Why data and machine intelligence will become the new normal in insurance

Now, more than ever, there is a focus on digitisation and sustainability. This is true across sectors but is particularly relevant for the insurance industry, who are viewed by many as being particularly slow to embrace digitalisation. In this article we consider trends like access to data enabling increased personalisation for consumers, and we look at business models that are emerging as the industry evolves into the next phase.

Customer touchpoints are becoming more and more digital

Imagine having two touch points a year with google. You’re allowed only two searches a year to get answers to two burning questions and google have only those two questions to use as data points to infer something about you. How likely is it that the advertising on that platform is going to be relevant to you? How likely is it the top search results will be of the most relevance to your enquiry? And how satisfied will you be with their service?

Although it is fun to consider what two topics would take priority, in reality many would view this as the dystopian plot of a science fiction novel. So many of us take curated content and relevant information for granted that anything else would see people leaving the platform in droves. This is the reality, though, for insurance companies.

In the US, there is currently only an average of 2.7 interactions every year for insurers with their consumers (Bain & Company, 2018). When 4 billion people connect online and more than 90% of people use mobile devices (We Are Social, 2018) there is no good reason for customer touch points to number so low. The way a product is sold matters almost as much as what is being sold when it comes to building customer loyalty, retention and the positive associations that come with seamless experience – word of mouth and positive recommendations are still some of the ways people make decisions about firms to try.

Data driven touchpoints

For many service providers, 2.7 touch points annually would be too few for them to adequately run their business on. Data is integral to improving processes and is being treated with, with it expected to reach 175 zettabytes by 2025 (Swiss Re, 2020). Any company that therefore does not collect data for their users, or, more importantly, actively collects data and utilises feedback information to better inform their service offerings, will be left behind. The most successful companies in the world are those that solve problems and pain points before the user even knows they exist; Amazon, for example, letting us know what people similar to us have also bought so that we know not to forget our batteries, or Asos, offering views of what clothes look like on a variety of body types. These customer informed solutions are informed by the data these companies collect on a second- by- second basis.

The stakes are rising. Demands and expectations on companies are increasing. In a world that demands ethical business practices, socially minded enterprise, transparency and sustainability, corporations need to keep up to date with the rapidly fluctuating needs and behaviours of their customers. They need to look at innovative solutions that give them a better view of their own supply chains and they need to make sure that what is important to their customers is important to them. Pre-empting consumer needs can only be done by analysing the vast amounts of data these companies collect. Amazon has filed a patent for a shipping system that pre-empts what you will buy and ships products in the individual’s direction before they’ve even bought it (Lomas, 2014). Other companies are getting in on the pre-emptive action; the Rapid team at Myntra, the overseer of Moda Rapido the fashion store, predict what fashion items will be fast sellers, employing more data scientists than they do designers to analyse trends in consumer buying behaviour (Flipkart, 2018).

Machine learning and artificial intelligence are the latest tools that purport to improve this pre-emptive ability and make better decisions. Currently, in sectors as diverse as finance, law and healthcare, it has been found that technology works best when used in conjunction with humans. Instead of letting machines run wild, we see the most effective companies using augmented human intelligence which combines a human being’s understanding of emotion and their flexibility of thought with the highly logical, analytical skills of machines. Rather than letting a machine detect a condition and then inform a patient about the likelihood of survival after diagnosis, we check with the doctor, who then informs the patient. Rather than letting a machine make bail decisions, the machine makes a recommendation and it is reviewed by a judge who takes contextual factors into account. For insurance, this augmented intelligence could revolutionise almost any area of the value chain; we could see insurers leveraging flexible, product- agnostic and fully integrated digital platforms that engage personally with the consumers.

Figure 1: Depiction of opportunities for platforms in insurance

In this diagram, we see both core services and market services operating on different but integrated platforms. Core services can be run by the industry players – the insurers and the reinsurers, for example – performing all relevant tasks of their value chain. For B2C and B2B market services and customer- facing functions it is more likely the industry will ‘plug’ into already existing interfaces, using third parties to reach more consumers. This happens already with insurance plug ins on travel portals, online commerce or ride hailing platforms. The same pattern applies to the rest of the B2B industry structure.

Although the goal is to automate as many services as possible for better customer experience and increased productivity, we have seen in other industries how there is still a degree of human oversight that is necessary when services either face the public or are making important decisions that require some emotional oversight. In claims management, for example, it is important that decisions are well justified and are appropriate before decisions are given to people. For now, a combination of machine and human intelligence seems the best way to limit mistakes.

Personalisation becoming the new normal

Personalisation is important as we see different digital behaviour across different markets, dependent on certain characteristics. Although online research of Life and Health (L&H) products is high in most countries, we see penetration rates vary wildly. In Sweden 68% research the product online and above 30% buy online. In Spain, 52% use the internet to research, but the online penetration is just below 1 percent (Swiss Re Institute, 2020).

