By Srivathsan Karanai Margan
The problem with evil is that in real life,
it is not necessarily ugly.
It can look really beautiful.
—Yuval Noah Harari
Personalization, which is now a part of the corporate zeitgeist, is the culmination of several decades of customer-focused and -centric initiatives. Personalization and its associated term, customization, are not new, but rather old concepts that only a privileged class of customers—high-net-worth individuals—could access. It turns out that notwithstanding net worth or any other categorization of their social standing, every customer has a deep desire of not being seen as a faceless, anonymous footfall count, but as someone special—and treated accordingly.
Global middle-class bulge, easy mobility, and proliferation of connecting technologies have empowered customers like never before—in effect democratizing customers’ desire for the celebrated exclusivity and causing a corporate mind-shift to deliver it. New technologies, while providing new opportunities for customers to voluntarily express details regarding their state-of-being voluntarily, are also creating new sources to capture it without customers’ awareness. The data deluge continuously generated by connected devices and the remarkable progress made to transmit, store, and analyze it are ushering in new ways in which a customer-centric company can engage with their customers.
The retail industry is an early adopter of the data-driven customer-centric approach to targeting individual customers in a highly differentiated way instead of focusing on undifferentiated mass markets. The industry has proved that such personalization increases sales conversion rate, customer wallet share, retention, and engagement, and decreases acquisition cost. Attracted by the benefits realized by pioneers and at the same time pressurized to meet the increasing demands of customers, nearly every customer-facing industry is joining the personalization bandwagon. As the drive to personalize becomes a contagion, even industries such as insurance that traditionally practiced the “build a better mousetrap” philosophy are joining the race. An intense corporate adrenaline rush is seen, rendering atomized products and services to a segment-of-one in a need-, preference-, and context-sensitive way for continuously staging personally curated experiences. The objectives are to get closer to customers, make their presence felt, and get a share of customers’ minds—and wallets.
Eager to “wow” their customers, companies could inadvertently personalize things in a way that customers feel to be intrusive. When a customer considers that personalization is crossing the “creepy” line, even a supposedly sincere attempt by the company could boomerang to bring bad repute and worse—customer dissociation. This article is an attempt to trace the indicators of such an invisible “creepy” line.
Personalization, the In Thing
Personalization, in very basic terms, is a company being aware of the needs of a customer and meeting them to produce a unique product, service, or experience. While customization also tends to create the same result, the difference lies in how the company engages with the customer to achieve it. For customization, a basic product or service is selected, which a customer modifies to suit his or her requirements by actively choosing from a given set of alterable specifications. On the other hand, in personalization, the same end result is achieved with the targeted customer remaining passive, making no effort, and the company proactively surmising the needs of the customer from the data it possesses about them. Based on the expansive nature of the data and sophistication of algorithms, the intelligence could spread from predicting the customers’ met to unmet and even unexpressed needs, as well as providing intricate details about when, where, and how they would prefer to have them served.
Irrespective of demographic segmentation, customers of today’s digitized world—by means of every tap on smartphones, click traits, wearables, sensors, and other connecting devices—leave behind an ever-expanding digital trail of who they are, where they are, and what they do. Along with this, behavioral metadata regarding when and how it was done gets shared. Companies leverage these erstwhile untapped data and less conventional data along with other first-party, second-party, and third-party data to associate every customer with a digital persona or identity. Intelligent algorithms deliver deeper insights on why a certain activity was done by a specific customer, at a given place and time, in a specific way and what that effectively means. All of this “under the hood” activity is used to pleasantly surprise customers with a personalized offering even before they look for it, thus giving them a feeling that the company knows them, thinks about them, and cares for them.
To receive these coveted personalized offerings, customers will have to accede to share with the companies their living profile seamlessly. For this, customers overcome privacy paradox, a tug-of-war between the desires for personalized content and the natural instinct to protect their personal information before deciding to share their data. After deciding to share, customers expect the companies not to treat them as a faceless representative of a generic market segment but provide services similar to what they get from a neighborhood corner shop.
