Pesronalisation has become nearly a sine qua non in consumer marketing. It frequently receives the status of ‘basics’ or ‘a must’ in digital marketing, like knowing the alphabet, among academic and professional-business publications. However, achieving a high level of performance in personalisation of message content or product recommendations is no simple matter; its execution is highly demanding of data (e.g., volume, accuracy and timeliness) and in adaptability in order to reach high quality of personalised outcomes. Companies often reach only intermediary levels of personalisation, in part of the time and for some of their target consumer segments. Nonetheless, technological advancement and intensified market competition keep pushing companies towards more automated, customised and autonomous applications that extend or exceed the capabilities of personalisation.
At an elementary level, personalisation implies the ability to personalise the content of messages (e.g., e-mails) by embedding personally identifying information (e.g., addressing a customer by first name), and background demographic and behavioural information (e.g., age or city of residence, last product purchased, respectively). It can also mean matching differing versions of message content (text and images), with changed appeals, to different and small segments of prospective and current customers. But more than that, personalisation involves the ability to match product offerings or recommendations more closely to the personal needs, wants and preferences of customers, as these usually can be learned from their past behaviour (e.g., their previous purchases, but also their search enquires, ‘likes’ in social media, reviews and own recommendations to their peers on the company platform). Personalised recommendations may be generated by retrieving related products that are similar in characteristics to a product viewed or chosen by the customer, or furthermore, potentially relevant products viewed or chosen by other customers similar in personal characteristics to the focal customer.
It is important to distinguish here between personalisation of product offerings and customisation, particularly mass customisation. Conceptually, personalisation concerns the use of information to enable adjustments for better fitting consumer characteristics at the individual level, while customisation is the act of modification, applicable for instance to physical products and services, as well as digital online platforms, advertising messages, or pricing. But here comes the complication, because customisation can be done in different ways, based on different sources of personal information.
Perosnalised offerings or recommendations entail matching the offering of existing versions or models of products to a customer according to different types of his or her personal information as suggested above. In mass customisation, a company makes it possible for a consumer to obtain a product version or model specifically created (‘tailored’) in accordance with his or her stated preferences for product features (e.g., appearance, functionality). Hence, these approaches share the objective of making offers that better fit the preferences of individual consumers; yet, applications of personalisation as described earlier customise the recommendations of existing product items, whereas mass customisation entails personally customising the physical product item for the consumer-customer (performed at a large scale for many consumers). (Note: Users often can customise by themselves virtual aspects of digital interfaces of software or apps easily and instantly, but this is done post-acquisition.)
Companies may customise the content and appearance of websites or apps automatically according to the browsing behaviour of users through data-driven methods of personalisation. Media companies use these methods for personalising the placement of fitting ads and other promotions. In e-commerce or online store websites, users may adjust their search for product items by selecting criteria according to their attribute preferences (i.e., allowing consumers-users to directly apply their preferences but with regard to existing products in the inventory or catalogue of the retailer).
The practice of personalisation raises a number of issues, some of them sensitive or thorny. The issues surround mostly the types of information collected, how the information is collected and recorded, and how the information is utilised by companies. Additional issues concern subsequently the effect of personalisation on the relationships of consumers-customers with companies (e.g., manufacturers of national brands, service providers, retailers online and offline). Misuse of information for different forms or objects of customisation can lead to ethical misconduct, create embarrassing situations for consumers and companies, and damage customer trust to the extent of terminating relationships. Done fairly and effectively, on the other hand, it can produce important benefits for consumers (e.g., facilitating decision-making, making more satisfying choices, increase convenience and save time in performing various tasks), and thereby creating advantages for the companies.
