Insurers are increasingly using vast amounts of information - ranging from crime patterns to historical data - in the process of analysis and underwriting. Kathryn McCarthy reports on the new trend.

The personal lines marketplace has become increasingly competitive over the past decade, with insurers relying on ever more sophisticated information in the underwriting process.

Today some insurers use in excess of 30 individual risk characteristics, compared to little more than a handful just a few years ago. The desire to rate products in this way means more detailed information has become available to insurers for underwriting, analysis and targeting.

"The extent of information available at postcode level, for example, is almost endless," says Jack Brownhill, motor underwriting director for personal insurances at Groupama. "Demographic and socio-economic information is boundless, as is data relating to exposure to natural weather events, geological structure, and so on." He comments that through government offices it is now possible to obtain extremely detailed information regarding crime patterns for both locations and vehicles. "Many insurers now build this information into their pricing and risk selection processes."

Brownhill says that by using increasingly sophisticated modelling tools insurers can now use large volumes of historical data to determine likely future patterns of activity. "They can do this, for example, in relation to previous accidents, losses or convictions for motor clients, and theft and subsidence patterns for household."

He says that gender and marital status are seen by a number of insurers as important rating factors, although the past insurance history of an individual or family continues to be one of the most useful barometers for rating. "With the existence of the claims and underwriting exchange (CUE) the ability to verify previous experience and to control fraud have grown in importance."

In a crowded market with some very sophisticated sources of information for underwriting purposes, insurers use these to rate more precisely. Insurers chasing lucrative affinity business deals and the trend for customer segment targeting influence both the level of premium paid and the extent of cover for different group types. As a result, information providers have a constant demand for their services.

Digging deep
One such information provider is Experian, a subsidiary of Great Universal Stores. Experian recently acquired CUE and now provides this information alongside its other services. Mike Prentice, sales director of the insurance services division, says the company supplies several types of information to insurers: "First there is vehicle information. When a registration number is put in we feed back the car specification and Association of British Insurers (ABI) rating code, cutting down on time and providing more accurate details for underwriting.

"Next we provide claims data through the CUE database. This simplifies and improves the proposal process, with a good record of claims history for verification.

"Then we provide the publicly available consumer information. This doesn't include private consumer credit information, but does include the electoral roll and records of County Court Judgments (CCJs) and bankruptcies.

"Finally we have a number of products, including our geo-demographic tools, which profile postcodes and individuals. We also have a perils product set that looks at the likelihood of certain risks at postcode level, including flood, theft, wind damage and subsidence."

A huge amount of information is available to insurers from external sources to arm underwriters with better knowledge. But the trend to gather and use more customer information lengthens proposal forms, and this is judged by some to be bad news for the consumer, who generally has a short attention span and little patience when it comes to buying insurance.

This view is held by the tele- and internet-broking fraternities that strive to reduce the number of questions asked at proposal to be more customer friendly and reduce transaction time.

Internet broker Quote & Buy provides online quotes from a range of participating insurers. Richard Jessel, partner, believes that while transactions would benefit from fewer questions, insurers show no signs of letting this happen. He says: "We provide quotes from a host of insurance companies. For each product we have a generic on-screen application form. Because all insurers have their own set of underwriting questions, we end up asking more, not fewer questions."

Jessel explains that to get around this problem a ball-park figure based on a few questions is quoted before the customer proceeds. "We only take them through all the questions if they like the initial price. Insurance companies are fantastically proprietary about everything, especially their policy information. They believe they are the best at what they are doing. So they all want their questions asked." Jessel explains that Quote & Buy's solution to the number of questions is its proposal form "wizard" which helps the customer through the form quickly, and many fields are filled by look-up tables. "This saves time, and the perception is that we are asking fewer questions than there really are. The key to it is to get to the ball-park figure quickly so you are not wasting anyone's time."

Speeding it up
Reducing transaction time by any means is important to many selling personal lines insurance. There is a growing sector of the market that sees the use of profile information as a way of reducing the numbers of questions asked. By profiling types of individuals a picture can be built up of their typical exposure to risk. Geo-demographics have been used as a marketing tool for years, but only recently as an underwriting information source. But in the future, when companies emphasise their consumer focus, and personal lines distribution shifts to big brand retail outlets and the internet, the likelihood of using profile data to underwrite insurance will increase.

