The growth of algorithmic underwriting could trigger structural overhauls of underwriting teams, open routes for specialisms, create entirely new senior positions and widen career paths for the next generation of underwriters, experts have told Insurance Insider.
Amid the proliferation of algorithmic underwriting, carriers using this technology explained to this publication how the dynamic of underwriting teams could change, the new skillsets that will be required, and how various positions across the pecking order will evolve.
At the senior level, it could see the creation of roles resembling chief product officers, mirroring a title seen more in consumer-facing technology companies but with a very different meaning in insurance, as they’ll be harnessing and synergising the skillsets of new-look underwriting teams.
Executives said this emerging role, which could function as the interface between technicians and underwriters, will need the technical knowledge, operational leadership capabilities, and communications skills to speak the language of both groups.
In an operating environment in which algorithms are automating low-value tasks and freeing up time and resource, they are in turn prompting a rethink and re-engineering of underwriting processes from the ground up. This has implications for underwriting assistants, but by no means did sources think this role will be rendered redundant.
Moreover, technology could eliminate any residual perception of underwriting assistants as data-entry dogsbodies, instead shifting to a situation that treats and trains them as data analysts, giving them pathways to pursue roles where they can feel fully engaged.
These developments could lead to carriers – at least those using algorithmic solutions – revamping the career trajectories they set out for new talent.
With this technology, there may also be fewer underwriters relative to premiums written in each carrier, compared with today, which should translate to expense advantages and faster response times for brokers.
And with less time required to assess individual risks, tracking underwriters’ productivity could also look different, as time will be freed up to engage more with brokers and track the output of those meetings.
Ultimately, the flipside of a more digitalised syndicated insurance market will see insurers and brokers start to harvest a lot more data, meaning that with a mass upskilling or hiring spree likely, data technicians will be in high demand.
Upskilling and hiring the new generation
For any carrier with the infrastructure to employ algorithmic underwriting effectively (more on that below), it may still have to resolve a skillset issue. This might become a market-wide challenge if, in 10 years, the majority are using an algorithmic solution.
Most executives this publication spoke to agreed there will be a need for more data scientists, data engineers, machine-learning operations specialists and, as an adjacent but related capability, individuals with technical knowledge in training generative AI tools.
James Slaughter, group CUO at Apollo – which launched a smart-follow partnership with algorithmic underwriting insurtech Artificial last year – acknowledged that some of these skillsets may be in short supply within the London market.
“[But] you don't have to go to the insurance market for data scientists and engineers; there are people who can bring those skills to the industry”, he said, adding: “It can also be undertaken via partnerships. Because of our Artificial partnership, we don't have to find the data scientists and the data engineers.”
At Ki, the first fully digital and algorithmically driven Lloyd's syndicate that writes follow-only business, CEO Mark Allan said the firm has hired some technology roles from the insurance sector, but as Ki is a blend of an underwriting business, a syndicate and a technology business, the vast majority of its technology team are new to insurance.
“So there is quite a sizeable job-creation opportunity here. That skillset around software engineering, products, data science, machine learning and generative AI just doesn’t exist in our market, but these aspects are core to the use cases for this technology," he said.
Gilbert Harrap, CEO and cofounder at InsurX, the digital capacity exchange that helps carriers offer algorithmic follow capacity, believes the market will need to undertake much more than a hiring spree for data scientists.
“The new generation of underwriters will need more data and analytics skills, and we’ll need underwriting grad schemes where the post-grads have relevant stem degrees, but not everyone will have to be an expert in computer science.
“I do think the London market is missing something by not allocating more budgets to train up underwriters to equip them with skills around data.”
Given the GWP figures recently unearthed in this publication’s analysis of smart-follow underwriting trends, these points may start to resonate with a growing number of carriers exploring adoption.
Questions of data proficiency
As algorithmic solutions are adopted more widely, Harrap believes some skills normally accumulated later in the journey of an underwriting assistant up to portfolio manager, head of business line or CUO, will be needed earlier, particularly proficiency in working with data.
Harrap added: “Aspiring underwriters, early in their career, should (therefore) be asking ‘what does this mass of data actually mean, how can I structure this data, or enrich it with other datasets? How can I use the data and enhance it to improve our portfolio and our underwriting performance?”
He believes hard, classic data analytics skills will be required for this, but beyond proficiency with Excel, Harrap thinks they’ll need the ability to query databases, for example using SQL, to slice and dice data in different ways to uncover insights.
However, this does not have to drive conflict with existing talent.
Slaughter explained that, at Apollo, even its most traditional Lloyd’s underwriters have found the use of smart-follow engaging, as their process for 20-30 years is being challenged for the better.
