Using AI to Solve the Biggest Challenges in Underwriting Data Cleansing - Pro Global

Using AI to Solve the Biggest Challenges in Underwriting Data Cleansing

Augmented Intelligence (AI) has the potential to transform the insurance experience from often frustrating and bureaucratic processes to something fast, on-demand and affordable.


November 1, 2021

Share this page

Augmented Intelligence (AI) has the potential to transform the insurance experience from often frustrating and bureaucratic processes to something fast, on-demand and affordable. Data Cleansing, in particular, is an area which is in need of innovation and simple solutions – spending unnecessary time and resource fitting data into another system is never ideal, particularly in areas such as exposure and catastrophe risk management where underwriters need to be able to make informed decisions based on quick, readily available and reliable data.

The consequences of not managing data cleansing effectively can be significant. At Pro, we have developed a solution – Cleansing Augmented Intelligence (CAI) – that can address these issues within one platform. CAI, which has been developed with specialist AI expertise, data cleansing knowledge and software houses including Analycat, is a unique solution to support and optimise data cleansing in terms of speed, accuracy, consolidation and data quality.

Making more efficient use of time
Our industry wastes a lot of time cleansing and re-cleansing the same data and this is particularly true within the world of exposure and catastrophe management. Huge numbers of SoV files can be cleansed one year but when they are re-submitted for renewal, in many instances they are re-cleansed from scratch even if they contain only minor changes. CAI recognises whole files and data types within files and reapplies the cleansing process – which can reduce the effort required to process a file by around two-thirds compared with manual processes.

Changing catastrophe model vendors is also an enormous operational undertaking – even when undertaken efficiently, moving a full suite of models for multiple perils and regions is likely to take at least 18 months. CAI can help alleviate the burden of changing models by providing multiple model formats from one cleanse. It will also smooth out the business-as-usual operational burden of running multiple models for those insurers that choose to do so.

Issues such as the re-cleansing of renewal data are very much magnified in an eco-system like the London Market. Indeed, the very nature of London Underwriting and its subscription methodology presents an additional inefficiency in itself with multiple parties all cleansing identical pieces of data. On a large commercial policy, for example, when the placing broker is included, a policy can have 40 participations. CAI is a system which can maximise the benefits of London’s combined data without compromising the individual underwriting interpretations which make the market so effective.

A big problem with manual data cleansing stems from subjectivity and the requirement for operators to make contextual assumptions during the cleansing process. With CAI, the assumption made to the original value is saved and stored for all schedules, including renewal variants. Therefore, the data remains consistent across an insurer’s portfolio and will not cause unwarranted fluctuations.

Intuitive Mapping
Geocoding is a central tenet of catastrophe modelling and exposure management. Without a thorough understanding of where exactly an insurer’s exposures are, it is impossible to accurately model and monitor risks. CAI’s intuitive mapping interface allows data cleansers to verify the original address information – with any amendments saved in the system and with no requirement to repeat the process the following year.

A challenge, particularly in the property D&F market is post-bind modelling. Commercial pressures and model bottlenecks such as data cleansing require that many policies are only modelled once the insurer is already on risk. With CAI, policies are cleansed on average 50% quicker than manual excel methods – allowing insurers’ to get more of their portfolio through on a pre-bind basis and ensure that the premiums charged are commensurate with the risk undertaken.

The reliability of exposure and catastrophe modelling output is only as good as the quality of an insurer’s data. Poor data quality and reliability can affect an insurer’s bottom line and tasks such as cleansing and validating data can drain valuable resources. That is where Pro’s Exposure Management Services team can step in and help either through a SaaS self-service offering or through the CAI managed service platform where we do the heavy lifting – allowing companies to focus their resources on where they bring most value, including in particular their underwriting discipline and performance.

MGAs: A Global Growth Market Looks Ahead

AM Best reports that premiums written by Managing General Agents (MGAs) stood at US$60 billion in 2021, up from US$51 billion in 2020 – record […]