12
Data Marketing for Customer Intimacy
We can probably all remember an excellent salesperson. A person who recognized you because it was not the first time you entered this shop, who called you by your first name or who, on the contrary, remembered that you prefer a little distance. A person who advised you according to your tastes and preferences, according to your latest purchases or a person who wished you happy birthday. This concept of “customer intimacy” is more difficult to bring to life in the digital sphere, which is by nature colder and more distant. However, the collection and use of data, generated by both connected consumers and businesses, make this approach realistic. These data can be used in a customer knowledge approach to develop, on a large scale, a service-oriented logic based on relevance and personalization.
As early as 1993, Roland Rust [RUS 93] formalized a paradigm shift faced by many organizations. This paradigm shift could be summarized by moving from a so-called “product” scenario to a so-called “service” scenario. The product scenario consists of companies offering their customers a relatively standardized offer and in observing in return a flow of aggregated information that gives an indication of the customer’s response to this offer and therefore of performance. The service scenario, on the other hand, reflects a more interactive plan between the company and its customers, which reflects the idea of a company that develops an active and individualized way of listening to its customers in order to design its offer and continuously adapt it to meet multiple and evolving customer needs.
A few years later, Christian Grönroos [GRO 97], explained this paradigm shift in marketing: from a marketing mix paradigm (the four Ps) to a relationship marketing paradigm. The latter is based on the construction and management of relationships between the company and its customers. This calls for the development of a long-term vision where value comes less from a series of anonymous transactions than from the development of customer value over time through the satisfaction, trust and commitment that the latter develops toward the company.
More than 20 years later, Samsung and LG compete in ingenuity in this logic of switching from a product paradigm to a service paradigm. As their recent announcements at the CES in Las Vegas demonstrate, their connected refrigerators have a resolutely service-oriented value proposition:
- – they recognize the different family members by their voices;
- – they integrate home automation systems and now control lighting, heating and surveillance cameras in the home;
- – they offer recipes adapted to each member of the household’s tastes, adapt to diets and product expiry dates;
- – they prepare shopping lists and integrate with Alexa, Amazon’s voice assistance system.
In short, this is a very characteristic case of a value proposition where the product however (iconic) becomes only one component of a broader service offer. Better still, these systems, equipped with artificial intelligence, progress every day, with each interaction, and thus gain in perceived relevance. Far from being anecdotal, the case of connected refrigerators is indicative of the major acceleration in recent years of the adoption by companies of customer orientation. Three vectors contribute to this acceleration: the evolution of consumer behavior through digital technology, the large-scale generation of data and major technological advances in the analysis capacity of these large volumes of data.
This move toward a greater service or customer orientation implies a customer knowledge approach. This is made possible by increasing access to data that allow companies to detect weak signals, intentions, individualize their discourses and adapt their offer in real time.
While data collection has long been restricted to companies that had direct access to their customers (e.g. retailers, banks or airlines, pioneers in setting up loyalty programs), new players, historically more distant from their customers, now have the ability to collect and leverage customer data. The paper published by IDC, “Data Age 2025”1, reveals the extent of the phenomenon. The volume of data generated worldwide in 2016 was 16 zettabytes2 and this volume could reach 163 zettabytes by 2025. This exponential growth is mainly driven by the penetration of connected objects (Internet of Things [IoT]), intelligent sensors and connected cars.
Figure 12.1. Evolution of the volume of data created since 2010 and projection to 2025
(source IDC “Data Age 2025”, April 2017)
Access to these data is even becoming a battle in which two types of players emerge: the Data Rich and Data Poor. It is not, moreover, the business line that determines the wealth of data: Uber and taxi companies have the same activity but have an incomparable data asset. In this battle for data, the GAFA have, through their audience pool, a unique position that redistributes power balance maps, such as the Amazon Dash buttons that allow many advertisers to market their products, or Google Home, which offers new applications every week, including those of many brands that see the opportunity to have a conversation with their customers.
So where and how do we collect these data? If new actors, whatever their position, have an unprecedented opportunity to collect data, how can these levers be identified and exploited? These questions are all the more essential as the volume of data collected is coupled with an unprecedented operating capacity. For example, the computational power required by a Google search at the moment matches that of the entire 11-year Apollo space program.
