Data Data Data
This is a summary of my keynote at Ivans Insuretech Boston 2025.
In my opening keynote I reflected on data in insurance from the vantage point of a strategist, a business perspective if you will.
And I talked about why data lead insurance is still something we need to make happen.
I know that you are immediately thinking "sheesh, why doesn't this guy know that insurance already has massive data scale."
Well I get that, but being data heavy doesn't mean we are data rich. And the adage I've heard repeatedly in insurance of "data is the new oil" in previous decades, and in many ways this is true, only we often don't extract it, refine it and use it operationally.
And this is something I believe has lead to somewhat of a stall zone, especially in the face of an AI onslaught, in insurance and we therefore need data lead transformations to continue.
DATA SCALE
The truth is that data is imperative and vital to insurance, it's the lifeblood of the actuarial muscle that makes it all possible.
For example, calculating weather risk, a typical catastrophe model might attempt to use 150 years’ worth of meteorological data. That represents ~ 30 terabytes of data, over 100M locations, while simulating approximately 100,000 atmospheric and hydraulic scenarios.
This in turn yields 200 billion records to feed the financial models.
These types of large scale, elastic compute, on demand scenarios are where the cloud excels. But also an example of where we dominantly think about data in insurance.
Collate, analyse, extract insight, build or update the model and then apply the outcomes mostly through underwriting and pricing.
Vital but relatively narrow. And massively constrained by the ability to truly shape propositions.
DATA VELOCITY IS BECOMING MORE VITAL
"90% of the world’s data has been generated and gathered in the last two years."
And that's because the world is recognising us more as individuals and shaping life experiences to suit specific interests and activities.
This is set to accelerate massively. Especially with AI.
Big Data has the ability to change the way we see people, which impacts on insurance and risk pricing.
It should also change how we operationalise and use data to create better relationships, to build risk mitigation, to orchestrate claims ecosystems and more.
Recommended by LinkedIn
But there's a problem.
Legacy and modern legacy technology.
And it has created a fracture. With data for analytics being separate to operating data models that treat data as a perishable asset, constantly mine it for insight and act on that insight. Increasingly close to real-time.
THE NEW DATA OPERATING MODEL
As you lot will know... Nearly 80 percent of enterprise data is unstructured, coming in the form of emails, text documents, research, legal reports, voice recordings, videos, social media posts and more. For insurers looking for answers, this unstructured data is a goldmine. However, compared to structured data, it’s much more difficult to analyze.
Fortunately, evolving technologies, such as natural language processing, can enable insurers to unlock the value.
Natural language processing, a component of artificial intelligence (AI) that can understand human language as it is spoken, has evolved to a point where it can be used to understand a user’s questions (text or speech) and mine insights from vast amounts of unstructured data. Through training models.
This can now also be added to vast amounts of connected and real-time data. Like your car telematics. Your connected home. Your commercial building's "Building Management System" and so on. Outside of risk, pricing etc. insurers need to use “customer data” so shape better experiences and services. Hélène Stanway Iain Wilcox and Daren Rudd 👓 now this only too well.
However, the tech stack and operating model needs to transform to make this and any meaningful use of AI a reality.
This new DNA for insurance makes insurers look more like eCommerce businesses in the way they operate and in the way they conceive and build product & experiences.
THIS DATA MODEL AS AN ECOSYSTEM
Operationalising data needs us to think differently.
Insurers can use large volumes of data to improve pricing strategies, streamline the claims process, and make better underwriting decisions. And yet they typically struggle to make any changes to product, services and experiences beyond this.
And make no mistake insurance is a data product and a data-lead service. We just keep pretending we still produce documents we call policies, only that concept is really defunct.
Instead what all of us really buy is an adaptive personalized cover typically adjusted somewhat to our risk tolerances and bank balances. We just make this super hard to achieve because we don’t fundamentally address the underlying way we operate, and in turn how this allows us to operate around data.
Data is a big part of insurance today, but it is an even bigger part of insurance tomorrow.
Superb keynote Rory. This is an area where I get so many mixed messages. In public, I hear awesome success stories pushing the boundaries of AI on top of ever richer real time data. In private, I hear not much has changed, with struggles to access and use data. Also hearing more about data ROT - risking making it more difficult to extract insight and action from our data swamp. You’re right in all you say. And in counter fraud we retain an insatiable appetite to use data of all types, in so many ways. The challenge remains ensuring appropriate architecture and orchestration to: Capture, store and make the data available Select and enable the weapon of choice Support the plumbing in getting the outcome to the production environment (be this user or intervention in a customer journey). Love to hear your thoughts on where we are genuinely succeeding and the extent of vapourware.
Thank you. Spent my morning in conversation on this topic. Good to read in retrospect your observations.
Rory Yates insights- 1) "90% of the world’s data has been generated and gathered in the last two years." 2) "80% of an insurer's data is unstructured and hard to analyse.2 3) Insurers are used to analysing data around a policy and not a customer Rory argues that a new data operating model is vital with five key action points.
Need to compare notes - I’m covering some similar topics (including being Data Fluid!) at ITC tomorrow!