“They've found it easy to see the opportunity to enhance where they add value, rather than as an attempt to engineer them out of their job.”
Ki’s Allan said that even in organisations where algorithmic or augmented underwriting is being used, the market will still need lead underwriters to review risks and provide customised expertise.
“I don’t think we’re going to rapidly move away from that”, he added.
Allan said Ki often refers to its own operational process as a bionic underwriter: “You retain the core capability, talent and IP of the underwriters, but you add a better data toolkit around them that allows them to surface better information on that risk, to enhance their ability to underwrite it.”
Evolution of the underwriting team
Experts also believe algorithmic solutions will reshape structures of underwriting teams.
Farris Salah, head of smart-follow at Apollo, believes there will be more specialisms, with some roles focussed on the relationship side with brokers and clients, others focussed on product innovation, some on the analytics, and others who work on engineering and modelling.
“They'll all come together, working in harmony, to solve underwriting problems.”
Within these new-look teams, the market might see the emergence of a job title akin to underwriting product officer or head of product, which would be separate to the insurance products, and focused more on the interface between the technicians and underwriters.
Slaughter explained: “In our business, we have two non-insurance people doing that job – one is an ex-special forces intelligence officer. Both these roles function as a bridge between the technologists and the underwriters, or CUO. That skillset is probably the hardest to find, but it is perhaps the most critical, otherwise communication breaks down and that becomes very hard to manage and resolve.”
Ki’s Allan said: “Head of product is typically a discipline in a technology company, it’s a role that leads the way in terms of scoping and articulating what needs to be done next. It’s almost like a prioritisation layer, it involves exploring what problems a company is trying to solve for their customers, how to solve them with the technology they’ve got and then directing the engineering efforts to the next priority.”
Allan believes that, in the future, an underwriter will need to understand the technology as well as what to prioritise and why, but with assistance from data engineers.
While algorithmic solutions will enable underwriters to spend more time on the high-level, strategic elements of underwriting and relationship management, sources agreed they'll need to be deeply engaged in the carrier’s roadmap for how the underwriting technology evolves.
Another recurring point in sources’ views on team changes was that underwriting assistants may gradually transition into data analysts.
Harrap believes they’ll need to develop a good business-context understanding of what the data they'll be receiving means, and develop hard skills to use tools such as SQL, Python and other programming languages to interrogate and transform this data into actionable insights.
He added: “The way I see the future of insurance is that the data is no longer solely the remit of actuaries. In the future, underwriters and portfolio managers will also need to be able to interrogate and analyse large datasets, working hand-in-hand with their actuarial teams.”
Slaughter sees the underwriting assistant role being carved into two: junior underwriters starting their career and data analysts or curators – who will be engaged with overseeing the structure of data, its quality and timeliness, as they feed it to the underwriter.
The essentials before adoption
However, before getting to the human element, the essential data foundations for algorithmic underwriting have to be invested in: a model must be designed to receive, access, manage and curate data, with governance around this framework.
Apollo’s Salah explained that it is only after developing integrated systems to enable true automation of data processing that a company can think about meaningful optimisation – which is where human underwriting skills come in.
“Often companies will jump to the last step first, hoping an AI tool can solve all their problems without thinking about the necessity to structure the data first.”
Perhaps more companies will explore the fundamentals, given the options opening up to adopt this technology. The latest came through Whitespace’s announcement that it will launch Digital Follow in H2 this year.
Tim Rayner, CEO at Specialty Business Solutions at Whitespace owner Verisk, believes any firm adopting algorithmic underwriting will require deep analytics capabilities, because, to use it, they’ll have to analyse what they’ve written before to produce the rules that can be automated.
“That’s 50% of the equation; the other 50%, which is even more critical, is the need for reporting and monitoring mechanisms to ensure you don’t automate something that’s bad for your business. You need that ‘break glass in an emergency’ element in the rules. The algorithm needs to be overseen by an analytical mind to ensure protections are baked into the rules for certain scenarios.
“You need good data, an individual with solid data skills to analyse that data to come up with recommendations, and a seasoned underwriter to sense check those recommendations, and the final step is that ‘break glass’ control.”
When to invest?
For firms with the operational pillars in place, there are numerous considerations around the skillsets required, the hiring that might be needed, the speed at which operational processes need to change and, perhaps, a cultural-mindset shift. And then there’s the expense of adopting the software itself.
However, in a market where there’s so much emphasis on trimming expense ratios, a tipping point where the advantages for early adopters becomes evident in their underlying performances could come sooner rather than later.