No Such Thing as Too Much Data

Little wonder then that speciality insurers, like many others, cite management of data and extracting value from it as one of their biggest challenges and simultaneously a significant opportunity for underwriting success. They have never held more data than they do now which sounds like a good thing until they tell you (as many have told us) that they have never found their data harder to use.

Competition in the marketplace rewards firms that can move fast and underwrite with greater precision. The expectations of regulators are always intensifying. Brokers want faster service, boards want sharper portfolio intelligence and clients want more insight into their risk exposure.

All of these are pressures that could be addressed by faster, better use of data – and the data, of course, is there to be used. Policy records, bordereaux, claims files, exposure schedules et al are abundantly available, but they are fragmented across systems, and rarely joined up at the point where it matters.

This gap carries real cost. A recent McKinsey report estimates that 30-40% of underwriters’ time is spent on admin tasks such as rekeying data or manually executing analyses, time which could be used in steering the portfolio or other, more value-enhancing activities. The firms that close this gap fastest - reducing the manual effort in data-processing and more efficiently mining data for insights – will have a material advantage over those still adrift on the data lake.

At every stage of the underwriting process, data issues threaten commercial opportunity. But at each pain point technology offers solutions that, if appropriately implemented, have the potential to ease the burden on teams and add efficiency and value all along the chain.

Beginning at the point of submission and intake, expensive, time consuming review of unstructured and often disparate data inputs, leads to less available focus on risk judgement and commercial value. But advances in machine learning and natural language processing can reduce the reliance on manual assessment and create comparable, structured data for review and workbench data-entry.

OCR and AI data extraction can then be used to automate document checking, creating management dashboards that classify accuracy by broker or class, improving quality and speed of intake and reducing the prevalence of incomplete submissions and the manual effort needed to review.

Risk assessment has traditionally been one of the more time-consuming stages of the underwriting journey, involving aggregation across multiple systems of loss history data for catastrophe modelling. Where AI can shine is in its ability to quickly and automatically generate insights gleaned from multiple trusted sources, all available to underwriters on an integrated platform enabling faster, more consistent decision-making and pricing.

And finally management dashboards can surface auditable tracing of underwriting steps alongside KPIs, empowering prioritisation and data driven decisions – a far cry from the lack of measurable end-to-end time to price that has long decreased consistency and limited opportunity for optimisation.

The data faucet is not going to be turned off at any point soon and underwriters, along with the rest of us, are going to be dealing with greater volumes of more granular data every day. But just as the challenge of managing this magnitude grows, so does our ability to use technology to organise, parse and mine data for actionable insights. Being overwhelmed by the data tsunami is not an option. With the right tools, volume can deliver value and vital competitive advantage.