International Journal of Research in Marketing 34 (2017) 22–45

P.K. Kannan a,⁎, Hongshuang “Alice” Li b

a University of Maryland, College Park, MD 20742, United States
b Kelley School of Business, Indiana University, United States

a b s t r a c t


We develop and describe a framework for research in digital marketing that highlights the touchpoints in the marketing process as well as in the marketing strategy process where digital technologies are having and will have a significant impact. Using the framework we organize the developments and extant research around the elements and touchpoints comprising the framework and review the research literature in the broadly defined digital marketing space. We outline the evolving issues in and around the touchpoints and associated questions for future research. Finally, we integrate these identified questions and set a research agenda for future research in digital marketing to examine the issues from the perspective of the firm.
© 2016 Elsevier B.V. All rights reserved.

a r t i c l e i n f o


Article history:
First received on January 22, 2016 and was
under review for 5 months
Available online 3 December 2016
Guest Editor: Michael Haenlein


Keywords:
Digital marketing
Online
Mobile
Internet
Search engine
User generated content
Omni-channel marketing

1. Introduction

It has been nearly a quarter century since commercial use of the Internet and the World Wide Web began. During this time the business landscape has changed at a frenetic pace.Large multinational corporations such as Google, Facebook, Amazon, Alibaba, eBay and Uber, unheard of twenty years ago, have emerged as key players in our modern economy. In 2015, online sales accounted for 7.4% of overall retail spending in the U.S., the highest percentage since tracking began in 1999 (Phillips, 2015). Sales made through mobile devices have increased at a rapid rate to between 22% and 27% of all online sales (Rao, 2015; Malcolm, 2015). Corporations now highlight the importance of creating a “digital relationship” with customers (Phillips, 2015). Moreover, digital technologies and devices such as smartphones, smart products, the Internet of Things (IoT), Artificial Intelligence, and deep learning all promise significant transformations of consumers’ lives in the near future.It is against this backdrop that this paper seeks to understand how the developments in digital technology are re-shaping the process and the strategy of marketing, and the implications of this transformation for research in the broad space we call “digital marketing”.

Our objectives for this paper are three-fold. First, we develop and describe a framework for research in digital marketing that highlights the touchpoints in the marketing process as well as in the marketing strategy process where digital technologies are having and/or will have a significant impact. Next, we organize the developments and extant research around the elements and touchpoints comprising the framework and review the research literature in the broadly defined digital marketing space.Using the framework, we also outline the evolving issues around the touchpoints and associated questions for future research. Finally, we integrate these identified questions and set a research agenda for future research in digital marketing.

In our discourse, we examine the research issues in digital marketing from the perspective of the firm – that is, we examine the strategic, tactical and implementation implications of the research conducted in the domain of digital marketing and focus on substantive issues of managerial relevance rather than on behavioral or methodological research per se. However, these issues could lead to fundamental questions that could be answered in the domains of consumer psychology, marketing analytics, economics, or computer science.In order to be as comprehensive as possible in covering the key substantive research developments in the area of digital marketing, and given our focus, we have narrowed down our search without compromising the representativeness. Our search for relevant literature focuses on four marketing journals: International Journal of Research in Marketing, Marketing Science, Journal of Marketing Research, and Journal of Marketing, focusing on articles published between 2000 to 2016.We started at Web of Science and searched for articles with the keywords “digital” or “online” as either the research topic or part of the article title, which provided us with 305 “seed articles”. As we read these papers, we eliminated those that were not directly relevant and included other relevant papers cited in these seed papers. This expanded our list to other journals not covered in our initial search. For each topic discussed in our paper, we selected the earliest papers in this list, and added a few most frequently cited papers in that topic to discuss under each topic making up our review.To this list we also added the most recent papers to render the review as current as possible. Thus, the review of extant research is not meant to be exhaustive but rather representative in order to cover the issues with sufficient depth and focus on future research issues appropriately. Our review complements recent review articles on digital marketing and related topics. The article by Yadav and Pavlou (2014) focuses on marketing in computer-mediated environments and reviews literature in both marketing and information systems. The article by Lamberton and Stephen (2016) focuses on consumer psychology, motivations, and expressions in digital environments to highlight a few.The article by Wedel and Kannan (2016) focuses on modeling and methodological issues in marketing analytics necessitated by the advent of digital, social and mobile environments. Our review cites these articles at the appropriate sections for further details on issues we do not cover. The paper is organized as follows.In Section 2 we present the framework and identify touchpoints in processes where digital technologies play a key role. In Sections 3 through 7, we review the literature around each element and touchpoint of the framework and discuss briefly open areas of inquiry. In Section 8 we present more details on these open areas of research and present an agenda for future research and conclude in Section 9.


