MGAs are leading the pack to invest in machine learning, working to ‘deploy it far quicker into their distribution and into their business than insurers will’

The march of the machines continues in the underwriting sector, with MGAs taking the lead in using technology to improve underwriting processes and customer service levels.

The use of artificial intelligence (AI) and machine learning (ML) by underwriters has the potential to remove many frictional costs, as well as freeing up underwriting professionals to focus on the more complex risks in their portfolios.

However, along with the undoubted opportunities of technology comes a number of challenges - not least around the collection and use of data.

But will mainstream insurers be able to follow the lead of more nimble, tech-centric MGAs?

Automation answer

One MGA, C-Quence, is using ML to help pre-populate policy documentation and record referral information – this enables an automated response or more detailed information to be sent to the underwriter in answer to subsequent queries.

C-Quence said its C-Q Elements platform is underpinned by more than 50 external data sources and thousands of algorithms. Natural language processing and ML means the platform learns and anticipates how risks evolve to continually improve automation capabilities.

In turn, this speeds up quotations and underwriting by reducing the number of questions asked, which means that in some cases, C-Quence can quote and bind a policy within three minutes.

Elliot Biggs, chief information officer at C-Quence, explained: “We call the fully automated quote to bind process the ‘Happy Path’ because it is very efficient and fast.

“Underwriters are only involved in more complex risks where we have insufficient data to learn from past activity or where we want to take a more considered view until clear and consistent decision-making patterns have emerged based on similar risk profiles and characteristics.

“Underwriters are freed from repetitive tasks to concentrate on the more interesting, rewarding and valued aspects of their profession.”

Levelling up service standards

Biggs said that ML can help underwriters radically improve the service levels currently experienced by brokers and their clients.

He added: “Traditional underwriting approaches are often too slow, too costly and can suffer from inaccuracies and inconsistencies due to the variability of the process. Automation and reengineering of repetitive processes is part of the answer.

“At C-Quence, we have achieved high levels of automation by codifying thousands of underwriting rules and rating algorithms, leveraging our vast underwriter experience and augmenting this with ML.

“We constantly evaluate the inputs and outcomes of manual underwriting actions and constantly evolve our automated platform based on learned behaviours, leveraging ML alongside the ingestion, structuring and management of vast quantities of data from multiple sources in real-time.

“These are profound advancements which will fundamentally impact processes and business models across our industry by delivering better, balanced outcomes for customers, their brokers and underwriters.”

Personal versus commercial lines advancements

Using ML to improve manual and repetitive processes can offer insurers, brokers and MGAs a route to providing a better service to customers in certain business segments and product lines.

For example, Mike Keating, chief executive of the Managing General Agents’ Association, said automation can be particularly beneficial for the SME sector, which can tap into automation advancements arising from the personal lines market.

He explained: “Looking at the ‘S’ of the SME market, the use of algorithms and what has been done in personal lines for quite some period of time [around] AI and ML - providing it is set up correctly and reflects the market, the product and the pricing of the business you want to do, the first thing [automation] will do for MGAs is remove a lot of frictional cost.

“If you do that in a customer segment that aligns to AI underwriting, then you can concentrate on other areas of business growth.”

However, “there’s still some way to go” in terms of applying automation to commercial line risks, Keating added.

He continued: “In the commercial space, there’s still some way to go.

“In personal lines, it is clearly far more advanced - certainly in terms of using life science data. It is a force which the right type of MGAs will need to embrace.

“In terms of deploying [machine learning] and investing in it, you have to make sure you do the full end-to-end business case. It must add true value to your business, but also to your customers.”

Standardising data

However, to get to the point where firms can begin to add value to their business processes, insurers must standardise data around underwriting actions, outcomes and claims trends.

This can prove a major challenge due to the manual way most of this activity is currently undertaken and because of the unstructured nature of the data which is available.

Biggs said: “Data is at the heart of ML models and their ability to learn and be tuned. Without this data and the mechanisms around its capture, structure and storage, it won’t be possible for firms to progress their ML capabilities.

“This is where C-Quence is different, building our platform around the capability to capture data and ensuring that we can capture and manage vast quantities of information and patterns in real-time.”

The benefits of ML are not only available to new ‘digital-first’ market entrants, but MGAs not encumbered by legacy systems may have an advantage - provided they take a ‘data-first’ approach.

Keating said: “If you’re going to deploy this, you’re going to deploy a richer blend of data in order for the AI underwriting to work. If you’re getting that data, then your risk selection and your pricing will be more forensic and more laser focused.

“AI underwriting is the continuous refinement of the technology to meet the needs of insurers, brokers and MGAs.

“MGAs will go and make the investment. First and foremost, they will have good insight from their brokers and their own businesses into what products would actually lend themselves to AI and ML underwriting.

“They will make the investment and critically, they will deploy it far quicker into their distribution and into their business than insurers will.”