We live in a time where digital technologies are promised to improve our lives by knowing us better than we perhaps know ourselves. Recommendation algorithms on Netflix and Spotify magically deliver us enjoyable series and music to consume. Targeted advertisement suggests us interesting products that we didn’t even know that we needed. Self-tracking tools, such as activity wristbands and smartwatches guide us to lead healthier lives and reward us for reaching the goals.
However, I would claim that we all have very different kinds of experiences, too. We get frustrated when we are suggested the same kind of music repeatedly. We are annoyed when an advertisement follows us around the web or when it seems to be based on crude stereotypes of age and gender. We get irritated when our self-tracking devices interrupt us at the wrong time or when our health data does not seem to match with our offline reality.
These examples show that algorithmic technologies are not just passive tools that we use but entities with which we interact and have relationships. The examples also highlight that emotions and commonly shared values are crucially at stake in these human-technology relations.
When successful, these relations feel pleasurable, or one may not pay attention to them at all as the technologies constitute a well-functioning background infrastructure. In other cases, the encounters may feel irritating and invasive, making people think about the limits of technological control and question whether such technologies can be trusted.
In my PhD dissertation I examine how these relational aspects are at stake in the making of a specific kind of digital technology: behavioural life insurance policies and their markets.
Two concepts are central to understand this technology: datafication and insurance
When we use algorithmic technologies, we leave behind digital traces of ourselves: our actions are logged, stored, sold and used for different purposes. This process of quantifying everyday actions and turning them into digital data is called datafication.
Datafication holds the promise that digital data can be turned into valuable information and used for improving services. The dominant idea is that the more data you gather, the better outcomes you will have.
Service providers are using data to optimize and to personalize their offerings. This means that the new forms of data are supposed to help businesses to see their processes and their customers in a more detailed way. By gaining a better view, service providers hope to operate more economically and to tailor their offerings according to the customers’ data. This data-driven operational model is often seen as transferrable across different fields.
Insurance is one industry experimenting with these promises of datafication.
Be it social insurance or private policies, insurance is an essential way of protecting oneself against sudden financial losses. For instance, in Finland, citizens are covered by social insurance which many supplement with private insurance policies, such as home, health and life insurance.
Usually, insurance is understood as a collective form of risk mitigation. When you buy an insurance policy, you join a group of people who share similar risks. These people form an insurance pool that collectively carries the chance of harm. Each member of the pool pays a premium. These premiums are used for paying indemnities for the policyholders who face a misfortune. Insurance is, thus, a bit like the three musketeers – all for one and one for all.
In recent years, insurers have been developing policies that include data-driven technologies, such as accelerometers in car insurance and activity wristbands and smartwatches in life insurance. These technologies monitor policyholders’ actions and provide access to their so-called behavioural data. This could potentially disrupt insurance in four different ways.
First, detailed data about policyholders could help insurers make more precise risk calculations. Second, the data could be used in personalizing the pricing of insurance. Third, self-tracking technologies, such as smartwatches, could be used in carrying out behavioural interventions and in incentivising quote unquote good behaviour. Fourth, the new technologies could help insurers present themselves as interesting, acting thus as tools for marketing and customer relationship management.
By using these technologies, insurers try to tap into the benefits of optimization and personalization. From their perspective, behavioural policies create a win-win situation. They generate cost-savings for the insurance companies and help people lead healthier and happier lives.
However, these new insurance products have been widely criticised in media and research. They have been seen as a way to subject people to increased surveillance and control. Furthermore, critics point out that, in an extreme case, they couldunravel the collective base of insurance and make it a luxury of low-risk and/or affluent people.
The discussion on the effects of datafication in insurance is clearly quite polarized with optimistic and pessimistic views of the potential outcomes. Both perspectives, however, fail to pay enough attention to the contextual and relational aspects of the new insurance technologies.
Insurance is a strictly regulated field with strong industry traditions. It is sensitive to reputation risks as insurance is essentially a promise to cover the costs if something bad happens. Although a central form of protection, insurance is neither interesting nor appealing to consumers, constituting it a ‘necessary evil’ as one of my research participants said. Hence, implementing data-driven technologies in insurance and making people trust the new operations might not be straightforward.
In my dissertation, I explore the developing of behavioural life insurance products as a contextual and a relational phenomenon. I do this by analysing both the insurance professionals and the policyholders’ practices.
I draw from three literatures that focus on the roles of different actors, relations and practices in the making of technologies and markets: sociology of insurance, data studies and sociology of markets. Inspired by these streams of research, I see that behavioural insurance (and its markets) cannot be accomplished without strong enough connections between the products and the customers. Hence, I examine both the insurers’ practices of developing behavioural life insurance policies and the policyholders’ experiences of using them.
The main questions of the thesis are:
- What kinds of ideas and aims guide the development of behavioural life insurance products and how do they play out in the insurance professionals’ practices?
- How do policyholders experience the new insurance technologies and data practices and why do they engage with them in specific ways?
- What kinds of relations do the new data practices create and how?
- How is behavioural life insurance (and its market) co-constituted with the new (data) relations between insurers and policyholders?
I analyse behavioural insurance by examining two Finnish life insurance products. These products use self-tracking technologies, such as smartwatches, to encourage policyholders engage in healthier lifestyles and to collect data on their physical activity. These behavioural life insurance policies were introduced to the market in the late 2010’s. They offer their customers wellbeing services and financial compensation in return of their data. For instance, if a policyholder reaches a certain activity level, they are granted a bonus to their insurance coverage. The products are experimental market openings. This means that they are still being developed while being available in the market.