⁎ Corresponding author.
E-mail addresses: pkannan@rhsmith.umd.edu (P.K. Kannan), aliceli@indiana.edu (H.“A.” Li).
http://dx.doi.org/10.1016/j.ijresmar.2016.11.006
0167-8116/© 2016 Elsevier B.V. All rights reserved.

2. A framework for analysis
2.1. Definition and framework

The term “digital marketing” has evolved over time from a specific term describing the marketing of products and services using digital channels – to an umbrella term describing the process of using digital technologies to acquire customers and build customer preferences, promote brands, retain customers and increase sales (Financial Times, lexicon.ft.com). Following the American Marketing Association’s firm centric definition (https://www.ama.org/AboutAMA/Pages/Definition-of-Marketing.aspx), digital marketing may be seen as activities, institutions, and processes facilitated by digital technologies for creating, communicating and delivering value for customers and other stake-holders. We adopt a more inclusive perspective and define digital marketing as “an adaptive, technology-enabled process by which firms collaborate with customers and partners to jointly create, communicate, deliver, and sustain value for all stakeholders”.1

The adaptive process enabled by the digital technologies creates value in new ways in new digital environments. Institutions enabled by digital technologies build foundational capabilities to create such value jointly for their customers and for themselves. Processes enabled by digital technologies create value through new customer experiences and through interactions among customers. Digital marketing itself is enabled by a series of adaptive digital touchpoints encompassing the marketing activity, institutions, processes and customers. Significantly, the number of touchpoints is increasing by over 20% annually as more offline customers shift to digital technologies and “younger, digitally oriented consumers enter the ranks of buyers” (Bughin 2015). In view of the above, we identify key touchpoints affected by digital technologies and propose a research framework that is inspired by the marketing process as well as by the marketing strategy process. The conventional marketing strategy process starts with an analysis of the environment including the five C’s – customers, collaborators, competitors, context, and company (firm). While these elements are presented in our framework (Fig. 1), customers emerge as the central focus (in the left box) with other elements such as context, competitors and collaborators making up the environment that the company operates in. Our key objective is to understand how digital technologies (at the bottom in Fig. 1) interact with the five C’s as well as the interface among these elements. We specifically identify the concepts, institutions and structures that emerge from these interactions – platforms and two sided markets, search engines, social media and user-generated content, emerging consumer behavior and contextual interactions. This analysis forms the input to the actions of the firm, encompassing all elements of the marketing mix – product/service, price, promotion and place – as well as information gathering through marketing research and analytics, which informs the marketing strategy of the firm. We focus again on how digital technologies are shaping these actions, information acquisition and analysis, and marketing strategy. Finally, as the outcome of the marketing actions and strategies, we examine the overall impact of digital technologies in value creation – creating value for customers (through value equity, brand equity, relationship equity and customer satisfaction), creating customer equity (through strategies for acquisition, retention and higher margin),


1- We thank an anonymous reviewer for this suggestion.

and creating firm value (as a functions of sales, profits and growth rate). Our framework, therefore, identifies the key touchpoints in the marketing process and strategies where digital technologies are having or likely to have significant impact. It not only encompasses the elements identified in Fig. 1, but also the interfaces among those elements, as shown by the arrows in Fig. 1. The framework also highlights our emphasis on uncovering issues in digital marketing that will impact the firm directly or indirectly. Next, we provide an overview of these concepts and elements highlighted in our framework.

