6 Sep

Increasing Profitability Using AI in Financial Services and Insurance

Deepak Ramchandani Vensi

Margins are getting tighter and tighter in the financial services and insurance industry.

The combination of shifting consumer habits, increasing regulation and a growing pool of up-and-coming challenger startups has meant that the competitive environment has intensified significantly in recent years.

Direct Line recently issued a profit warning saying that its combined operating ratio⁠ - which measures costs as a proportion of premiums⁠ - will be between 96% and 98%. The closer the ratio is to 100% the less profitable the company will be.

And they’ve recognised the need to take action to get back to profitability: “We have already taken actions including increasing prices and deploying new pricing capability to restore margins.”

But when competition is growing, you can’t just rely on increasing prices.

Instead, companies must generate new sources of both efficiency and revenue to get margins back to a comfortable point.

That’s where AI comes in.

In this blog, we’ll look at how AI can help your business to increase efficiency and drive new sources of revenue to combat flatlining margins in the increasingly competitive financial services industry.

How Does AI Improve Margins and Help Create New Revenue Streams?

There are two main ways that AI can improve profit margins.

The first is using AI to find new areas in your business that can be automated. This can drive massive efficiencies and save considerable amounts of time, money and effort.

The second is to use data in order to get to know your customer more deeply. This can create opportunities to increase revenues and even open up entirely new revenue streams.

Let’s take a look at each in turn.

1) Driving efficiencies with automation

The first is driving massive efficiencies through automation. This means optimising operations with precise forecasting, predictive maintenance, quality control and using intelligence to reduce risk. It can also be used to identify inefficiencies in the first place and spotting where costs can be cut.

There are two kinds of automation worth considering: process automation and intelligent automation.

Process automation performs routine business activities by following the same actions that a human would do via a software interface. It is best used for highly repetitive tasks with fairly predictable outcomes.

This is for things like:

- Supply chain optimisation

- Inventory management

- Back office automation

- Generating financial insights

For example, you could train an AI to reply to a common refund request or sift through insurance applications to identify patterns and make initial decisions on what to do with the application. Or internally they could be used to rapidly generate reports, invoices and so on.  

This is where there is massive potential to generate considerable short-term savings as well as freeing up time and resources.

Intelligent automation is more sophisticated and can be used to automate nonroutine tasks that require intuition, judgement and creativity. This is done through capabilities like image recognition and natural language processing.

Current examples in the industry include:

- Analysing portfolio data and generate automated reports for their clients

- Improving regulatory compliance by monitoring employee communications for evidence of noncompliance

- Fraud detection and claim evaluation

Use cases for intelligent automation are boundless, but typically more complex than process automation use cases.

However, they have the potential to transform what were previously not only tedious but actually rather complex, manual tasks that had to be performed by a human into automated tasks with equal, if not superior, results.

2) Creating opportunities with data

The second way is by helping you to go deep with your data so you can for instance get to know your customer much more intimately, opening up new avenues for improved customer satisfaction and experiences, enhanced loyalty and better interactions.

This is more at the revenue generation and value-add end of the AI spectrum, where the power of AI is used, not to create efficiency, but to create knowledge.

AI helps businesses to take the customer data that they have and to turn it into something that can be used to improve the services, offerings and customer experience that the customer receives.

This includes things like:

- Personalised services and offerings

- Data-led business decision-making

- Individualised pricing

- Personalised marketing

- Customer churn and retention

- Customer experience (e.g. chatbots)

This is where there is huge potential to open up incrementally better and new sources of revenue by creating virtuous circles whereby you get to know your customer better, which allows you to serve them better, which allows you to get to know them better and so on.

As you improve your service, you improve your data, which helps you to further improve your service.

For example, research by Futurum Research has shown that when chatbots are supported by AI they can use the conversation history and emotional recognition technology to better understand the context of the customer’s inquiry and provide a better service. They even get better at learning when human intervention is required and when they need to pass the customer on to a human customer agent.

They found that “customer experience delivered by humans in partnership with machines (AI) can boost loyalty, help the company understand customer needs better, and improve cross-selling and up-selling opportunities”.

AI Example Use Cases in Financial Services

Let’s take a closer look at some example use cases to help bring this to life a little.

Individualised and optimised insurance pricing: AXA Insurance

Currently, insurers create insurance policy quotes based on a few simple values. So for car insurance, for example, this might be age, claims history, type of vehicle, driving history and so on.

If you are the same age as someone, have never made a claim and own the same kind of car, you will receive the same policy quote.

AI can instead draw on a vast range of personal and historical data in order to much more accurately assess the risk for each customer and provide a quote that much more accurately reflects their true position.

Let’s look at how AXA Insurance has done this.

AXA Insurance knew that 7-10% of their customers cause a car accident every year. Of those, 1% are ‘large-loss’ cases that require payouts over $10,000.

Their data science team managed to create an experimental deep learning machine learning model that was 78% accurate in predicting those customers that would cause a large-loss driving accident.

This will help AXA optimise their pricing structure as well as allow the creation of new insurance services such as real-time and personalised pricing at the point of sale.

Using deep learning models to prevent fraud

As online shopping and transactions increase, cybercriminals are leveraging ever more complex attacks and fraud attempts, netting billions of dollars every year.

Financial services firms - and consumer banks, in particular - are having to keep pace with the sophistication of the cyber threat actors.

American Express has maintained the lowest fraud rate in the financial services industry for 13 years in a row, according to The Nilson Report.

They have adopted AI to help them identify and prevent fraud in real time.

American Express has adopted fraud detection technologies optimised with deep-learning-based models to monitor each transaction on the $1.2 trillion that annually passes through their platform in real-time.

The bank is leveraging NVIDIA TensorRT, a high performance deep learning inference optimizer that minimises latency and maximises throughput.

The new GPU-accelerated technology has helped the bank improve fraud detection accuracy by up to 6%. This is achieved while operating within a tight two-millisecond latency requirement, which is a 50x improvement over CPU-based configurations.

Final Thoughts

But you can’t just jump straight into pumping out AI-driven projects and ML algorithms!

You need to have the fundamentals in place first.

Check out our other articles that go into greater detail on the key ingredients required and the common roadblocks to getting your AI projects into production!

- 5 Blockers Preventing Scaled Adoption of ML & AI in Financial Services and How to Overcome Them

- 11 Ways to Get Your Machine Learning Model Into Production

Mesh-AI is a global consultancy that uses data, machine learning and artificial intelligence to deliver transformative outcomes for enterprise organisations.

Our mission is to make data your competitive advantage.

Get in touch to see if we can help you to use AI to restore your profit margins and help open up new revenue streams!

Interested in seeing our latest blogs as soon as they get released? Sign up for our newsletter using the form below, and also follow us on LinkedIn.

Latest Stories

See More
This website uses cookies to maximize your experience and help us to understand how we can improve it. By clicking 'Accept', you consent to the use of these cookies. If you would like to manage your cookie settings, you can control this in your internet browser. Find out more in our Privacy Policy