The primary focal point of personalization for industries is marketing. Companies face a tricky challenge of shrinking attention spans of customers to overcome and attract customers in a digital world. Studies indicate that the average human’s attention span, which was 12 seconds in 2008, had reduced to 8 seconds in 2013. It is also said that due to the information overload, people are becoming more selective about what they give attention to. Whatever the cause, companies get a very short time window to navigate through the cognitive, affective, and conative stages of consumer behavior. This urgency has created a sense of desperation for companies to prove to customers that they know their needs perfectly and have the perfect fit and exclusive offer for them. After all, a good personalization approach executed accurately could pave the way for new opportunities, create customer loyalty, and increase revenue.
It is common wisdom that companies, by all means, aim to achieve good personalization, but still, mistakes happen and result in bad personalization. I identify bad personalization of three types:
- Spurious: Personalization that occurs because of poor customer profiling and improper solution matching, thus resulting in a mismatch. If luck favors, it could result in an acceptable generic offer otherwise, the customer dismisses it as spam.
- Insensitive: Personalization that occurs when sensitive variables (age, gender, ethnicity, religious affiliations, sexual orientation, mortality, morbidity, etc.) are manipulated for deriving inappropriate or controversial inferences. These result in activities that can be deemed devious and/or discriminatory. This could also be branded as evil personalization for being insensitive and could trigger serious repercussions.
- Intrusive: Personalization that occurs when a company overdoes certain things or oversteps into private spaces that customers consider to be sacrosanct. Customers feel uneasy and perceive that their privacy is being intruded.
Whatever the category of bad personalization, it detriments the company with bad publicity, customer resentment, dissociation, and possibly lawsuits.
An Eerie Feeling of Intrusion
A paradox regarding any bad personalization termed as intrusive is that the findings are not wrong, but they are just more accurate than what is tolerable or what is expected. Considering the repercussions, seldom would a company wantonly commit an intrusive personalization. However, the intricate challenge is that the line that differentiates between a spot-on good personalization and a creepy intrusion is invisible. It is subjective and varies depending on the perception of every individual customer. It requires a very delicate balance and sensitive understanding to distinguish, because what could potentially stage a “wow” experience for one customer could make another customer cringe.
It is impractical to have a standardized and objective approach to draw a visible line to denote the space beyond which any personalization attempt would be termed as intrusive. However, certain indirect methods could provide broader guidelines and raise red flags to identify what could turn out to be intrusive. For example, the so-called Johari window—a model created by Joseph Luft and Harrington Ingram in 1955 to help people with self-awareness, personal development, group development, and understanding relationships—could be used to understand personalization better.
The Johari window consists of four quadrants, such as:
- Open area: What is known to a person about him/herself and is also known to others
- Blind area: What is unknown to the person about him/herself, but known to others
- Hidden area: What is known to a person about him/herself, but unknown to others
- Unknown area: What is unknown to a person about him/herself and is also unknown to others
The “open area,” also known as “area of free activity,” is the most effective and productive region for collaborative work. The main purpose of the tool is to develop this region for every person. The “open area” could be increased by reducing—the “blind area” by asking for and receiving feedback; the “hidden area” through the process of telling that includes self-disclosure and exposure; and the “unknown area” through collective and shared discovery. The expansion of the “open area” happens through a collaboration between a person and others.
From the angle of personalization, “others” in the Johari model could be replaced with a company that embarks on the personalization initiative. The typical ask and tell methods suggested by the model would be synonymous with the active participation of a customer for customizing the offering. Owing to the active involvement of the customer to create a best-fit offering, customization is never considered to be intrusive. Even where the limitation of choices provided could result in a bad customization, it is not perceived to be intrusive. When the best-fit offerings are created proactively by the company, without any active involvement of the customer, it would be synonymous with personalization.
If a personalization approach is made within the “open area” and the company avoids nefarious personalization, it could result in a noncontroversial personalization. However, in this instance, the customer has shared personal data and is aware that the company knows them, and hence the personalization becomes an expected one. When such personalization is done, it may not trigger any surprise, but it could result in dissatisfaction if not done well. To create customer thrill by catering to their unexpressed needs, companies will have to attempt personalization in the other three regions. The success of personalization in those regions depends on factors such as being accurate to the right degree, being sensitive, and the customer benefiting from the offering. Any personalization mistakes in these three regions run the risk of being perceived as generic, spurious, or intrusive.
For customers, personalization from a nearby corner shop comes from personal bonding, a human touch, interpersonal relationships, and longstanding relationships. To achieve such personalization in a corporate and digital environment would require technological devices to imitate the human touch of a nearby corner shop. This could either trigger an angry response about a faking system or an unsettling perception of the uncanny valley.