2.2. Key concepts and elements

Digital technologies are rapidly changing the environment (Box 1 in Fig. 1) within which firms operate. Digital technologies are reducing information asymmetries between customers and sellers in significant ways. Analysis of interactions of digital technologies and the elements of the environment starts with the examination of how consumer behavior is changing as a result of access to a variety of technologies and devices both in the online and mobile contexts. We focus on how this affects information acquisition with regard to quality and price, the search process, customer expectations, and the resulting implication for firms. Next, we examine digital technologies’ facilitation of customer-customer interactions through online media – word-of-mouth, online reviews and ratings, and social media interactions (social media & user generated content (UGC)). The emergence of platforms – institutions created through digital innovations which facilitate customer-to-customer interactions for ideation in new product/ service development, those that connect customers and sellers in platform-based markets and those that leverage two-sided markets for their revenue generation – is also examined as collaboration enablers that connect a firm to its market using digital technologies In the same way, firms have to contend with search engines as both collaborators and platforms on which they compete with other firms in acquiring customers. Thus, we also review the research on search engines and the interactions among customers, search engines and firms. Finally, we examine the interactions of digital technologies with different contexts of geography, privacy and security, regulation and piracy, and their implications for digital marketing (contextual interactions). Within the company (Box 2 in Fig. 1), digital technologies are changing the concept of product in three ways in order to provide customers new value propositions – augmenting the core product with digital services, networking of products using digital technologies to release the dormant value inherent in the products, and finally, morphing products into digital services. We examine these trends and the opportunities they create for customizing and personalizing customer offerings, by varying not only the core product/service but also the augmented digital services. The developments in digital product lines and tailored offerings to customers lead to pricing challenges for firms. The reduction in menu costs associated with digital technologies also leads to opportunities for dynamic pricing and yield management in product and service categories traditionally sold with list prices. These developments along with the use of online auctions for products/services, search keywords, display ads, and name-your-ownprice strategies, have given rise to interesting research questions that we review. In addition, the interface between pricing and channels (both offline and online) is becoming an important issue as more firms adopt online and mobile channels to target and transact business with customers.

Over and above traditional means of communication such as print, radio, and television, the digital environment provides new means to reach customers and promote products and services via e-mails, display advertisements, and social media (promotion).There has been much focus on the effectiveness of such new media and its incremental contribution over traditional media in building brands and affecting outcome variables of interest. Newer forms of promotional tools such as location-based mobile promotions and personalized promotions are increasingly used and we explore the implications of their use for firms as well as for customers.We also focus on the rise of new channels for customer communications and promotions, not only online and mobile, but also sub-channels within each of these environments such as social channels, search engines, and e-mail that help firms to provide significant value to customers as well as acquire the right customers and increase customer value. The impact of digital technologies on outcomes (Box 3 in Fig. 1) could span across different dimensions – in creating value for customers and in extracting the value for the firm. The outcomes are a reflection of how the firm has been able to benefit from the opportunity provided by digital technologies to create value for their customers and also create value for themselves. As Fig. 1 suggests, firms can leverage the interactions of digital technologies with the environment and with its own strategic and tactical actions in leading to the outcomes. We focus on research that models this relationship across various dimensions of outcomes – value equity, brand equity and relationship equity (Rust et al., 2004), customer satisfaction, customer value as a function of acquisition, retention and profitability of customers, and at a more aggregate level, firm value as a function of sales, profits and growth rate. Research on understanding how different channels and media contribute to these outcome measures and how this understanding affects marketing actions will also be discussed. Marketing research (Box 4 in Fig. 1) focuses on the acquisition and processing of information generated as a result of the use of digital technologies to understand the specific elements of the environment, actions and outcomes that inform the marketing strategies of the firm. Examples include researching the browsing behavior of customers at websites and mobile sites, comparing search behavior in online environments to searches in mobile environments, and examining online reviews, social interactions, and social tags to understand how a firm/brand is being perceived by the market. While the substantive issues are discussed in the context of the environment and the company, all such research involves the development of specific methodologies and/or metrics. Within this section we highlight the managerial questions that could be answered using data within the firm and environment, however, we do not focus on the methodological aspects as these issues are well covered in extant research (see, for example, Wedel & Kannan, 2016). There are issues related to marketing strategy (Box 5 in Fig. 1) that are partly captured in one or more of the elements or interfaces, and we discuss specifically in Section 7 those issues which are not captured elsewhere. We do not track the developments in digital technologies per se, but in discussing the impact on the customer touchpoints they are implicitly taken into account. In the various sections, we also outline the descriptions and capabilities of new technologies that lead to new opportunities – for example mobile technologies, virtual reality, wearable computing and IoT, etc.

3. Digital environment

Table 1 provides an overview of the state-of-the-art research developments under each of the five main areas we focus on.

3.1. Consumer behavior

It order to understand the impact of digital technologies, it is important to understand how consumers’ buying process – prepurchase, purchase consummation and post-purchase stages – are changing as a function of new environments and devices. Consumers’ information acquisition, search and information processing are also affected, and as a result, decision aids can play an
Table 1
Digital technologies and marketing environment: research issues and state-of-the art.