12.1. Multiple customer data sources
Gradually, all behaviors generate data. This is explained, on the one hand, by the increasing time spent by individuals on digital media where most interactions generate data and, on the other hand, by the fact that digital media is entering new spheres: homes (home automation and voice assistants are only the first steps), physical points of sale (via WiFi or Beacon technologies3), cars (which become connected before they become autonomous) and media in the broad sense (radio podcasts, replay/VOD television). In short, two levers coexist:
- – the first one is the rapid penetration of digital technology. The adoption of smartphones demonstrates this. In volume, we are halfway to connecting the entire population with these devices;
- – the second lever is about usage and adoption of such digital technologies: opportunities for interaction with technologies are more numerous and more frequent.
Let us look at these developments through the prism of customer journeys. Let us take the example of a consumer who wants to buy a connected TV.
Consumers are now beginning their purchasing journey through the consumption of Web content (nearly 80% of consumers consult the Internet before making a purchase according to FEVAD [French Federation of e-commerce and distance selling] figures). This beginning of the online journey gives rise to multiple interactions: exposure and reactions to targeted advertising campaigns, keyword searches, visits and behavior on brand and retailer websites offering connected TVs, subscriptions to newsletters, etc. The purchase, whether online or at the point of sale, in turn generates transactional data. Omnichannel is the simultaneous use of several channels: the use of the smartphone in store to check prices in real time is a typical example. The rest of the journey leads to the emergence of media consumption (increasingly in digital form, via VOD, radio, online platforms) and interactions on social networks (sharing experiences, online opinions, word of mouth) and finally the continuous use of a product as a service generates more and more data via the IoT.
It is therefore in light of this proliferation of touchpoints, at these different moments of a “typical” journey, that opportunities for data collection emerge. Throughout this journey, data hubs are being set up, some of them are already mature while others are emerging.
Figure 12.2. The “Data Journey”, an example of data generated throughout a customer journey
(source: Numberly)
12.2. The different customer data hubs
If we take the case of a customer wishing to acquire a connected television, here are, in order of appearance, the hubs or “groups” of potentially generated data.
- – The Web content consumption hub, and more generally behavioral data
The first stages of this connected journey constitute a Web content consumption hub, often represented by the prism of paid, owned and earned media of brands or retailers. It is a mature Hub, which covers the scope of the Data Management Platforms (DMP), these behavioral databases resulting from digital interactions, often anonymous and traceable in particular through cookies.
- – The transactional data Hub
Transactional data are a second essential Hub, whether online or offline via relational programs that allow a transaction to be linked to a customer ID. These data are critical in measuring the effectiveness of early-stage actions, which aim to generate conversions.
- – The postpurchase data Hub
The postpurchase stages of the journey are, more and more, key collection moments. Indeed, new media consumption patterns are contributing to the proliferation of the number of digital touchpoints that can be activated by brands and retailers. A recent OpinionWay survey indicated that 40% of French people had already listened to a radio podcast [AUD 17]. The trend is the same for television content, where replay, VOD and connected TV make it possible to characterize audiences more and more accurately. As such, Amazon’s investments in its Prime Video service are significant. The media, consumed via digital services, is becoming more than ever a very rich source of characterized information. The rise of programmatic media buying, i.e. real-time and individual bidding, thus allows the encounter and convergence between media and data. This media Hub, consumed at home or on the move, is emergent.
- – The social interactions Hub
The next Hub is that of social interactions. More mature, it brings together platforms for sharing opinions, rating, recommendations and influence and is a major source of word of mouth.
- – The usage data Hub
The last Hub of this journey is emerging but highly strategic. Indeed, the IoT is a new opportunity for brands and retailers alike because it generates new and particularly valuable data: usage data, context and recurrence of the use of products and services. This continuous flow of information, relating to the way in which customers use a product or service, could allow the drastic acceleration of service-oriented and interactive logic where the offer becomes totally adaptable, individualized and therefore scalable. Samsung and LG’s approaches to their connected refrigerators, similar to Axa’s with You Drive, are the first steps in the large-scale exploitation of data from the IoT.
