at a clear disadvantage. We are working with the software houses and aggregators to agree go‑live dates that, disappointingly, currently range from June 2014 to 2015.”

If some insurers remain outside the scheme longer term, aggregators will have to continue asking all the current questions as well as requesting the DLN, blunting some of the consumer benefits.

Data protection concerns
The other drawback of the scheme relates to concerns over data protection legislation, especially among government departments. Paul Baxter, head of Chaucer Direct, feels the DVLA is being too cautious and not making the most of the two‑way flow of information that will see it getting up‑to‑date address and insurance details.

Graeme Trudgill, head of corporate affairs at the British Insurance Brokers’ Association, believes this excessive caution is a common problem in the industry’s data‑sharing projects with government departments: “It is very, very disappointing that the government won’t co‑operate. We are sharing data for My Licence and other projects so why can’t they share more with us?”

Trudgill adds that a similar problem is holding back the development of the employers’ liability insurance database for tracing mesothelioma claims: “They won’t give us access to the employers’ reference number database and other Companies House information. The barrier is that the civil service lawyers cite the Data Protection Act and then all the usual different excuses come out.”

Data‑protection issues aside, the next step for the IIADD project is a no‑claims discount database expected to be up and running by the end of the year, although Lexis Nexis has been in the market with a similar product since last September.

Both these initiatives, like much of the work done by the Claims and Underwriting Exchange, is about using existing data more effectively – something the insurance industry could still do more of, according to commentators.

“It is sometimes easy to go chasing the bright new shiny things and forget to look at what you might have under your own bonnet already. I have no doubt there are some hidden gems,” said Haynes. “I am sure there is a vast amount of potentially valuable data that needs to be got off paper and into a database. It might not be immediately obvious what to do with it, but there is almost certainly value in it.”

Too often insurers keep their data in silos that don’t communicate with each other, says Richard Holling, insurance sales director at Target Group: “Claims systems are just not linked to quote systems, for instance. Insurers haven’t got a 360‑degree view of the data they already hold.”


What is big data?

Big data is not just about the size of databases, although most big data is so large and complex that it becomes difficult to work with using most relational database management systems.

It also refers to “things one can do at a large scale that cannot be done at a smaller one to extract new insights or create new forms of value, in ways that change markets, organisations, the relationship between citizens and governments, and more”, according to Victor Mayer‑Schönberger and Kenneth Cukier’s bestselling book on the topic Big Data.

A key characteristic is the lack of immediate connection between different databases: “Most strikingly, society will need to shed some of its obsessions for causality in exchange for simple correlations: not knowing why but only what.”


Optimism over abilities
George Marcotte, managing director of financial services analytics for UK and Ireland at Accenture, is more optimistic about the ability of insurers to identify and utilise existing data: “Their history has been to exploit their internal databases. They are maturing in their understanding of the old data they have and are looking for value.”

Marcotte adds that insurers are using text mining and video tagging to find the added value in their old data to help refine their understanding of the risks they are underwriting, and to price them more accurately. Sharing key elements of this data with brokers is also key in improving risk profiling. He notes: “Insurers understand the profitability of their portfolios. They need to share more of that intelligence with brokers to help them attract more of the right risks.”

Biba corporate affairs manager Andy Thornley takes this viewpoint a step further, insisting that data sharing between brokers and insurers could also help improve customer service: “In the future, with brokers and insurers being able to pull in data from different sources and share it effectively at the point of quote, it will become a better experience for customers.”

More efficient data sharing between brokers and insurers has also levelled the playing field between brokers and those in the direct channel, says an Ageas spokesman: “There are clear fraud benefits for brokers, for example through the use of Syndicated Intelligence for Risk Avoidance and other fraud tools. Brokers can now expect better pricing, whereas before it could be argued direct writers had an advantage due to greater pricing sophistication – through credit scoring, for example.”

The fact that computer systems are now more advanced means insurers face fewer challenges in sharing and utilising existing data. Haynes explains: “The tools that exist combine a lot of diverse data. You can share it through online portals and you don’t need the monster data warehouses of old now that we have the cloud.”

Where legacy IT becomes an issue is with point‑of‑sale data collection, according to Holling: “If insurers want to rate on a new factor, they have to have the ability to collect it, store it in their system and access it. This isn’t always easy.”