Research developments

a. Stages of buying process, purchase funnel, and impact of digital environments and digital devices
b. Information acquisition, search, information processing and decision aids in digital environments
c. Buyer behavior across digital and non-digital environments
d. Customer trust and risk perceptions in digital environments
a. Electronic Word-of-Mouth (eWOM) and motivation for eWOM
b. Dynamics in eWOM posts and their impact on sales
c. How eWOM posts influence other posts?
d. Social networks, identification and targeting of influencers
e. eWOM and fake reviews
a. Network effects in online platforms, information asymmetry and impact on sales
b. Impact of competition on two-sided content platforms
c. Issues in crowdsourcing and using platforms for innovations
a. How should search engines price and rank keywords?
b. How should advertisers choose specific keywords and bid on them?
c. Relationship between rank, click-through rate and conversion rate, and decision support for optimal bidding
d. Synergy between organic search and paid search
a. Interaction between geography/location and digital environments
b. Impact of regulatory environment – Privacy concerns and effectiveness of digital marketing
c. Impact of piracy of content

Area of focus

Consumer behavior

.

Social media and UGC

.

.

Platforms and two-sided markets

.

Search engines

.

.

Contextual interactions

important role in the new environments. Recent marketing research has provided insights into consumer behavior, customer trust and risk perceptions in these processes across digital and non-digital environments. This sub-section will review these issues. It is well known that consumers move through different stages in the buying process starting with awareness, familiarity, consideration, evaluation and purchase. If consumers receive value consistently by purchasing a brand, they are more likely to become loyal customers. In conventional offline environments the consumer journey is fairly extended, especially in the consideration and evaluation stages, whereas in the digital environment these stages can be quite compressed or even eliminated (Edelman & Singer,
2015). Customers can gather information from focused research at search engines and read other customers’ reviews on retailers’ sites or third-party forums not controlled by the seller, and the initial demand to purchase could be created just by seeing a post on social network. Thus, in the digital environment, customers can move through their decision journey in fundamentally new ways.
Our key research focus is to understand how buyer behavior is affected by the digital environment, specifically through interactions with search engines, online reviews, recommendations, and other similar information not produced or controlled by the firm or brand. In addition, even as the environment itself changes depending on the device that customers use – personal computers (PCs), smart phones, tablets, or wearable devices – how do these devices and environment affect buyer behavior? Such research issues focus on the elements unique to the devices or environment and examine their impact on consumer decision making and buying behavior. A good example of an early paper focusing on such research is by Haubl and Trifts (2000) who investigated the nature of the effects that interactive decision aids may have on consumer decision making in online shopping environments. Another example is by Shi, Wedel, and Pieters (2013) who used eye-tracking data to examine how customers acquire and process information in their online decision making. Shankar et al. (2010) developed propositions on how the characteristics of mobile devices may influence consumer behavior, and Xu, Chan, Ghose, and Han (2016) examined the impact of tablets on consumer behavior in digital environments.Focusing on the role of decision aids in evolving consumer behavior, Shi and Zhang (2014) found
that consumers evolve through distinct behavioral states over time, and the evolution is attributable to their prior usage experience with various decision aids. Decision aids can be constrained by device features, and thus the optimal design of decision aids could vary across devices.

Research in the practitioner’s realm offers a new perspective of the digital buying journey wherein interactive social media and easy access to information may expand rather than narrow customer choices. Furthermore, customers can influence other potential buyers through online reviews, social media, and so forth, during both the pre-purchase and post-purchase stages (Court et al., 2009).
The customer decision journey, more often than not, spans across digital as well as traditional offline environments.This buyer behavior across environments has been the subject of several papers. For example, do customers who shop across the two environments spend more money than those who use just one channel? Kushwaha and Shankar (2013) addressed this question with a compiled database of around one million customers shopping across 22 product categories over 4 years. In their analysis, a print catalog was the only offline channel and its customers were compared with customers who use the online channel, or both. They developed a conceptual framework where the monetary value of a customer relies on two features of the product category – whether the product is utilitarian or hedonic and whether the product is of low or high perceived risk. They found that the multichannel customers are not necessarily more valuable than single channel users. For example, the offline-only customers have a higher monetary value than multichannel customers on low-risk utilitarian product categories, and the online-only customers spend more on high-risk utilitarian products than multichannel shoppers. Neslin et al. (2006) provided a comprehensive review on the customer behavior in the search, purchase and after-sale stages of multichannel shoppers. They identified five key challenges for future research, including data integration, understanding customer behavior, channel evaluation, resource allocation, and channel coordination. In addition, the large volume of individual-level touch point data adds more complexity to these challenges.
Information search plays an important role in the customer’s decision journey. Early research by Ratchford, Lee, and Talukdar (2003) examined how the digital environment affects automobile purchases and revealed that the Internet shortens the consideration and evaluation stages of the customer journey, and customers would have searched even longer if the Internet was absent. A later study by Ratchford et al. (2007) in the same automobile context, found that the Internet substitutes for time spent at the dealer, for print content from third-party sources in pre-purchase stage, and for time spent in negotiating prices in the purchase consummation stage. These results highlight the importance of the reduced search costs and thus more efficient purchase processes
in digital environments.

