To assess big risks such as terrorism and oil spills and to calculate the price of insuring against them, underwriters receive information from brokers, usually in spreadsheet form. They then add other information, including their own claims data. This is combined with pricing information from actuaries that weights various aspects of the risks, which is also generally created in Excel. All this data combined is used to decide whether to offer insurance cover and at what price. There is a lot of manual work.

Just last week I met a senior executive offering aviation insurance who typically spends six days a month pouring over spreadsheets. Mostly not productive analysis but rather data prep, correcting errors, and trying to make up for missing data.

Talking with the CEO and founder of, Eric Giroux, his company's predictive analytics platform saves up to 30% of an underwriter's time spent on data preparation and reporting. Time then freed up to focus on performance improvement and growth.

According to hyperexponential, underwriters spend three hours a day on data entry which ties in with's finding.

In the specialty and commercial markets one of the major issues is the sheer volume of email submissions (enquiries) they receive. The challenge is how to find the business they want to write hidden in the (hay)stack of emails and documents sitting in their inbox. One lead underwriter said that her team had to leave 90% of emails unread. And she wasn’t sure that the 10% they did quote was the best business! There aren’t many industries that have to ignore 90% of business opportunities!

The challenge is more complex than that of course. Spreadsheets are one factor, not least in BDX. But to write better business there are other data sources to combine, normalise, and analyse. Claims data, telematics and sensors, flood and climate data for starters. 

Organisational barriers – such as insufficient access to data (54%), legacy systems (51%) and a lack of skilled talent (47%) – are preventing insurance firms from excelling at customer service, according to new research from information technology company Capgemini.

Its World property and casualty insurance report 2024, published on 17 April 2024 by in-house think tank The Capgemini Research Institute, identified these internal drivers as possible reasons why 27% of policyholder respondents looked to switch their insurer in the last two years.

Renewing capacity requires evidencing your underwriting discipline and insights to attract capacity providers. Distribution is only optimised by betting on the right brokers and market segments.

Ultimate loss ratios are an outcome of improved pricing, accurate reserving, and indemnity cost management.

“I think the underwriter’s job has got harder and harder over the last decade or so, because all we have done . . . is give you more information to consume, understand, process, but not given you any new tools to do it with,” said Nigel Walsh, head of insurance at Google Cloud, which provides analytics and product development tools. Data has been “stuck into spreadsheets and those spreadsheets got more complicated [with] more versions,” he added. “As those got bigger, and bigger, those things took a lot longer to run . . . and you never really knew if you were working on the latest set of data, or the latest version.” Spreadsheets can also struggle to cope with the vast reams of real-time data on insured assets such as oil tankers and airlines that is now available."

FT  April 7th 2024

Depending on spreadsheets is a great vulnerability but so is not leveraging the right data both internal and external. GenerativeAI is seen as a great contributor to tackling some of these challenges but will not take away from the need to understand your data and know what insights are required to achieve strategic and tactical goals.

Generative AI is trained on data and that data must be as relevant, current, and complete as possible. Otherwise, predictions are going to be flakey. That's not to mention the inherent vulnerabilities of deploying large language models and GenAI. ( See further reading).  

And you cannot just tackle this on a departmental, functional level. We have focussed on underwriting, pricing, and actuarial analyses above. To truly transform and focus less on admin and more on performance improvement and growth the enterprise-wide data must be strategically managed. An insurer typically has many technology stacks and data silos inherited through many mergers and acquisitions. Whether carriers with 25 or more such incompatible stacks, MGAs and brokers merged into global groups. The challenge is the same.

There are many gaps to bridge between Policy Admin Systems (PAS), Core technology platforms, and point software applications like rating engined and underwriters' workbenches,  claims, payments, and counter-fraud. Typically a data lake, or data warehouse project is meant to underpin all these whilst in reality data and relevant insights are still not readily available to business unit leaders having to compete in a rapidly changing world where old assumptions become outdated.

A recent survey by Gartner found that 47% of digital workers struggle to find the information or data needed to perform their jobs effectively. The survey also revealed that 66% of respondents agreed better business outcomes could be achieved if IT provided universally accepted and supported applications and devices to get work done. Without the right applications, digital workers struggle to find the information they need, make wrong decisions due to a lack of awareness, aren’t alerted with relevant notifications, and miss out on important updates.

Companies like Aiimi tackle that problem for clients like the FCA, Liberty Mutual, Anglian Water, GigaClear, and other blue-chip enterprises. With a mixture of data scientists and engineers plus a data management platform powered by AI that will ingest all these incompatible data sources and make them available for point applications, core platforms, PAS, and specialist platforms like: -

  • ( MGA underwriting, actuarial and pricing),
  • hyperexponetial and Cytora ( broker and carrier underwriting, actuarial and pricing) 
  • CGI underwriter's workbench and rating technology
  • Point solutions like RightIndem (Claims) , Friss/Shift (Fraud), Imburse, Shift, Mastercard (Payments) and so on.

Companies like Aiimi underpin data lakes, platforms, and point solutions. They also have the skills and resources to operationalise AI: - conversational, extractive, generative, and natural language processing and machine learning. 

From putting enterprise knowledge into the hands and screens of employees to embracing the flood of generative AI products and services they underpin the transformation of the whole enterprise and not just underwriting, pricing, claims, distribution, and product development.

Freeing up people to focus more on performance improvement and probably doubling the amount of time available to plan and achieve growth.

Data was always at the centre of insurance from the early days in London's coffee houses to today. There is no easy path or shortcut to surface and turn data into actionable insights that deliver competitive advantage and customer satisfaction. There are many tools and platforms available to achieve that outcome that are backed by the advice and practical guidance to operationalise AI, data and insights. The beguiling attraction of GenAI should not camouflage the need to make data ‘GenAI ready’ and operationalise AI. 

A good place to start is with the companies mentioned above 

Further Reading

Saints & Sinners - proving effective AI use cases in insurance

Travel insurers, pricing, and profitability

AI, Healthcare, speed of adoption and lessons for insurers

Enterprise ChatGPT, LLMs, and choosing the right use cases.