How insurance data science is like gelato

A few quarters ago I re-joined a conversation that I started with a respected Chief Actuary over a dozen years ago. We frequently connect on trends in the industry.

Now this is no ordinary actuary with just a focus on reserving and rate filing, but a clever, highly educated, and practiced mathematician who made a career change to become an actuary and who is currently a leader in the industry. Having vestments in actuarial and analytics has appeal.

As I said, the original dialogue began a dozen years ago, as multivariate and general linear modelling methods were becoming table stakes for granular sophistication around risk-based pricing predictive analytics. Being smarter about risk is a competitive advantage.

Unlike a lot of conversations, this one started with grieving.

The actuarial profession was suffering from a skills gap. Predictive analytics skills were becoming necessary, and a new type of worker, a data scientist, was entering the gap to become valuable to the business of insurance. Many executives in those days proclaimed data scientists as the future of insurance, and these new outsiders were taking leadership roles, sometimes with actuaries reporting to them.

It was dually bewildering and depressing - and cause for career concern.

The subject of our conversations revolves around what is different about the practice of analytics between and among actuaries and data scientists as applied to insurance and risk.

They framed the discussion like this -

.....“If you cannot describe the difference between a taste of ice cream and a taste of gelato, it is obvious that you have never experienced tasting gelato.”

Having come from meagre beginnings, and not being normally susceptible to the lure of desserts, I in fact, had never experienced gelato – so I ran towards the opportunity to learn by metaphor as we ordered the last course of our meal at the Casualty Actuarial Society meeting. One bite changed my palate and perspective forever with a spoonful of handcrafted, artisanal pistachio heaven.

The texture, substance, and flavors bombarded my senses of taste, touch, sight, and smell (even sound as I murmured my delight). Even still today – if memory is another sense, that epicurious event remains imprinted as a moment of epiphany that still rings in my recurring cognition.

The entirety of the whole experience they described as data science. Data science is an aspiration of wonder dolloped as consumable understanding of complexity and made as simple as placing a scoop in a dish. Actionable information that can be easily served and consumed is data science at its best.

Practicing Fellow and Associate Actuaries with decades of tenure never had the academic experience and exposure to data intensive, high computation, multivariate algorithms which are becoming a competitive advantage for companies that invest in these capabilities. Rate filings and reserving can still be accomplished with a word processor and a spreadsheet, but sophisticated pricing and scoring systems require more, as do advanced marketing, underwriting, distribution, billing, call center, and claims systems.

Actuarial methods traditionally have had a narrow application to the confidence in a financial forecast that serves to maintain the solvency of the company in meeting its future obligations. This is served inside a restraint-laced regulatory framework with oftentimes slow and deliberate change cadence. Data science is not that.

Our conversations have evolved in recent years as the skills gap chasm grows even wider.

Data, AI, and cloud have grown more important with the emergence of new data, new analytics, and new computation capabilities vastly outside of the traditional actuarial remit and even beyond the IT department capabilities of many of the largest and most sophisticated companies.

The list of impactful technologies is extensive and growing. Here is an active list: computer vision, natural language processing, speech analytics, text analytics, geospatial frameworks, machine learning, telematics, end-to-end experiences, IoT, digital distribution, AI, household and living situation models, customer behavior scoring, knowledge graphs, social network information, etc.

Productizing and monetizing insights into actions across the enterprise are the dividing lines between data science and traditional IT and actuarial organizations.

Organizing for success is taking a wandering path. Companies are learning how to learn, how to ask better questions, and how to source the mix of build and buy levers to use in different AI and data science strategies and deployments. Everyone is finding new appetite for more data, more compute, more impact, and more recursive segmentation and feedback loops for more accuracy with faster cycle times and smaller expenses.

Inevitably we resolve that these career paths are separable with distinct changes in how and where knowledge, skills, and abilities are applied. Actuarial - slower, regulated, embedded. Data science - faster, nimble, scalable.

Both are on the dessert menu, but best not served together. Essential and regulated tasks are not often the shiny, new, and sexy projects requiring bleeding edge skills, powerhouse computation, and new sources of exotic and even streaming data.

Grieving is growing again. And this time the actuarial syllabus cannot maintain pace.

Maintaining curiosity, learning how to learn, and how to best improve customer experiences and business solutions using data science is part of creating a data driven culture. As my recent Celent survey report on “Analytics Strategy Execution” concluded, culture change is a key barrier to success for most companies, and it is the key reason for the emergence of new analytics executive roles in the industry.

Like the CIO took a Board seat two decades ago, Chief Data and Analytics Officer roles are emerging across the industry on an enterprise-wide basis. Their remit is to influence customer insight, digitization, pricing, marketing, underwriting, claims, end-to-end experiences for customers and employees, compliance, and other links in the insurance value chain.

Nine months into my Celent tenure, I am having this same conversation everywhere with CIOs, CDOs, CDAOs, CEOs, CFOs, CUOs, CMOs, Claims Executives, and Chief Actuaries alike. Ping me if you are curious about the survey results or how to think differently about your data and analytics strategy at mellingsworth@celent.com.

This blog entry has been reprinted from Celent.

For reprint and licensing requests for this article, click here.
Data science
MORE FROM DIGITAL INSURANCE