The specific manner in which the consumers’ digital search unfolds and how the process is affected and moderated by searchand decision-aids in an ever-changing digital environment is, in and of itself, an important topic. Many of the research findings in the general area of search can be applied to specific digital settings. For example, Seiler (2013) developed a structural model in which the search decision is jointly modeled with the purchase decision. The customers decide on how much information they need to gather by trading off the perceived purchase utility with search cost. Using customers’ shopping data in traditional brick-and-mortar stores, Seiler (2013) showed that customers do not search in around 70% of their shopping trips due to high search costs. If the search cost is reduced in half, as in his counterfactual analysis, the elasticity of demand can be more than tripled.In the online setting, when search cost is significantly reduced, researchers found higher demand elasticity in various product categories (Degeratu, Rangaswamy, & Wu, 2000; Granados, Gupta, & Kauffman, 2012).

Kim, Albuquerque, and Bronnenberg (2010) integrated the sequential search process into a choice model. They used webcrawled data of viewing and ranking for all camcorder products at Amazon.com for a one and a half-year data window and assumed these data are aggregations of individual-level optimal search sequences. Their results showed that consumers usually search among ten to fifteen product alternatives. While the ranking and filter tools offered by the retailer can help customers reducesearch costs, these tools also concentrate demand on the bestselling products. Bronnenberg, Kim, and Mela (2016) examined customer online search behavior for multi-attribute, differentiated durable goods such as cameras, and found that on average a customer conducts 14 searches online across multiple brands, models, and online retailers over a 2-week period. However, the extensive search is confined to a small set of attributes and 70% of the customers search and purchase within the same online retailer. They also found that customers first search with generic keywords and narrow down to specific keywords, echoing the research findings by Rutz and Bucklin (2011).
Trust is an important element that influences customers’ selective information gathering and search behavior in the digital environment. Shankar, Urban, and Sultan (2002) introduced a conceptual framework for online trust building using stakeholder theory, which approached the trust building from the perspective of different stakeholders such as customers, suppliers, and distributors. From customers’ perspectives, they want retail sites to be trustworthy and their transaction information and personal information to be protected. However, such customer needs may not quite align with supplier’s efficiency perspective. In one of the earliest empirical studies on customer privacy concerns in online shopping, Goldfarb and Tucker (2011a) conducted a field experiment and found that targeting can undermine the effectiveness of a display ad. According to their research, an ad that is both obtrusive and content-based targeted has less impact on a purchase than an ad that is only obtrusive or targeted, possibly due to ustomers’ privacy concerns.

Understanding how emerging digital technologies affect consumer behavior is an important research area. It is the key to understanding the role of various touchpoints in determining customers’ purchase journey, extending the work of Court et al. (2009). Do these touchpoints always compress and shorten the purchase journey as described by Edelman and Singer (2015) or is there a tipping point where the journey gets extended? How are these findings change across devices? Does switching across channels and devices increase or decrease the search cost? Theory-driven research focusing on the impact of devices on consumer behavior is critically needed.

3.2. Social media and user-generated content

An important characteristic that sets the digital environment apart from the traditional marketing environment is the ease with which customers can share word-of-mouth information, not only with a few close friends but also with strangers on an extended social network. In the digital environment, customers can post reviews on products, services, brands and firms at firms’ websites as well as third-party websites and social networks, and these reviews reach a much larger number of potential customers. Toubia and Stephen (2013) focused on the important motivation question: why do people contribute on social media? Their research distinguished between two types of utility that a contributor derives from social media: (1) intrinsic utility, the direct utility of posting content and (2) image-related utility derived from the perception of others. These two types of utility can be empirically distinguished because the former depends on posting behavior whereas the latter only relates to the number of followers a person has on the social network. In their field experiment, Toubia and Stephan randomly selected 100 active non-commercial users on Twitter and added 100 synthetic followers to each user over a 50-day period. They found the intrinsic utility outweighed image-related utility when the Twitter users had fewer followers, whereas image-related utility became more dominant as the Twitter users gathered more followers. Moreover, the image-related utility was larger than intrinsic utility for most users. It is important to identify the influential individuals in a social network. In their seminal paper, Watts and Dodds (2007) proposed a hypothesis that there is a small group of influencers, the impact from whom can cascade to others. Trusov, Bodapati, and Bucklin (2010) developed a latent measure of influence and empirically examined the influence on individual log-in behavior with social network data. Katona, Zubcsek, and Sarvary (2011) studied the diffusion of influence and found that an individual’s position in the network together with specific demographic information can be good predictors of adoptions. An individual is more likely to adopt if she is connected to more adopters or if the density of adopter connections is higher in her group.