Some of the reasons I believe could make customers deem a personalization attempt intrusive are:
- Unknown source of personalization: Personalization is made by a company with which the customer does not have an existing relationship
- Unknown source of data: Personalization is made from data that the customer never shared knowingly with the company, or they are not aware that the company has the information
- From stolen data: Customer shared some details of their private self with one company, but find personalization reflected in the offering of a different company
- Consciously hidden area: Personalization is made in areas concerning private self that the customer has consciously suppressed from revealing
- Information asymmetry: Personalization is done at a very granular level to the extent that it gives an uneasy feeling of excessive information asymmetry
- Being surveilled: Personalization is being solely made on digital shadows—the unconscious digital trails a customer leaves—thus creating uneasiness of being surveilled and put under a microscope
- Too specific to me: Personalization seems to be very specific to the individual customer and does not have any appearance that it could be broadly applicable to customers with a similar profile
- Chained liberty: Personalization approaches that limit the choices available to customers and make them feel that their basic freedom is being constrained
- Not done privately: Personalization that is supposed to be very personal to the customer is exposed or communicated to others
- Unwanted: Personalization is done in a context where the customer does not want it to be done
- Unexpected: Personalization is done in an area where the customer does not expect it
- Undecipherable: Customers unable to comprehend the human logic behind the personalization approach
- Being used: Customer feels the personalization offer serves to benefit the company
As of now, the competitive edge for the companies that embark on personalization journeys is determined by the volumes and variety of the new data sources that they have access to and the capability to stage a best-fit personalization. To customers who were deprived of any such exclusive care earlier, even a basic personalization done in “open area” would be quite exciting. However, as this space gets competitive, “open area” personalization loses its charm and companies compete to expand into the other three regions of the Johari window. Over the course of time, the access to data becomes equal to all the companies, and the competitive edge is determined by the sophistication of the intelligent algorithms in finding new data associations.
Insurers Learning to Dance
Personalization is still in the early stages of adoption even in the retail industry, and challenges exist while catering to large market sizes. A highly regulated industry like insurance invariably faces a lot of structural, actuarial, legal, and technological challenges to overcome while embarking on personalization initiatives. The edifice of insurance is built on deep-rooted generalization principles and has progressed for over three centuries on push marketing, fear merchandise, and mandated purchase. With limited customer service touchpoints, the industry has followed a templated model for generalized services and customer indifference. The industry outsourced a major portion of the task to personalize to the intermediaries who established a longstanding relationship of trust with the customers.
I think the information technology application landscape of most of the traditional insurance companies is overcrowded with legacy systems, which pose a big challenge for them to experiment with anything new. The industry is embracing the data-driven personalization craze by either setting up their own legacy-free startups or establishing partnerships with insurtech companies. The traditional insurance discourse is being redefined, and insurance behemoths are learning to dance.
To start with, insurers are experimenting with the tested personalization range of offerings from other industries such as targeted marketing based on propensity to buy, contextual marketing, and persistency assessment. While performing personalization in these areas, insurers will face the same personalization predicaments that other industries face. Insurers are creating industry-specific personalization offerings such as targeted marketing with pre-underwritten products, usage-based insurance, on-demand insurance, peer-to-peer coverage, and dynamic pricing. Insurers are atomizing the products and services to offer them to customers, exactly matching their needs depending on the context, location, period, and method. Though the end product or service is exactly what the customer wants, most of these initiatives provide customers with a wide range of options to select from. Hence most of these initiatives would only satisfy the definition of customization and not that of personalization.
Insurers are embracing risk prevention and happy-state extension-oriented business models for risk prevention and protection. The status of an insured or insured object is continuously monitored through sensors, wearables, or other partners. Insurers prescribe certain risk behaviors to customers, engage with them to ensure strict adherence, and intervene on identifying any aberrations. The risk categorization and pricing models in insurance are based on the data pertaining to every individual customer obtained in real time rather than representational and empirical datasets. The definition of personalization is revisited from the core insurance perspective as one where customers could potentially manage the risk, reduce losses, and in effect control their premiums. In this model, the concern and care extended by the insurers and their ecosystem partners during pre-risk, at-risk, and post-risk conditions could be deemed textbook examples for good personalization.
