machine learning in insurance building.jpg

Author: Mitchell Feldman, CDO  
Last updated: March 26, 2018

Looking for patterns in large volumes of data is hardly new for insurance companies.

In fact, big data is almost the foundation upon which the insurance industry is built. It’s how decisions are made and policies are determined.

And insurers are not oblivious to how technology can enable this; half of insurance executives are prioritising technology investments to capture new client insights over the next three years. On top of this, artificial intelligence (AI) and machine learning in insurance are expected to be bigger future trends than in many other industries.



’90 percent of the world’s data has been created in the last two years,’ says George Lee, CIO at Goldman Sachs. ‘The ultimate question is really what insight and value can we draw from that data.’


Data analytics as we know it today is changing.

It is evolving into something more powerful with the advent of machine learning and artificial intelligence. Taking advantage of new technology today can, and will, make a significant difference tomorrow.

insurance guide


5 key Machine Learning Opportunities in the Insurance Industry

The opportunity for insurance machine learning is there and ready to be put to use by any insurers willing to embrace the technology.

In fact, there are already plenty of ambitious firms making the most of this potential and here are five of the key successes and opportunities we’ve noted.

1. Running Models through Machine Learning

While running business-critical data through tried-and-tested in-house actuarial models might be common practice, it is no longer the most efficient method. With today’s technology, insurers can take these and run them through machine learning paths and patterns which have been created by industry leaders like Microsoft.

Starting small, insurers can learn which models to exploit by conducting small pilot tests and implementing at scale when the right model is found.

2. Using the Internet of Things

Machine learning is more powerful the more data it has access to and nothing can collect user data quite like the Internet of Things (IoT).

Organisations will already have, or should be collating, vast amounts of data to provide better value to their customers and IoT is a ripe source for such data. Currently, the insurance sector is lagging behind in data analytics and needs to up its game to compete with the likes of retail and transportation.

machine learning with the insurance sector.jpg

For example, RedPixie worked with a large insurer to collect telemetry data from cars, analysed their driving patterns, and used this data to develop risk analysis models. On this project, we were able to demonstrate how car fleet managers could reduce their workload by more than 70 percent using machine learning, filtering out ‘noise’ like speed bumps or pot holes.

Further reading: find out how Microsoft Azure machine learning works.




Make Risk Management less Risky

Download your free guide to successful IoT implementation

 Email me a copy →



3. Creating a Competitive Advantage

By running models through machine learning and utilising IoT, organisations can create new revenue models, quickly explore niche opportunities, and create incomparable client value.

Three or four big name players currently dominate the insurance market, but access to tools such as machine learning will help smaller firms capitalise on specialised services.


Spending less money automatically gives insurers a competitive edge, but machine learning can offer even more value. Deeper analytics lead to innovations in products and services that push an insurer ahead of its competitors by optimising their risk profile and pricing policies more effectively.

4. Better Allocation of Staff Resources through Superior Automation

Although less reliant on manual processes than they once were, insurance companies still rely heavily on people to make decisions and analyse information.

With a move towards smarter automation, the digitisation of these tasks will help insurers save more money.

A recent example of this is the work that Lapetus, a US-based startup, has done with machine learning for life insurance. To eliminate the lengthy, human-led process of assessing a customer using old actuarial models, Lapetus has designed a machine learning programme that uses just a single photograph to underwrite policies.

lapetus tech.jpg

From just one selfie, the machine learning algorithm looks for indicators like gender, rate of ageing, and body mass index to provide an accurate prediction of life expectancy.

By automating tasks like this, organisations can redistribute their workers across different areas of the business to create new elements not previously considered.

5. Maintaining Compliance

With machine learning, insurers can review and analyse all forms of structured and unstructured data, including pictures, videos, and audio, to a much higher degree, using a larger pool of data.

With this capability, insurers will be able to improve compliance and prevent the mis-selling of products – something that’s particularly important ahead of the GDPR deadline in May 2018. As a damning example, in 2015, a US insurance company paid a $271,815 settlement to UK courts over a compliance violation in their London-based offices.

Big name Insurers lead the Machine Learning charge

When it comes to machine learning in insurance, AIG and Zurich are already taking big steps.

Insurers have always looked for patterns in data, according to Monika Schulze, Global Head of Marketing at Zurich, but it is time to use more complex algorithms that outclass the ability of any human touch.

'[Fraud mitigation] is where I see insurance applying machine learning, to improve the P&L,’ says Monika Schulze. ‘It is a much faster process and it is easier to reduce errors by using machine learning to process large amounts of data.’

George Argesanu, Global Head of Advanced Analytics, Personal Insurance, AIG agrees:

‘The level of sophistication and tools has changed over time and I look at Machine Learning and AI as transformative for the way we try to solve the same problems, while also gaining insights from places where traditional methods fail,’ 

Insurers such as State Farm, Progressive and Allstate are also moving into the machine learning sphere by using virtual assistant algorithms to determine safe driving behaviour and telematics.

insurance guide

The future of Machine Learning in Insurance: when, not if

Adopting machine learning in the insurance industry will do more than provide a competitive advantage – as attractive as that is for insurers. More importantly, it will help to prevent the 350 cases of insurance fraud worth £3.6 million uncovered every day. The number of fraudulent claims is rising every year, and with it the overall cost to insurers.

 If you haven’t already factored machine learning into your future business objectives, there are a number of steps you can take today:

  • Start prioritising business outcomes that can really transform your business
  • Engage with an IT partner who can help you make the right decisions
  • Implement a good cloud structure that allows you to scale up
  • Consolidate your data environments so that databases can be well utilised
  • Test small cases of machine learning with cost-efficient Azure environments (or an alternative), noting successes and scaling accordingly
  • Measure the ROI and remain as agile as possible

 Outperforming the competition and battling insurance fraud won’t be easy but it’s no longer a matter of if fellow insurers should adopt machine learning: it’s a matter of when.

Want to learn more about the exciting ways technology is impacting the insurance industry? Download our free guide ⇓

 insurance guide

Topics: Digital Transformation, Insurance, Machine Learning

KNOWLEDGE IS POWERGet weekly email insights from RedPixie

IT Partner Success Choose correctly and boost your value by 56% →
surface banner.jpg