In the end, these different data hubs are, in addition to the colder or more static information (sociodemographic, for example), considerably enriching customer knowledge by helping to reconstruct an increasingly exhaustive view of consumer behavior.
| Examples | Provenance | |
| Sociodemographic data |
|
|
| Transactional data |
|
|
| Behavioral data |
|
|
| Usage data |
|
|
| Social interaction data |
|
|
12.3. The difficult consolidation of customer data
The prism of the customer journey thus seems interesting in order to study this data marketing ecosystem. In particular, the types of data generated along such paths are very heterogeneous. Transactional and personal data, collected and used for several years, are enriched by data often referred to as behavioral and anonymous. Indeed, undoubtedly, many interactions throughout these “data journeys” take place without a customer being identified by name. Such data typically includes cookies placed on a web browser, geolocation data or an anonymous platform (Twitter, Facebook etc.) and device.
In other words, the spectrum of the so-called “PIT” (personally identifiable information) data is supplemented by the so-called “non-PIT’ (non-personally identifiable information) data. These allow weak signals to be picked up and intentions to be detected. They also make it possible to increase customer addressability by reconciling several identifiers around the same customer: an essential dimension for real-time activation through relational channels. Data marketing is part of the migration from a model based on the depth of data from the PII world (highly characterized, accurate and nominative data) to a model based on the volume of data from the non-PII world (volume, varied, real-time and nonnominative data) that has contributed to the growth of the term “Big Data” in relationship marketing.
Figure 12.3. Illustration of the multiplication, over time, of customer data to be reconciled
12.4. The intersection of media and data to serve customer strategy
Data collection is not an end in itself. The major challenge of data marketing is to develop a company that actively listens to its customers, a company that learns from the continuous feedback it receives, a company that can identify weak signals and provide a relevant response in real time and finally a company that can recognize (in every sense of the word) its customers during key interactions.
The touchpoints activated by companies are closer to the customers’ life journeys. The challenge for advertisers therefore becomes “to be where the customer is”, on their journey, in a flow. This translates into a significant trend in relational practices that could be summarized by the idea of “being present where the customer consumes content”. As such, the changes in content consumption patterns speak for themselves. A comScore study [COM 18] shows that in 2012, American consumers spent an average of 28 minutes more watching television than using other digital media every day. Forecasts for 2018 revealed that American consumers were expected to spend 126 minutes more on digital media than on television.
Figure 12.4. Comparative evolution of the time spent per day in front of the TV and digital devices from 2012 to 2018 with regard to the American market
(source: comScore)
This trend of strong digital growth is mainly driven by mobile phones, which capture most of the extra time spent on digital media.
Figure 12.5. Comparative evolution of the time spent per day in front of digital devices (laptop, smartphone and tablet) from 2012 to 2018 with regard to the American market (source: comScore). For a color version of this figure, see www.iste.co.uk/ngoala/augmented.zip
The first issue in data marketing operations is probably the convergence between two universes that were relatively distinct until recently: the media universe on the one hand and the Customer Relationship Management (CRM) universe on the other.
The rise of programmatic media buying allows advertisers to activate new one-to-one channels such as displays, but also video, audio (radio) and smart television. To be relevant, one-to-one activation requires customer knowledge: it is the meeting of media and data that is played out.
Customer knowledge from CRM, which is often older and deeper, could therefore find a new source of value: its ability to interconnect with behavioral data. The reconciliation of PII and non-PII data should be a major area of interest for customer relations in the coming years. These behavioral data, collected by advertisers, is called “first-party” data. These first-party data constitute an essential new asset for advertisers in this logic of a programmatic CRM, very much in line with the idea of being present where the customer consumes content. Recent figures from Pivotal Research according to Nielsen DCR4 indicate that the Google ecosystem (Google, Youtube, Waze, etc.) represents more than 27% of the time spent on digital media by Americans, while the Facebook ecosystem (Facebook, Messenger, Instagram and WhatsApp) represents just over 16% of the time spent on digital media. The advertisers’ first-party data could be considered as a particularly strategic new asset in the “data battle”.
This data convergence is at the heart of multiple service operations for customers.