Companies tackle often enough challenges and hurdles in obtaining and gathering information about consumers, even their own current customers. Companies can differ quite largely in their capabilities to collect detailed behavioural data about their customers; that is, not all companies are like banks, telecom and media companies, or online retailers — especially the global ones — that almost have the data at their fingertips from activities performed by their customers in different channels. Yet, even when data is more readily accessible, companies face problems with data accuracy, consistency and timeliness. Additionally, data on the ‘observed’ actions of consumers may be inadequate when information is lacking in background characteristics of consumers (e.g., demographics, lifestyles). Behavioural data may provide some indications for inferring interests or activities, but it would be restricted to product domains of the company. One option is to ask customers to fill-in information (degree of cooperation varies). Another option is to acquire third-party data on demographics, lifestyles, and behaviour (e.g., activities in social media networks). However, data from third-party market sources tend to be incomplete or include inaccuracies which may lead the client companies to make wrong assumptions or conclusions .
Data on buying behaviour, especially actual choices and purchases, are used primarily for revealing the preferences of customers-shoppers. This knowledge can be applied for prediction of future purchases of similar and complementary products or services, initiate product offerings, and fine-tune recommendations. Yet, predictions based on past behaviour have their limitations. Pre-purchase behaviour (e.g., search and browsing product pages) can be done for exploratory, learning and other enquiring purposes without real purchase intention — but it can provide a broader, eye-opening perspective on consumer behaviour beyond preferences to be applied. Past purchases can be misguiding when consumers do not necessarily wish to repeat their previous choice, because they want to try new and different things, and seek variety in their purchases and consumption (‘psychology of complication’ as phrased by Howard and Sheth, 1969).
Furthermore, there is little sense in sending a consumer an offering right after a purchase is made for the same or similar product, specially when it is normally used over a longer period of time (e.g., yet another mobile phone), a rather frequent practice. Even if the companies want to make a cross-sell of complementing products or related services, or up-selling upgrades (e.g., replace your model version 12 with version 13), it should allow time for the customer to learn how to make the best use of the product and become accommodated with it before he or she is ready to make the next step. Companies may supplement, cross-validate and enrich their knowledge on preferences by asking customers about their stated preferences through periodic questionnaires, but stated preferences may also be learned from the attribute criteria customers use when searching for products in the online store website or app platform.
Personalisation carries a price for consumers: In order to get more relevant, accurate, timely and useful messages, offerings and recommendations, one has to allow the collection of more data on his or her behaviour from devices in use, and provide more detailed information. Allowing for data to be collected and registered often does not require active consent of the consumer (though this is changing these days). Consumers who care more about the information about them divulged to the company (and third-party partners) can restrict some of the information shared, but then accept that they will get less personally relevant ads, offerings, or other content. One of the more sensitive issues, for example, is information on geo-locations of consumers, places where they may sometimes be less interested in getting personalised messages (e.g., in the dressing room) or others knowing their whereabouts in general — personalisation frequently infringes on privacy. This could engender a legitimate argument about justifiability of personalisation in different contexts: Sharing personal information, behavioural and other aspects, and forsaking some privacy, may be more justified in return of the practical benefits received from personalised product recommendations and concrete offerings than if personalisation is utilised for the purposes of marketing communications (advertising and other merely promotional messages).
Getting personalised content of various sorts does not necessarily and truly mean receiving personal treatment. Suppose, for instance, making a visit to a store where the shopkeeper-seller greets you by name, updates you on new product arrivals or makes suggestions based on knowing you, and shows how he or she is happy seeing you around in the store again — that is personal treatment. Personalisation could be seen as an attempt to substitute at a large scale, through intelligent technology, for the lack of personal interactions between retail sellers and customers, and revive some of the personal familiarity of retailers with their customers that was more common forty years ago and earlier. Still, this practice cannot genuinely compare with personal treatment, suggestions and advice. It might have helped if sellers in stores had a virtual assistant at hand that would aid them to access personal information on identified customers so as to provide genuine personal treatment that is useful and friendly.
Recommender systems, which rely on personalisation methods, leave the final choice if and what to buy from the products recommended to the consumers. Autonomous shopping systems take a step further and affirm a choice of products judged as most required and best fitting for the consumers-shoppers, and proceed to order them on their behalf (de Beliss and Johar, 2020 ). An autonomous shopping system may choose and order, for example, groceries that need to be replenished in the refrigerator or food items required according to a recipe (which may subsequently be mixed and cooked in an autonomous cooker machine); such a system may next be used for buying new clothing items like shirts, or perhaps making travel reservations for a vacation. It needs to be clarified, however, that the systems de Bellis and Johar refer seem to do more than making shopping decisions. Consumers can gain some important benefits from utilising an autonomous shopping system such as reducing tasks of human decision making, and thereof alleviating stressful conflicts of difficult trade-offs, and freeing-up time for engaging in more personally meaningful and pleasant activities. But there are also disadvantages in using such intelligent systems which could dissuade consumers from adopting them: for example, they may be reluctant to forgo their choice autonomy or forsake ensuing feelings of decision satisfaction.