One form of online customer interactions that has been studied extensively is the online review (e.g., user generated content and electronic word-of-mouth, or eWOM). Just as with traditional offline word of mouth, eWOM encompasses customers’ knowledge about the products, their usage, experience, recommendations, and complaints, and is generally perceived as trustworthy and reliable. Moreover, eWOM may have richer content and larger volume than offline word of mouth, and it is much more accessible and can be shared widely in the digital environment. Given the importance of eWOM, it has been the subject of extensive research over the last decade, addressing issues such as: the motivation for eWOM posts; the impact of eWOM posts on sales and the dynamics of such posts; how eWOM posts influence other posts and reviews; and the identification of the most influential people in the network, known as “influencers”. More recently, research has also focused on deceptive reviews and their motivations.

Godes and Mayzlin (2004) were the first researchers to investigate the impact of the online review. They examined the volume and dispersion of the online review and found that the dispersion is a good predictor of the ratings of a TV program. Chevalier and Mayzlin (2006) studied the important relationship between online reviews and sales using online book reviews. They found that online reviews are generally positive and that these reviews can increase a book’s sales rank, but that negative reviews have a stronger impact than positive ones. Moe and Trusov (2011) identified two dimensions of online reviews – product evaluations and social dynamics – and found both influence sales. Apart from the relationship between eWOM and sales rank, researchers are also developing tangible metrics to measure the return on investment (ROI) of social media. Kumar et al. (2013) introduced a metric to measure the viral impact of eWOM and its associated monetary value. Wu, Che, Chan, and Lu (2015) developed a learning model to evaluate the monetary value of a review and found more value is derived from contextual comments than numerical ratings. In addition to the organic eWOM created by customers, can firms drive sales by generating their own eWOM?

According to Godes and Mayzlin’s (2009), the answer is yes. In a large-scale field experiment, in which they collected data from customers as well as non-customers, they found less loyal customers are likely to have a greater impact on eWOM campaigns. Chen, Wang, and Xie (2011) compared the impact of eWOM and observational learning at Amazon.com, where eWOM is created by customers and the observational information is provided by an Amazon feature that shows the customer what other customers purchased (as an aggregate metric in percentage) after viewing the same product. This observational learning feature was discontinued by Amazon in late 2005 and resumed in late 2006. The researchers collected one and a half years of data covering these two feature changes at Amazon and used a first difference model to measure the impact of eWOM, observational learning, and their interactions. The results showed that negative eWOM is more influential than positive eWOM, whereas the reverse is true for observational learning. These findings imply that it is profitable for retailers to provide observational information and the impact of such information can be strengthened by eWOM volume.

One selection issue that needs to be taken into account is that not every customer contributes to online reviews and a customer’s decision to write a review needs to be modeled. Ying, Feinberg, and Wedel (2006) developed a selection model to capture this decision process and also examined the valence, volume and variance of ratings. They found that more active reviewers post lower ratings than less active reviewers and that over time these active reviewers become the majority of the reviewer population, which explains the declining trend of the proportion of favorable ratings over time. Another explanation of the declining trend in positive online reviews is offered by Li and Hitt (2008). They identified a selection process where the customers who purchased later and thus reviewed later had lower utility from the product and the lower rating over time represented the lower valuation by these later customers. In addition, Godes and Silva (2012) contributed to the research of the dynamics of eWOM by explaining the temporal dynamics. Moe and Schweidel (2012) focused on why consumers post ratings and modeled the individual’s decision to provide a product rating and the factors which influenced that decision. The researchers showed that there were significant individual differences in how consumers responded to previously posted ratings, with less frequent posters exhibiting bandwagon behavior while more frequent posters tended to differentiate themselves from other posters. These dynamics affect the evolution of online product opinions over time.