True to prediction, the trend over the past ten years has been a move to use more information in insurance rating, says David Campbell, partner at top five accounting firm Pricewaterhousecoopers. "In the future, those selling personal lines will increasingly be strong retail brands. But these third parties will not want to gather a lot of customer information at the point of sale." Campbell explains the desire for retailers to reduce the transaction time so insurance can be sold over the counter. "With this mindset they are looking for different ways to reduce the amount of information required from the customer on the spot. Experience on the internet is that lots of questions put people off, so information sources are used to fill in a lot of the information."

Campbell believes the pressure to cut the transaction time and the bundling of insurance with other products will eventually lead to insurers using less data. "This type of less refined rating is driven by the changes in distribution and the need to sell quickly and cleanly. There is a sense in which insurance is seen as arcane by consumers. Some retailers have taken this and turned it around to gain a competitive advantage. The whole transaction becomes customer-driven, and insurance is the subsidiary purchase."

Most insurers are resisting the move to ask fewer proposal form questions. In the information age knowledge is power, and if insurers have to purchase profile information to understand their customers, they are in a weaker position.

Indeed, not all insurers rely heavily on external data in the underwriting process. The larger the company is, the more proprietary information it has at its disposal. The merger of Norwich Union (NU) and CGU is a perfect example, with 23% household market share. "The integration of CGU's data is almost complete, and we are in a good position to get prices and products right,"claims Nick Pierson, head of household marketing at NU. "We don't use personal data from outside sources. When we are setting up prices for postcodes we review internally our claims data and experience."

Pierson explains NU uses external sources such as geology maps and lifestyle data and combines them with its own data. "I don't think it is necessarily that good to buy information at individual level, as this goes against the concept of the insurance pool. Clearly some individuals are cross-subsidising others, which is how it should work. But customers don't want this, as price is the main selling point of personal lines. There is a balance to be had between becoming too granular with data and spreading the risk."

Picking cherries
For the future Pierson believes factors such as internet technology will allow some insurers to underwrite at individual level. "Some players will cherry pick using technology, and will become more granular and go down to individual level. The internet could accelerate individuals going online and creating their own products. Perhaps some insurers will see a market for doing this. We have no immediate plans to go down this route."

Some underwriters may well go down the minimalist underwriting route for their big affinity retail clients, but mainstream insurers realise the value of having their own customer data and will continue to resist moves to change their processes.

Axa Insurance uses its own proprietary data but combines this with information from other sources. Philip Bird, direct and corporate household product manager explains why the major source of information comes from its own policy data. "For any five year period we have two to three million exposures we can look at to analyse risk. And we have got much more sophisticated in how we look at our own data."

Bird observes that geo-demographic information classifies on a socio-economic basis and can be used as a rating factor, but Axa uses this type of system to profile its account. "Third party data sets have a wider spread for geophysical analysis, such as analysing subsidence and flood risks. These risks are relatively less common and we won't necessarily have a complete enough picture from our in-house data, so we use these data sets. We mainly use information from providers who take their information from insurance industry sources."

At present, historic proprietary data is commonly used for sophisticated rating of future risk exposure by underwriters. But for insurers, insurance is a gamble and past history can only give an indication of what the future will hold.

In the future, a proportion of the market may well choose to limit the amount of data captured at the point of sale, and those insurers will turn to external data sources to build up a more detailed understanding of their customers. But in this process insurers will have to bear in mind issues such as data protection and discrimination.

The personal lines marketplace is in flux, and with changing distribution trends and increasing technology there will be many different philosophies when it comes to the use of external information sources. But it does seem that sophisticated pricing and risk selection is here to stay and insurers will continue to find new ways of selecting the customers they are looking for, and pricing their policies accordingly.

What information sources can be used for
Postcode & House Number
Obtains correct address and validates from Postcode Address File (PAF)

Name & Surname
Checks made against electoral register and publicly available info such as CCJs and bankruptcies

Vehicle Registration number
Cardata check obtains car specification,

ABI code and insurance group.
HPI check searches database of 57 million UK registered vehicles

Claims History
Obtains details held on the CUE database, which holds 85% of all household and 70% of all motor claims over the past five years

Payment Details
Credit card verification
Bank sort code and account number validation

Rebuilding Value
Obtains rebuilding costs for all types of property, UK-wide

Flood Data
Information on coastal and fluvial flood risks

Subsidence Data
Information on past and future likelihood of subsidence risks

Wind Data
Met Office service records wind speeds to within 100 metres of every UK property, to validate storm claims

Scoring systems
Provide a score on the consumer.
For example: Proposer not on the electoral register + a bankruptcy record = low score = not a good risk

Consumer Classification Systems
Places consumers in identifiable market segments, at person, household and postcode level