In the next 3-5 years, the digital insurance consumer will likely remain the millennials, with higher levels of income and education. It is important though to not assume homogeneity and develop solutions based on lazily assessed group characteristics. Personalisation is more important now than it ever has been. Beyond functionality and ease of access, emotions and personal growth are key drivers in consumption behaviour and like in any other group, there are a diverse set of expectations and desires amongst this group. Tailoring services and online buying journeys to the individual rather than the group is paramount; in the same way that offering life insurance immediately following a bereavement could be viewed as inappropriate, so too an offer of a social insurance be offensive to a staunch individualist. Certain benefits, although appealing on the surface to members of ‘the group’ may not work at a more nuanced level – a donation with every policy bought to an environmental charity will not appeal to every millennial. Hopefully, in this personalised version of the future, the benefits will be sufficiently tailored that people will see what is most relevant to them, not to whichever ‘group’ they may belong to.

Figure 2: Schematic showing of drivers of insurance purchasing behaviour model

As this group ages, we will see all age groups become attracted to digital insurance, with income and education less relevant factors in purchase decisions. As the aim of insurance is to improve resiliency, it is vital that insurance products adapt to become more accessible to vulnerable groups as soon as possible, not just as groups today age.

Vulnerable groups of people such as those in lower income brackets and those with chronic mental health conditions are often underinsured; if we focus solely on digital journeys for those with high income brackets and advanced education, we risk increasing inequality by leaving these groups further behind. If people in lower income groups are not represented and visible in the data we base our customer surveys on, we risk not adapting products to their needs. If they lack access to the technological devices necessary to start the online journey, they may never hear about the array of microinsurance products that have been developed.

Microinsurance was developed to service low income people and is used most frequently in developing countries. It still works as a risk pooling product, but coverage values (and therefore premiums) are lower than the usual insurance plans because the individual is insured only against specific perils. Insurers have realised that these markets have been historically underserved and with digitalisation lowering overhead costs, many are attempting to develop tailored, well-informed products that serve these markets. Having more data on both consumers and the risks they face means products are more likely be beneficial and increase the resiliency of both the individuals and the communities that need the protection.

The growth opportunity is an exciting area for research and development teams and could also go some way towards changing the perception of insurance as a more exclusive luxury that only the wealthy can afford.

Insurance as a service becoming commodity

Changing the perception of insurance is currently high on the priority list for the industry; it is one of the motivators behind the increasing pressure on the industry to digitise. No longer should insurance be seen as a burden some necessity forced on us by an archaic institution. It should instead be a service provider seamlessly integrated along with other services that manage our best interests; in theory, no one should be more invested in preventing bad things from happening to us than the insurance industry.

Figure 3: New capabilities to be established

In order for this paradigm shift, insurers will need to build a superior partner network, identifying the correct ecosystem partners and suppliers, especially in the technology and data space. Access to unique and comprehensive data sets for improved risk modelling will be contingent on the development of partnerships as assembling data layers in a way that creates unique insight is where the advantage will be over competitors. Other capabilities that need to be improved are the analytical expertise to make sense of all the data and the context, scalable underwriting expertise for more dynamic responses and developing deeper customer insights that truly put the customer at the centre of the ecosystem. Behavioural insights to personalise offerings will be essential.

In this version of the future, insurers will no longer rely on two touchpoints with their customers; these business models simply would not survive. In the same way that the majority of the population cannot imagine two google searches a year, so too will the insurer of the future not be able to recall a time that they interacted solely at the point of sale and point of claim/ renewal. Perhaps, even better for the industry, customers will be unable to recall that time too.


Bain & Company (2018) Customers know what they want. Are insurers listening

Kemp S. (2018) “Digital in 2018: world’s internet users pass the 4billion mark”, wearesocial.com, 30 Jan 2018, https://wearesocial.com/blog/2018/01/global-digital-report-2018

Swiss Re Institute (2020) Data driven insurance. Sigma 01/2020

Lomas N. (2014) “Amazon patents “anticipatory shipping” – to start sending stuff before you’ve bought it.”, techcrunch.com, 18 Jan 2014, https://techcrunch.com/2014/01/18/amazon-pre-ships/2014/01/18/amazon-pre-ships/

Stephen S. (2018) “What’s cooking at myntra? AI, fast fashion and ratatouille”, flipkart.com, 25 May 2018, https://stories.flipkart.com/myntra-rapid-ai-fashion-industry/

Swiss Re, op. cit.



Original post: https://www.swissre.com/institute/research/topics-and-risk-dialogues/digital-business-model-and-cyber-risk/data-and-machine-intelligence-insurance.html

Leave a Reply

Your email address will not be published. Required fields are marked *