The past decade has witnessed a surge in research on online reviews. Based on 51 studies, You, Vadakkepatt, and Joshi (2015) conducted a meta-analysis on the elasticity of the volume and valence of online reviews. They found that the valence elasticity (0.417) is higher than the volume elasticity (0.236) and these elasticities are higher for private and low-trialability products. In an interesting study on online deception, Anderson and Simester (2014) found that approximately five-percent of the online reviews at a large retailer’s website were for products never purchased by the reviewers. These tended to be more negative than the average review and the authors conclude that it is unlikely that all were written by competitors or their agents as these reviewers seem to have purchased a number of other products at the retailer.In a recent meta-analysis, Rosario et al. (2016) found that eWOM had a stronger effect on sales for tangible goods new to the market, but not for services. They also found that eWOM volume had a stronger impact on sales than eWOM valence and negative eWOM did not always jeopardize sales, but high variability in reviews did.

Lamberton and Stephen (2016) provided a detailed survey of recent research developments encompassing substantive domains of digital, social media, and mobile marketing topics from 2000 to 2015. They focused on digital technologies as a facilitator of individual expression, as a decision support tool, and as a market intelligence source, which complements our above treatment by providing more details about social media and UGC research.

Current research on UGC mainly centers on the study of structured data – the number of stars or likes and their statistics such as mean and variance. However, the content of the reviews and posts themselves contain valuable and direct information expressed by the customer. While sentiment analysis has been used to capture valence information, only a few empirical papers have tapped into the unstructured textual content of online reviews (for example, Tirunillai & Tellis, 2014; Büschken & Allenby, 2016). Future research needs to focus more on the semantic analysis of UGC.

3.3. Platforms and two-sided markets

Several platforms and platform markets have emerged in the digital environment, including those that connect individual customers with other individual sellers (eBay), those that connect customers with a multitude of firms/sellers (Alibaba, Amazon, media sites, and various advertising exchange networks), firms with firms (business-to-business platforms) and firms with the crowd (crowdsourcing and innovation platforms like Kickstarter). In all the above examples, platforms are independent thirdparty entities that connect buyers, sellers, firms, the crowd, and so on. To this list, we also add customer communities that firms organize so that they can observe and interact with the crowd (firm-sponsored platforms). Innovation platforms (for example, Dell Ideastorm, Cisco’s open innovation platform) and other social communities are good examples. The revenues for the independent two-sided platforms come from one or a combination of commissions, performance-based charges (for example, Google charges advertisers by cost per click), and impression-based charges (for example, the cost per thousand impressions charged by the ad networks). Two-sided markets are well-studied in traditional network markets and much of the research is
readily applicable to online platforms too (see, for example, Parker & Van Alstyne, 2005). In this section we will specifically focus on studies that draw upon the unique characteristics of the digital environment in examining the relevant research issues.