Among these operations, techniques for customizing the relationship are gaining momentum. This is a historical pillar of customer relations that has undergone major recent changes due to the digitization of our economies. The personalization, because of the reconciliation of CRM and first-party data, is done in real time, in an automated and large-scale way. One example is dynamic content optimization (DCO) techniques, which, on programmatic channels, allow the message to be personalized according to:
- – the detected device of exposure: by democratizing formats in “responsive design” for mobile reading, but also by detecting the operating system (iOS or Android) to automatically redirect Internet users to the appropriate Store to download a mobile application;
- – geolocation: by indicating the nearest point of sale in a drive-to-store logic, or by creating digital catchment areas, allowing for the targeting of only Internet users passing through this area;
- – CRM data available on the target: Coca-Cola is thus able to recognize certain customers who browse the Web to wish them a happy birthday with a bottle personalized with their first name;
- – real-time weather forecasts or prices charged by competitors because of third-party data available to advertisers.
The same relational campaign can thus have, in real time, an infinite number of variations depending on the Internet user’s context, their profile in the advertiser’s databases and their history (purchase, visit or reaction) in the relationship.
Figure 12.6. Example of a relational programmatic campaign for a customer named Tom on his birthday
Such approaches to differentiated customer knowledge and activation are accessible to an increasing number of advertisers, some of whom have historically been further away from their customers. Manufacturers or industrial advertisers, who do not distribute their products directly (B-to-B-to-C), now have, via first-party data collection, a new gateway to these relational approaches. By constructing this asset, resulting from the interactions of Internet users with their digital ecosystem (paid, owned, earned), these advertisers are able to evaluate, for example, the engagement of Internet users on their brand sites. They can determine an engagement score by the quality of visitors (time spent, recurrence, number of brand sites visited, number of product pages visited) and then set up a differentiated activation of their strategic targets because of programmatic media: the most engaged Internet users via VIP treatments, recognition; moderately engaged Internet users via more transactional messages, etc.
Table 12.2. Example of website visitor engagement scoring based on non-PII data. For a color version of this table, see www.iste.co.uk/ngoala/augmented.zip
12.5. Leveraging data: market research in the era of customer data
While they bring real benefits to customers, emerging language and image recognition technologies (chatbots, voicebots and other robots based on artificial intelligence, virtual and augmented realities) are also an opportunity for brands to collect new types of data. The collection of this type of data will intensify in the coming years (an Oracle study [ORA 16] even predicts that more than three-quarters of brands will rely on virtual reality and chatbots to manage customer experience by 2020), and their large-scale exploitation will undoubtedly be one of the challenges of the coming years.
These new data availabilities also revolutionize market research approaches, which were historically mainly based on the exploitation of PII. The use of such data, which is always crucial in customer marketing, makes it possible, for example, to answer strategic questions such as:
- – What are the most attractive product categories for a first purchase?
- – Can we predict the value of a customer by their very first basket?
- – How can we identify, through segmentation, the top 10% of high value customers?
- – Which customer profiles are responsive to promotions versus fashion trends?
These studies are considerably enriched by the use of non-PII and Data Science approaches, particularly through machine learning. We are now at the heart of the challenges of migrating from a data depth model to a data volume model where algorithms contribute more than ever to the performance of marketing actions.
The progress made by these algorithms in recent years is significant. In language recognition and image recognition, for example, algorithms nowadays have lower error rates than humans (see Figure 12.7).
In relationship marketing, these models make the following possible:
- – to direct acquisition actions to target profiles similar to the most loyal customers: so-called “look-alike” approaches;
- – to determine the probability of a positive reaction to an advertising message based on customer history to target only the most reactive if they are targeted and would not react if they were not targeted: so-called “incrementality” approaches;
- – because of NLP (natural language processing), i.e. learning natural language, to perform large-scale semantic analyses. This is a relevant approach for the analysis of customer verbatim, or even more so for the analysis of Chatbot conversations.
Figure 12.7. Comparative evolution of error rates in image recognition: algorithms versus human between 2010 and 2016 (source: [ELE 16]). For a color version of this figure, see www.iste.co.uk/ngoala/augmented.zip
In physical stores, retailers are competing with innovative technologies to collect data while providing a better consumer experience. In 2017, for example, Facebook tested a new loyalty feature within its mobile application “Rewards”. The Application offered users the option of scanning QR codes deployed in physical stores to benefit from targeted promotions5. This technology was an interesting sales promotion tool for the retailer, which generated additional traffic in stores. But above all, it was an opportunity for Facebook to collect new data on its users and to generate additional advertising revenue. The same applies to connected loyalty cards, which will enable retailers to offer personalized shopping experiences in the future, with product recommendations and discounts, and will give them the opportunity to learn more about their customers: the Kiabi clothing brand, for example, tested a connected loyalty card in stores in January 20176.