After naming enablers of adoption, de Bellis and Johar discuss in depth barriers to consumer adoption and utilisation of autonomous shopping systems in four areas, where consumers may suffer a perceived loss of: (1) autonomy and control; (2) meaningful experiences; (3) individuality and identity; and (4) social connectedness. In each class of barriers they propose recommendations for ways companies may deal with and overcome those barriers. Firstly, people want to feel control and ownership of the outcomes of decisions they make — companies should present the system’s properties (particularly a physical device of the system) as more friendly and humanised (anthropomorphize), and to emphasise the contribution consumers can continue to make (framing). Even in daily tasks (e.g., at home) consumers may find meaning, such as when these experiences let them perceive themselves as productive and competent — this aspect should not be underestimated, and companies may highlight the other meaningful activities consumers can pursue while an autonomous system is in use.
Of particular interest are the suggestions of de Beliss and Johar regarding the need of consumers to protect their individuality and uniqueness of their needs and preferences. In order to help consumers maintain their feelings of uniqueness, they propose employing mass customisation and personalisation. Mass customisation, reliant on preferences explicitly stated by consumers, would be employed for letting consumers take part in designing the physical device of an autonomous system, including exterior features (e.g., related to visual design like colour and size) and interior features (e.g., related to functionality such as type of integrated tools). Personalisation, based on application of customer data and its leveraging, would be employed in the context of services the system provides (e.g., increase interest, foster creativity which may lead to surprising but explicable and useful outcomes, solidify relationship with the brand of the system).
- Personalisation and customisation are entangled much of the time, but there are applications of them where they diverge. Customisation is enabled largely by applying the information produced by methods of personalisation on the basis of formerly collected and analysed customer data. Yet, mass customisation is a special form of customisation that relies on explicitly stated preferences; the use of attribute criteria for screening products in e-stores is a form of personalisation that actually allows shoppers to apply their stated preferences but is not like mass customisation; there is also a form of customisation where consumers-users may apply their personal preferences as in ‘self-service’ but is done after acquiring the product (i.e., by modifying features of appearance and functionality of software-type products — the company may gather information about these adaptations).
There is undeniable promise and potential, often proven, for benefits to consumers in implementations of personalisation. But the value of personalised applications to consumers truly depends on how they are executed and delivered: effectively, accurately, fairly and non-intrusively. While in some cases the personalised content should be prompted timely, to be relevant and practical, the time and place may not always be appropriate for consumers. It is also in the responsibility and interest of consumers to be aware of the benefits yet also of the implications (non-monetary costs) of taking advantage of personalised applications. Personalisation is the ground for enacting additional functions and capabilities and is definitely not the ultimate goal or outcome. Customisation in different forms, robotic and autonomous systems, and other artificially intelligent tools surround personalisation or are adjacent to it, reliant on its results or enhance and extend them in new ways, and the developments in this field keep on rolling.
Ron Ventura, Ph.D. (Marketing)
 Drawing on a discussion as part of Adobe’s Chat series: Debunk (with Magento), “Myth: Customers Always Want More Personalisation“, November 2020 (last viewed January 2022), (registration may be required), Ben Davis (Econsultancy, moderator), Scott Morrison (The Boom), Brian Green (Adobe), Peter Weinberg (LinkedIn — The B2B Institute).
 Ibid. 1, an argument raised by Peter Weinberg.
 Autonomous Shopping Systems: Identifying and Overcoming Barriers to Consumer Adoption; Emanuel de Bellis and Gita Venkataramani Johar, 2020; Journal of Retailing, 96 (1), pp. 74-87