I have worked in the Insurance Industry for over 23 years for both large insurers, brokers and software houses.
Almost all of these organisations were sat on a gold mine of data across a range of software applications, data sources and departments.
Often data was difficult to access and analyse in one place and was likely to be unstructured, particularly where text-based legacy systems are used.
Insurance organisations that embrace and implement advanced analytics such as machine learning, artificial intelligence and predictive analytics will have a distinct competitive advantage to those that don’t.
Adopters will benefit from improved efficiencies, reduced operating costs, easy identification of new product opportunities, niches and more through having a consolidated data view across distribution, underwriting, risk consulting and claims.
But if insurance organisations have disparate, potentially unstructured data systems, where do they start and what are the short, medium and long-term objectives of a winning data strategy?
Why I love the AALP Data Maturity Model
Identifying and understanding where your business is in terms of its data maturity can be a mammoth task.
I love the AALP Data maturity model that we use at Software Solved as it makes the process of identifying and planning a data maturity strategy simple.
The AALP model breaks the process down into smaller, more manageable stages that unlocks value as the organisation progresses, quickly providing a return on investment.
AALP Stands for Access, Analyse, Learn, Predict.
Here is an overview of each stage:
1. Access
You might think that being able to access all of your data is obvious but many businesses with a variety of disparate systems struggle with this first crucial stage.
If you can understand where and how your data is stored and how it aligns to your business objectives it will enable you to consider how this data can be integrated and accessed centrally.
This could be via data integrations between your chosen business intelligence tools and your various software applications or via a data warehouse.
2. Analyse
Once you have established easy access to your organisations data you will be able to consider how to analyse this data and what insights you want to glean.
This could be using smart visuals to simplify how your data is summarised or, it could be focusing on unlocking new value in your data.
Examples could be, identifying opportunities for improving efficiency, reducing cost, profitable business lines or segments, new niches and products.
Having centralised data can enable your business intelligence tools to bring all of this data together across your key business functions such as distribution, underwriting, risk consulting and claims.
This enables data driven decision making and the development of regular automated reporting that help identify insights quickly across your organisation.
Your business intelligence tools will also enable you to identify issues with data accuracy and any areas of unstructured data that need resolving before moving on to more sophisticated data applications such as Machine Learning.
3. Learn
When data analysis becomes a business as usual task, and you are unlocking more value from your data you can start to explore how machine learning and artificial intelligence can benefit your organisation.
Before the learn phase commences it is important that you are satisfied that the accuracy of your data can be relied on and that any potential bias has been identified and excluded.
At this point, the organisation is effectively trialling machine learning and artificial intelligence algorithms to find those that best suit their data needs.
The organisation may also be performing other manual data mining investigations to understand ‘What If’ type queries or just to basically understand if the data exists to answer specific questions.
Here are some basic examples of the application of Machine Learning in Insurance:
⦁ Advice: from managing the initial interaction with the client to determining which cover a client may require (like how Amazon recommends products to its customers). An obvious example would be a chat bot that automatically provides answers to basic customer queries, thus reducing employee time on the phone.
⦁ Claims: automating processes from claims registration to claims settlement. Some Insurers are using Machine Learning to identify cases that may be fraudulent.
⦁ Sales: Automatically identify cases that are likely to lapse in advance for a focused sales effort
An Insurance Providers Data Analysts may also use Machine Learning to discover predictors in claims activity which can help identify assumptions and feed these into its pricing models, underwriting and risk analyses and actuarial analyses.
If your organisation gets this stage right, the benefits can be significant.
It also paves the way for implementing data automation and advanced levels of trends and insights.
4. Predict
The last stage in the AALP model is Predict.
When an organisation reaches this point, it should have developed a good understanding of which machine learning and artificial intelligence approaches work for them and have begun to implement them into their data production systems.
This will move them away from manual data insights to automation.
In addition, the Predict phase lays a strong foundation for experimenting with predictive analytics which has multiple applications and benefits for insurance organisations.
Imagine being able to automatically predict outcomes with a high degree of accuracy using a number of internal, open source and external data sources?
Examples could be:
⦁ Predicting future loss probabilities based on current data trends
⦁ The impact on profitability following a rate, excess or cover change
⦁ Benchmarking your enquiries against your current clients to predict profitability based on your historical claims, risk and underwriting data overlaid with recent (or even real time) data such as crime and weather patterns.
⦁ Assessing how a constantly changing business mix might affect your portfolio in real time
⦁ Automatically attributing a risk score by property peril based on data insights from internal, paid and open source data such as how the current crime rate affects theft and malicious damage
⦁ Predicting the impact on profitability following the investment and implementation of a risk improvement
⦁ What your book could look like if a specific cover or product is introduced
The list can be endless.
How long would it currently take to model this data? Imagine a world where this level of insight was instant.
What do you think?
Has this helped you visualise how you could tackle your data challenges?
If you need some help on understanding your organisations data maturity and implementing a winning data strategy, Software Solved can help in a number of ways by providing:
1. A Data Maturity Assessment which helps a business develop a blueprint of its data and systems progression across a number of key areas including data access, visualisation, machine learning and AI and Predictive Analytics.
2. Consultancy on Data Warehousing and Data Integrations to help you successfully navigate complex projects.
3. Flexible Resource for instances where you need additional data analysts, developers, business analysts and project managers on a short-term basis to help you shape and implement data projects.