Finally, the rise of mobile tracking technologies at points of sale (beacons) will accelerate data collection outside the digital environment. Beacons are terminals installed in stores (in its surroundings or inside the shelves) that operate using Bluetooth technology. They make it possible to identify and target consumers that are close to the terminals, provided that the company knows how to reconcile a consumer and their smartphone through the use of its database. This opens up a huge field of possibilities in terms of customer knowledge and targeted promotions. According to the Business Insider website [BUS 18], by 2018, there will be 4.5 million boxes in operation in the United States, including 3.5 million for the retail sector. Note that 50% of the main North American retailers have already initiated experimentation phases since 2014. For retailers, this is the fastest speed of technology adoption since the introduction of mobile credit card readers. As is often the case, the European market should logically follow the adoption pattern of the American market by accelerating the equipment of its points of sale in the coming years.
12.6. Data marketing… tomorrow
The use of customer data is rapidly evolving and makes it possible to identify tomorrow’s challenges in a forward-thinking manner. But it is also, sometimes, a tricky exercise. Nevertheless, several dynamics seem to be structuring the market.
A major trend in data use is the migration from a relatively “push” approach to relationship marketing to a much more “pull” approach. This migration is entirely in line with the logic of ubiquitous commerce, where commercial touchpoints are integrated into consumers’ life journeys. Because of artificial intelligence, this acceleration is intensified with conversational marketing, in particular by Bots (Chatbots, Callbots, Voicebots). The integration of artificial intelligence into the marketing cycle is full of promise, and in particularly the idea that a conversational agent can detect a customer’s dissatisfaction and if necessary make the decision to stop for a given time makes CRM promotional campaigns toward this customer seems very realistic.
Another structuring trend for data marketing lies in the next stages of programmatic advertising. While today’s programmatic advertising is essentially based on cookies and is largely limited to the world of the web browser, a major challenge lies in extending programmatic advertising to new connected platforms to escape the browser world. As such, the car, conversational bots, connected objects and the television could become the new gateways to one-to-one marketing. These platforms will undoubtedly generate an unprecedented amount of data. It is up to marketers to find a service-oriented, ethical and transparent exploitation for consumers, and finally to give it meaning and to find sources of value for all stakeholders.
12.7. References
[AUD 17] AUDIBLE/OPINIONWAY: Livre audio, podcast et autres contenus parlés: L’enjeu culturel de l’audio parlé en France, available at: https://m.mediaamazon.com/images/G/08/AudibleFR/fr_FR/img/site/mt/newsroom/CP_Audible_Opinionway_20170324.pdf, 2017.
[BUS 18] BUSINESS INSIDER, 6 predictions about digital trends in 2018, available at: https://www.businessinsider.fr/us/6-predictions-about-digital-trends-in-2018-2018-6, 2018.
[COM 18] COMSCORE, “Le Futur du Digital en Perspective”, White paper, available at: https://www.comscore.com/fre/actualites_et_evenements/Presentations-and-Whitepapers/2018/Global-Digital-Future-in-Focus-2018, 2018.
[ELE 17] ELECTRONIC FRONTIER FOUNDATION, AI Progress Measurement, available at: https://www.eff.org/fr/ai/metrics, 2017.
[GRO 97] GRÖNROOS C., “Keynote paper from marketing mix to relationship marketing – towards a paradigm shift in marketing”, Management Decision, vol. 35, no. 4, pp. 322–339, 1997.
[ORA 16] ORACLE, Can Virtual Experiences Replace Reality? Research report, available at: https://www.oracle.com/webfolder/s/delivery_production/docs/FY16h1/doc35/CXResearchVirtualExperiences.pdf, 2016.
[RUS 93] RUST R.T., OLIVER R.L., Service Quality: New Directions in Theory and Practice, SAGE Publications Inc., London, 1993.
Chapter written by Grégoire BOTHOREL and Virginie PEZ-PÉRARD.