The reciprocal effect is a challenge peculiar to personalization in the insurance industry. The happy-state extension models are based on rewarding a positive risk mitigation behavior and penalizing a negative one. When a risk monitoring system observes an intentional or unintentional blip in the risk mitigation behavior of the customer, insurers could enforce increase of premium, imposition of other restriction, denial of claim, or denial of insurance. The customer’s perception about personalization could drastically change to being seen as intrusive when insurers initiate corrective measures in response to the increased risk, and effects turn to disadvantageous to the customer.
And the Balancing Act
Connecting technologies have changed the landscape of data-sharing forever. The data deluge is likely to further swell as the digital life becomes hyperconnected with the wider adoption of technological instances such as implantables, smart cities, digital twin, connected vehicles, connected homes, connected workplaces, and so on. Notwithstanding the privacy calculus that any customer performs by weighing the risks and benefits of data-sharing, the default digital design is going to be skewed toward sharing large amounts of data. In such a scenario, it will be very delicate to perform the balancing act to stay within personalization range and not to veer into the territory of intrusion.
To address the data privacy concerns of customers, regulators across the world are enacting regulations to seek explicit consent from customers to collect, store, process, and use their data. The consent also covers permission to obtain data from third parties and share data with other partners. The conditions regarding soliciting additional information from third parties, including data brokers, is a moot point today as privacy regulations are relatively subdued. Nevertheless, in due course, the first-party, second-party and declared-data flow could be humongous to churn, thus making the need for unauthorized third-party data irrelevant.
Customers keep interacting with several stakeholders for their needs and would provide multiple companies with separate consents to share different types of data for varied purposes. There could be association among these companies for sharing data, which they churn to find new data associations. To decipher the source and interoperability from the overwhelming network of stakeholders could be a cognitive overload for customers and could give them a feeling of being surveilled.
The evolution of platforms and ecosystems—wherein companies that produce different products and services come together to build a landscape for collaborative existence—could throw at customers a challenge of a different kind. With the ecosystem operating as a single entity, when a customer shares data with one company in the ecosystem, he or she may not realize that it would get mutualized with other stakeholders within the ecosystem. A partner company could interweave this with the data that it may already possess about the customer to produce an offering that could invoke concerns of invasion of privacy.
Increased access and wider availability of genetic information and its interpretation could entirely recast the scope of personalization. Any individual who voluntarily shares her or his genetic information involuntarily provides private information pertaining to other family members. With the volume, variety, and velocity of the data deluge continuously increasing, personalization is becoming an impersonal activity driven by intelligent algorithms. In performing a balancing act in such an algorithmic landscape, the burden of choice to decide on to whom, what, when, where, and how anything should be personalized without crossing the line of intrusion solely rests with the companies that embark on the journey. To avoid crossing the creepy line, the simplest rule to follow is to focus on being more sensitive than being more accurate. To begin with, it is suggested to ensure the appropriateness of the method of data collection and have a “human-on-the-loop” to validate the intrusiveness factor of the personalization initiative.
Companies will be wise to adhere to the general principle of respect for the dignity of customers; the best course of action would be to initiate a dialogue with the customer to expand the “open area” region and increase the amount of declared data, especially regarding the safe personalization spaces. In case the list of all areas in which customers would want personalization is too vast and difficult to capture, it is suggested to capture the areas where they would not want it. Companies should realize that the best personalization is not the one where “wow” experiences are staged nonstop. Instead, it is something that is so smoothly interwoven in the lives of the customers to the extent that it is not even visible. The presence of it may not even be recognized or appreciated, but its absence intensely missed.
SRIVATHSAN KARANAI MARGAN works as an insurance domain consultant at Tata Consultancy Services Ltd.
 Communication Theory. “The Johari Window Model.” Accessed at https://www.communicationtheory.org/the-johari-window-model/.
 Earnix and Insurance Innovators. 2018. “The Age of Insurance Personalization.”
 Jebbit. 2018. “Overcoming the Personalization vs. Privacy Paradox.”
 Mcspadden, Kevin. 2015. Time Magazine. “You Now Have a Shorter Attention Span Than a Goldfish.”
 Prezi. 2018. “2018 State of Attention.”
 Steiner G. A and Lavidge R. J. “Hierarchy of Effects Model.” Accessed at https://bbamantra.com/response-hierarchy-models/.