I explored the new data-driven policies by conducting a fieldwork in two Finnish insurance companies. During the research collaboration, I interviewed 16 insurance professionals. Furthermore, I conducted 11 focus group discussions with insurance customers. In total 46 customers participated in the study. Finally, I collected document data and tested out the services myself, hence getting a sense of the products. I conducted the final analyses of the research by juxtaposing these varied empirical materials and by analysing them thematically.
The results of my study show that the use of digital data does not automatically optimize or personalize services. Instead, implementing data-driven technologies in insurance is quite difficult for three reasons. First, in the Finnish context, regulation sets strict boundaries for data collection and use. Second, the products are met with suspicion in the market, both within the insurance companies and among potential customers. And third, the use of the technologies entails a plenty of practical issues, such as lack and poor quality of data.
Because of these difficulties, at this stage of product development, Finnish insurers’ main goal is to use the new services to promote healthier lifestyles and closer customer relationships. They envision that self-tracking technologies could teach people about healthier habits and incentivise them to engage in them. Importantly, they hope that the new technologies could help insurance companies to become a part of customers’ everyday lives. The idea is to create a kind of data-driven partnership that promotes customer intimacy and loyalty.
Policyholders’ experiences of behavioural life insurance products are, however, ambivalent. On the one hand, they are interested in the new products and, at times, find them useful and enjoyable. On the other hand, they feel that they are perhaps not adequately rewarded for their data and have concerns about the scope of data tracking and use. Importantly, policyholders are uncertain which data is collected from them and for what purposes. This can affect the trust that people have on the new insurance operations.
Generally, customers accept behavioural policies if they are voluntary, as they are in the Finnish case. They, however, resent the idea of forced self-tracking and the feeling of not being able to decide for themselves. This becomes clear when policyholders describe their self-tracking practices. The data-driven technologies do not readily recognize their needs in changing life situations and interrupt them at the wrong time. Life-style interventions that customers experience as helpful in one situation might feel intrusive and annoying in another. This makes the disturbing sides of technological control visible and pushes people think whether their decisions are really theirs to make.
In the end, policyholders discard the technologies easily. The previously mentioned issues with trust and control play into this dynamic, but in most cases, policyholders simply forgot about the self-tracking practices. The policies suppose that customers should use the activity wristbands and other devices continuously. However, as people’s self-tracking practices are usually episodic, they drop the services when they no longer serve them.
My study shows that the relationships between customers and the new insurance policies are complex and that they break quite easily. It also highlights the importance of emotions and values in the making of these relationships.
Emotions are central to both insurance professionals’ design practices and customers’ everyday experiences with technologies. Insurers picture the wants, needs and moods of the customers as they try to find the right combination of services and operations to align with customers’ lives.
Correspondingly, customers describe their immediate relations with digital services in affective terms. Emotions alert when technologies become too intrusive, creepy or irrelevant; they notify when technologies are crossing important boundaries and threatening crucial values, such as trust and autonomy.
The new data-driven insurance technologies, thus, push people to consider the preconditions of these relationships. In this way, insurance transforms from an invisible and self-evident infrastructure to a visible actor, but perhaps not in the way that insurance professionals intended. With the introduction of digital technologies, the uncertainties related to datafication at large can spread to the field of insurance and undermine consumers’ trust in the industry. This can be challenging to insurance business which is sensitive to reputation risks.
My research shows that the techno-solutionist vision of an optimizing data logic that would be easily transferrable across fields does not work frictionlessly in insurance. Instead, the results highlight that technology development must consider the situational and relational issues at stake in a specific field, including regulatory and market contexts and customers’ experiences.
Furthermore, my research underscores the fact that the datafied arrangements are not self-evidently mutually beneficial. Subjecting oneself to continuous data-tracking does not necessarily lead to perfectly tailored and personalized services. In the case insurance customers, the benefits were often underwhelming and paired with the uncertainties related to the new data relations. For an individual consumer, the entirety of the data arrangements is likely to be difficult to understand. Hence, the imbalance of knowledge and power between consumers and service providers is clear.
Thinking about future developments, it is crucial to consider how socio-technical arrangements, such as behavioural life insurance policies, could become mutually respectful. Moreover, attention should be paid on how they could consider commonly shared values, such as autonomy and trust, in a satisfying way.
Respectful co-existence with data-driven technologies requires that the technologies do not tamper with us constantly but leave room for people to reflect upon, tinker and adjust their data relations. This calls for questioning the paradigm of optimization and personalization. Do personalized services truly improve the quality of our lives? How could we ensure that these technologies are not used for exploitative practices? And ultimately, how could people have more possibilities to define which kind of data-relations they engage in and with what conditions.
This is not to say that issues related to data relations could be solved by increasing individual control. Instead, the respect for people’s boundaries and the promotion of central values should be a collective concern, inscribed already in the design of technologies, regulation and the larger sociotechnical context.
In effort to create truly mutually beneficial digital services, lawgivers, industries and researchers should pay more attention to people’s everyday experiences with data-driven technologies and the different emotions and values that are at stake. The focus on emotions and commonly shared values would help both in developing better technologies and in supporting socially sustainable futures.
The full dissertation is available at: https://urn.fi/URN:ISBN:978-952-03-2627-2