Looking to the future, there is vast potential in collecting new types of data that may not appear immediately linked, says Holling: “It could be something as simple as identifying someone as a dog owner. On the face of it there may not be an obvious connection between a dog owner and a safe driver, but there might be a correlation. People are often more careful with their pets than their children, so they might drive more carefully with pets in the car.”

This would take the insurance industry down the road to big data, where it could learn a few lessons from the likes of Wal‑Mart in the United States. Wal‑Mart reviewed years’ worth of till receipts and then matched that data against a wide range of other databases. It noticed some surprising correlations between hurricane warnings and the types of food people purchased. It didn’t matter that they couldn’t explain why that happened – the exercise had extracted some vital sales intelligence.

In the US, Aviva has been studying the idea of using credit reports and consumer marketing data as proxies for analysis of blood and urine samples, with the aim of identifying customers who may be at a higher risk of illnesses like high blood pressure, diabetes or depression. This method uses lifestyle data – including hundreds of variables such as hobbies, the websites people visit and the amount of television they watch – in order to eliminate the need for diagnostic tests for health insurance applicants. This reduces the cost from $125 (£75) per applicant to $5, and saves having to put people through the tests.

UK insurers are watching developments in the use of big data on the other side of the Atlantic with interest. “The industry is starting to embrace big data but recognises there is more to be done to maximise its potential. There is a lot to learn from some of the US models where they are very advanced in using data for insurance pricing, including the use of data on social media behaviour,” says an Ageas spokesperson.

Telematics, social media and other lifestyle data is already being used by insurers in the UK, explains Marcotte: “People are more self‑directed than advised nowadays. In fact, 80% are willing to share their information to get more personalised products.”

Through the eyes of an insurer, this means more refined risk profiling. Marcotte continues: “Take social media. People are more expansive about their experiences online and form online digital communities. Insurers can use that to create a better risk profile, and then create an interactive dialogue with those communities that can draw out additional data.”

Fighting fraud
The fight against fraud has also started to engage with social media. “The Insurance Fraud Bureau is now very sophisticated in taking in data from local authorities and social media to help identify fraud networks,” says Thornley.

Telematics is another big data opportunity for the insurance industry. It can have huge benefits for road safety and more sensitive pricing – but, on the flipside, it creates new challenges on the data‑sharing front, warns Trudgill. He says: “Telematics has lots of potential but it is still in its VHS v Betamax era. Insurers are using it in different ways and collecting different data – so who is to say where it is going?”

As well as creating a clearer consensus on what data should be collected via telematics, Jeremy Haynes says there is a debate about who owns it: “Who does the data belong to – the car owner, the device owner, the insurers? This needs to be clarified. I think insurers are still getting to grips with what to do with this data.”

As an insurer, Baxter has no doubt about whose data it is: “My view is that 100% of the data is owned by the customer. It doesn’t belong to insurers. The issue is about creating common standards so it can be transferred and shared.”

So where will all this data take the insurance industry? To better products, according to Trudgill: “The industry is getting so much rich data – the genie is really out of the bottle now. One result is that we will see more products designed for niche markets.”

However, Holling warns that there is a darker side to the issue: “You have to be careful you don’t end up with a social exclusion problem. There are people who don’t want to share all their information online and others who have a fragmented online profile because they move around – maybe in rented accommodation or in employment because they are freelancers or contractors. Just because you don’t have data on them doesn’t mean they are poor risks.

“As insurers get more refined in their risk assessment, you might end up with people who are viewed as a poor risk who can’t get insurance or only get it at prohibitive rates.”

Video: Data validation in the broker space

expertise-forum-data-validaThe insurance industry has invested heavily in its capabilities to detect and stop fraud in recent years. And this is as true at the front end with data validation as it is with claims fraud at the back.

However, there is added complexity when the business is brokered; an intermediary is the first point of contact for a prospective customer as the validation could act as a potentially frictional layer of information gathering between the underwriter and broker.

As part of the ongoing Expertise in Action campaign Post editor-in-chief Jonathan Swift recently sat down with Zurich’s underwriting fraud manager, UK personal lines, Alanda Reynolds; and head of personal lines, broker, Ian McManus to discuss how these validation processes can run smoothly and whether any lessons be learnt from the direct market.

The trio also talked about the value of education and issues of using third-party information sources.

Watch the video 

  • LinkedIn  
  • Save this article
  • Print this page  

You need to sign in to use this feature. If you don’t have an Insurance Post account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an indvidual account here: