Underwriting is the fundamental process that allows insurance companies to operate profitably while providing financial protection to their customers. However, underwriting has traditionally been a manual, labor-intensive task involving assessing risk based on limited information. As a result, the process tends to be slow, inconsistent, and reliant on generalized assumptions.
The emergence of sophisticated data analytics, AI, and other technologies is rapidly changing the game. Insurtech companies are leading the charge in using technology to transform underwriting in revolutionary new ways.
In this comprehensive blog post, we will analyze the most important technologies reinventing insurance underwriting, highlight innovative companies, discuss industry trends, and examine the future outlook. Let‘s dive in.
What is Insurance Underwriting and Why Does it Matter?
Insurance underwriting is the process of evaluating and pricing risk to determine appropriate premiums to charge customers based on their level of risk exposure. The goal is for the insurer to make a profit, while protecting the policyholder from potential financial losses.
Effective underwriting allows insurers to:
- Accurately price policies based on true risk levels
- Optimize profits on their book of business
- Maintain a suitably diversified mix of policies across industries, geographies, etc.
- Operate efficiently enough to meet customer expectations on policy issuance times
Getting underwriting right is crucial for insurers. Those with poor underwriting practices face losses when claims exceed collected premiums.
As an example, below is a comparison of top US insurers across key underwriting metrics:
| Insurer | Underwriting Expense Ratio | Combined Ratio |
|---|---|---|
| Berkshire Hathaway | 23.7% | 96.4% |
| Progressive | 19.1% | 93.7% |
| Travelers | 32.1% | 98.3% |
| Allstate | 22.3% | 95.0% |
Underwriting expense ratio measures the cost of underwriting as a % of net premiums written. Combined ratio includes both underwriting expenses and losses.
The data shows leading auto insurers like Progressive with tech-driven underwriting boast superior expense and combined ratios versus less advanced peers. This directly translates to stronger profit margins.
Top-performing insurers generate nearly 3X the return on equity compared to median companies, per McKinsey research. This underscores why perfecting underwriting is an imperative.
The Underwriting Technology Revolution Powered by Insurtech
Advancements in data & analytics, connectivity, and computing power have sparked a revolution in insurance underwriting capabilities.
Insurtech firms are aggressively developing and deploying technology solutions to overhaul legacy underwriting processes. Incumbents have also begun modernization efforts, though at a slower pace.
Here are the most pivotal technologies and techniques driving underwriting transformation in insurance:
1. Predictive Analytics & Machine Learning
Sophisticated ML algorithms applied to structured and unstructured data are enabling insurers to accurately model and score risk with a high degree of automation. Key capabilities include:
- Automated data ingestion & cleansing: Extracting insights from disparate data sources
- Predictive modeling: Estimating future claims and losses with precision
- Risk scoring: Assigning policyholders to precise risk segments
- Straight-through processing: Automating the entire policy underwriting workflow for accelerated cycle times
For example, Cape Analytics leverages computer vision and geospatial data to auto-generate accurate property risk scores. Their models can evaluate a property in just 90 seconds versus weeks through manual underwriting.
A major European insurer was able to trim 2 weeks from their home insurance underwriting process after deploying an ML-based solution, reducing loss ratios by 4 percentage points in the first year. This demonstrates the power of machine learning applied to risk selection and pricing.
2. Real-time Sensor Data Analytics with IoT
The proliferation of internet-connected sensors allows insurers to get far richer and real-time data feeds to enhance risk visibility.
Examples include:
- Auto insurers accessing driving behavior data from telematics
- Life insurers getting health & wellness data from wearables
- Commercial insurers monitoring sensor data on insured equipment
This translates to superior loss forecasting, personalized policy pricing, and incentives for risk-reducing behaviors thanks to IoT-enabled usage-based insurance models.
One specialty insurer in the healthcare sector improved loss ratios by 29% using IoT initiatives including asset tags to track utilization and conditions of insured medical devices real-time. This enabled proactive loss control and lower premiums.
3. Natural Language Processing for Unstructured Data Analysis
Only 20% of data is structured, while 80% is unstructured in formats like texts, emails, PDFs, images, audio, and video.
Using NLP and text analytics, insurers can rapidly process high volumes of unstructured content to uncover risk insights that were previously buried within mountains of documents.
For instance, text mining medical records improves health risk analysis for life & medical underwriting. Image recognition and OCR extract insights from scanned documents and photos.
A top 10 global insurer optimized their life & health insurance underwriting by deploying an NLP engine to extract intelligence from doctors notes and medical reports shared by applicants. This boosted risk assessment accuracy by 15% and premium optimization.
4. Advanced Simulation Modeling with Digital Twins
Insurers have long struggled assessing risk for rare but catastrophic loss events that have sparse historical claims data.
Virtual simulation modeling of digital twins allows more credible loss forecasting for such events. Software models the impact of simulated disasters like floods, hurricanes or cyber-attacks on digital replicas of insured assets. The automotive and property insurance sectors are leading adopters of this trend.
For example, Axis Insurance developed a cyber risk modelling platform using advanced simulation techniques. It runs multiple loss scenario simulations on highly detailed firm-specific virtual models to quantify financial impact and tailor cyber risk coverage limits accurately.
5. Blockchain for Enhanced Data Sharing Between Carriers & Partners
Strict data regulations hinder certain types of sensitive data from being shared between insurance carriers and healthcare providers. Using blockchain infrastructure for tamper-proof data sharing allows insurers access to richer datasets for underwriting life, health and workers compensation exposures, in a fully compliant manner.
6. Computer Vision for Automated Visual Risk Analysis
Cameras provide a rich stream of visual data capturing details on properties, buildings, equipment and manufacturing processes – all crucial for commercial insurance underwriting.
Advances in image recognition and computer vision techniques allow insurers to automatically extract visual risk indicators from photos and videos to enhance underwriting quality and speed.
For example, At-Bay offers automated cyber risk protection for tech companies. Their Computer Vision models scan thousands of a company‘s public images to identify security vulnerabilities like outdated software or lack of encryption visible. This feeds into their real-time underwriting system to score cyber risk.
7. Graph Neural Networks to Detect Sophisticated Fraud
The bad guys are using more advanced techniques like creating deliberately complex webs of fake identities and suspicious financial transfers to commit large-scale fraud.
But graph neural networks (GNNs) can help insurers uncover these deceptive patterns. GNNs combine graph theory and deep learning to model relationships and interdependency between entities to separate fraudulent networks from legitimate clusters.
I believe GNNs will become a vital tool enabling insurers improve fraud prediction, avoid unnecessary losses and price risk accordingly during underwriting.
Overcoming Resistance to Underwriting Technology Change
Despite compelling benefits, transitioning from traditional underwriting methods to tech-centric modernization faces barriers:
Talent Shortages: Many underwriters lack advanced analytics skills to build models or interpret machine learning outputs. Re-skilling strains resources.
Data Hurdles: Integrating complex legacy systems with new data sources poses engineering challenges. Ensuring robust data governance is equally vital.
Change Resistance: Stakeholders clinging onto status quo ways resist altering comfortable work modes.
Unproven Reliability Concerns: Suspicion exists whether machines can perform risk selection as adeptly as veteran underwriters.
Cyber Exposure Fears: Greater reliance on technology heightens vulnerability to outages and cyberattacks.
Here are tested strategies to tackle these change obstacles:
-
Take an iterative approach rolling out modernization initiatives by line of business and proving value delivered before going full steam.
-
Develop specialized underwriter roles supporting technology adoption like Analytics Underwriters rather than expecting all personnel to transform promptly.
-
Leverage cloud infrastructure for increased agility, lower TCO and heightened security standards relative to aging on-prem systems. This debunks cyber risk notions.
-
Commit to continuous model tuning and rigorous validation testing to address questions of predictive reliability and instill trust.
Overall, underwriting leaders have to delicately balance people, process and technology elements to fuel this transformation.
Innovative Insurtech Firms Driving the Underwriting Revolution
Established insurance carriers have been slow to modernize legacy underwriting systems and processes. As a result, the most exciting innovation is coming from nimble startups using cutting-edge technology to disrupt commercial insurance underwriting.
Here are 5 of the top insurtech innovators transforming small business underwriting:
Next Insurance: Using advanced predictive analytics, Next Insurance simplified the quoting process and reduced submission-to-bind times to days or minutes across their portfolio of small business products. Over 30 variables power their risk models driving automated underwriting.
At-Bay: At-Bay developed an AI-powered risk assessment engine that scours tens of thousands of data points to quantify cyber risk and underwrite in real-time. Their precision underwriting model cuts the typical 30-45 day underwriting process down to just minutes.
Slice Labs: Slice Labs delivers on-demand insurance products for the digital economy using sensor data models and automated underwriting. As an example, their mobile-based Clover product for on-demand commercial auto insurance leverages GPS and vehicle sensor data.
Planck: Planck uses a big data platform, predictive analytics and digital user engagement to offer data-driven commercial insurance policies with streamlined underwriting.
Acumen Social: They provide automated underwriting for employment practices liability insurance using an AI reviews employers’ public social media data to score policy applicants based on risk signals like employee dissatisfaction.
While the above firms focus specifically on commercial lines underwriting, there is no shortage of insurtech innovation across other P&C, life and health insurance lines too.
If you are interested to discover other insurtech underwriting solution providers suited to your specific needs, please fill out this form and our team would be glad to offer tailor-made recommendations!
The Critical Components of An AI-powered Automated Underwriting System
Drawing from my past expertise building commercial insurance underwriting models leveraging big data and machine learning, I wanted to share principles and best practices to follow for optimal outcomes:
1. Model Development Methodology
A structured model development approach entails:
- Data inspection, cleansing, feature extraction
- Exploratory data analysis to surface insights
- Input feature finalization based on predictive strength assessment
- Comparing multiple candidate algorithms like regression, neural nets, gradient boosting etc. and tuning hyperparameters
- Rigorous testing on out-of-sample data through walkthroughs with domain experts to fix model blindspots
2. Training Data
Models are only as good as the data they learn from. Prioritizing acquisition of quality data both in breadth and history is vital even if requiring upfront data alliance investments.
Some attractive training data categories include:
- Financial credit & liability signals
- Public records, licenses and ownership information
- Commercial property attributes data
- Brand online reputation and security ratings
- Claims and litigation involvement histories
- Business operation footprint intelligence
3. Risk Indicators
Each line of business and coverage type have specialty indicators that experience shows are strong predictors.
Cyber underwriting models account for technical infrastructure criteria like:
- Presence of firewalls, tokenization, encryption, SIEM
- Vulnerability scan ratings
- Software currency
- Backup power supplies
- Employee cybersecurity policy POS/NEGs
There are over 75+ predictive variables for most commercial risks. The key is distilling and testing to arrive at the optimal signals per product.
4. Regular Retraining
Underlying risk dynamics keep evolving so models require periodic refreshing with new data to maintain predictive accuracy.
Quarterly or bi-annual retraining cycles are recommended depending on pace of change in the insured industry. More frequent in hypergrowth tech sectors.
How is Underwriting Technology Impacting the Insurance Value Chain?
Transitioning to data-driven automated underwriting powered by AI and analytics unlocks a multitude of benefits across the entire insurance value chain:
Underwriters: Technology eliminates repetitive manual tasks allowing underwriters to focus on higher judgement-intensive complex risks. Tools augment underwriter expertise.
Actuaries & Risk Managers: Significantly enhanced data assets and sophisticated modeling techniques lead to improved loss forecasting, reserving analysis and portfolio risk monitoring.
Distribution Partners: Shorter quote-to-issue cycles coupled with straight-through processing allows agents and brokers to generate quotes and bind policies in minutes rather than days. This results in higher sales productivity.
Marketers: Data mining aids precise segmentation while behavioral analytics provides sharper targeting ability to optimize customer acquisition costs.
Claims Analysts: Pattern recognition abilities using past claims data results in more accurate fraud prediction and proactive loss control intervention.
Customers: An overall enhanced digital customer experience with self-service buying, near instant quoting and claims settlement. Additionally, usage-based insurance unlocks cost savings for lower risk customers.
The cascading impact is faster product speed-to-market, lower operating costs, improved loss ratios and happier customers for technology-forward insurers.
Industry Outlook and Predictions
Here we take a forward-looking view at the most important underwriting technology and market trends that will shape insurance in the years ahead:
-
Adoption of AI and data-driven automated underwriting will accelerate across both personal and commercial lines. Leaders predict over 50% of underwriting will be AI-led by 2030.
-
Commercial lines seen as the “greenfield” opportunity will witness the most underwriting innovation. Significant room exists to disrupt the archaic small business underwriting segment especially.
-
Core underwriting platforms offered by legacy system vendors will continue to lag cutting-edge capabilities, triggering more insurers to examine cloud-native digital insurance platforms.
-
Usage-based and behavior-based insurance models will keep gaining traction across both personal and commercial lines supported by telematics and IoT-based data.
-
As analytics models depend heavily on quality data, strategies to expand access to richer datasets will be a competitive advantage. This makes blockchain-based data sharing alliances more common.
-
Partnerships between insurers, MGAs and insurtech MGAs will keep increasing as digital-first insurgent business models focused on specific niche markets emerge. More M&A likely too as insurers look to rapidly gain digital underwriting competencies.
-
Underwriter roles will evolve to be more technical and analytical rather than manual paperwork shuffling as underwriting transforms into a digital, cloud-based workflow. New hybrid underwriting roles merging underwriting with data science skills will arise. Talent strategy becomes vital.
-
Regulators will aim to strike the right balance between data-sharing to improve underwriting outcomes and maintaining strong consumer data privacy safeguards. Managing this would require deft digital ethics handling.
The Bottom Line
Underwriting lies at the heart of the critical risk selection, pricing and portfolio optimization decisions that determine the financial success of insurers.
Legacy underwriting processes have been human labor intensive, relying on primitive actuarial techniques and dusty databases. Today underwriting is undergoing a significant transformation thanks to insurtech innovation.
Cutting-edge data analytics approaches powered by AI and machine learning allow underwriters to reinvent risk assessment and pricing processes.
Technology leaders like Next Insurance and At-Bay exhibit how modern cloud-native software development coupled with data science can simplify and enhance underwriting quality and speed.
The performance gap between laggard carriers still mired in antiquated underwriting systems and processes versus those embracing insurtech and automation will rapidly widen in the years ahead.
Stay tuned to our blog for more insights tracking the underwriting technology evolution!