Extant research in online platform markets has empirically examined the existence of network effects, that is, more users/ buyers will increase the number of advertisers/sellers of the two-sided marketplace (Parker & Van Alstyne, 2005). Tucker and Zhang (2010) conducted field experiments and investigated the influence of disclosing information on the user base and seller base of an exchange network. Their results suggested that a seller prefers an exchange network with more sellers due to its attractiveness to more buyers. Fang et al. (2015) applied a vector autoregressive models analysis to investigate the direct effects of
buyers and sellers on the platform’s advertising revenue, as well as the indirect effects of click-through rate and cost-per-click (CPC). Their results demonstrated strong network effects – more buyers boost the CPC for the sellers and more sellers increase the buyers’ click-through rates. The two-sided platform in their study launched a search advertising service within their data window, which allowed them to capture the different effects during the launch and the mature stages of search advertising services. Interestingly, they found the ROI at the mature stage is twice of the ROI at the launch stage. In the launch stage, they found the existing sellers bid higher than new sellers and have a stronger impact on click-through rates. The reverse is true during the mature stage. As for buyers, the new buyers have a greater impact on the click-through rates and price during the launch stage and this impact is even more prominent during the mature stage. Additionally, the impact of new buyers lasts three times longer than that of existing buyers. Godes, Ofek, and Sarvary (2009) examined the impact of competition on two-sided platforms in both duopoly and monopoly settings with analytical models. They found in a duopoly setting the media firms tend to charge more for their content than what they would charge in a monopoly case where no competition exists. This contradicts the common belief in the negative relationship between competition and price that the price is lower when competition is more intense. As a result of the network effects of a two-sided market, the profits from advertising may decrease at a higher level of competition, but the content profits could still increase. Jiang, Jerath, and Srinivasan (2011) examined the role of Amazon as a platform provider in linking small sellers with customers and the strategies it adopts in observing the demand for sellers’ products and offering the high-demand products themselves and examined, in a game-setting, the firm strategies. Chakravarty, Kumar, and Grewal (2014) focused on business-tobusiness platforms and examined their total customer orientation as a function of buyer-side versus seller-side concentration on the platforms and found that total customer orientation increased with buyer-side concentration.
Crowdsourcing platforms are another type of platform that connect firms to their customers (the crowd) and help generate ideas for new products and services. Such innovation platforms allow firms to repeatedly collect ideas from a dispersed crowd of customers and choose the best ideas to develop further. Bayus (2013) researching Dell’s Ideastorm platform found that customers who repeatedly submitted ideas were more likely to provide good ideas but once they won their success rate dropped. Luo and Toubia (2015) focusing on online idea generation platforms suggested that the platforms should customize the task structure of the idea contests on the basis of each customer’s (those who submit ideas) domain-specific knowledge in order to increase the idea quality. As innovation platforms become increasingly popular, there is more research attention on how to increase ideation quality in such platforms. In the context of firm-sponsored community platforms, Manchanda, Packard, and Pattabhiramaiah (2015) focused on “social dollars” (the economic benefit to a firm generated from customers being engaged members of the community) and found stronger social than informational source of economic benefits accrued for the firm from customers in the platform, highlighting the benefits of running such communities for the firm. A more detailed treatment of platforms can be found in Sriram et al. (2015), where the authors identified opportunities to advance the empirical literature in platform research. There is still a significant gap in our understanding of the processes within the platforms that can lead to more efficient and effective interactions and outcomes (for both firms and customers/crowd). For example, how can platforms maintain the engagement of customers/crowd in e-commerce interactions or in new product/service development ideation processes? How can social processes and commercial processes co-exist and complement each other on platforms? As the social networks facilitate commerce on their platforms – Facebook runs a virtual Marketplace to let users trade within a local community and Chat apps like WeChat and Line can be used to order groceries delivered to the door, the research on the relationship between social and commercial features of a platform assumes greater importance.

3.4. Search engines

Search engines allow customers to acquire free information on products and services and identify firms and brands that fit their search criteria. Search engines provide organic (natural) listings of websites as well as paid search listings in response to the keywords that users type in. In this sub-section, we first review research examining the impact of search engines on outcome variables of interest. Then, we focus on search engine decisions as a platform, and decisions by advertisers as the clients of the platform.We examine the ecosystem as a whole and center our discussion on the relationships among the decisions of various players. Finally, we highlight the research on the synergy between organic and paid search.

The effectiveness of search engines is supported by several empirical studies. Chan, Wu, and Xie (2011) found that the customers acquired through paid searches purchase more and generate higher customer lifetime value than customers acquired from other online or offline channels, indicating that search engines are an effective selection mechanism to identify high-value
customers.In addition, Dinner, Van Heerde, and Neslin (2014) found that paid search advertising is more effective than offline advertising, and Wiesel, Pauwels, and Arts (2011) also found the impact of a paid search is more enduring than that of e-mail.

There are three players involved in search engine marketing: the search engine, the advertiser/firm, and the customer. We have already discussed the role of search engines in a customer’s decision journey. In this sub-section, we focus on specific issues from the perspective of the search engines and the advertisers: (1) how should search engines price and rank keywords, and (2) how should advertisers choose specific keywords and bid for those keywords for the most efficient and effective customer acquisition. The generalized second price auction is widely adopted by search engines to determine the prices and rankings of listings for each keyword. It is well-known in economics that the generalized second price auction outperforms the first-price auction, but its implementation at search engines may not always be optimal. Amaldoss, Desai, and Shin (2015) compared the generalized second price bidding and the first-page bid estimate mechanism implemented at Google (the estimate offers the minimum bids to appear attractiveness to more buyers. Fang et al. (2015) applied a vector autoregressive models analysis to investigate the direct effects of buyers and sellers on the platform’s advertising revenue, as well as the indirect effects of click-through rate and cost-per-click (CPC). Their results demonstrated strong network effects – more buyers boost the CPC for the sellers and more sellers increase the buyers’ click-through rates. The two-sided platform in their study launched a search advertising service within their data window, which allowed them to capture the different effects during the launch and the mature stages of search advertising services. Interestingly, they found the ROI at the mature stage is twice of the ROI at the launch stage. In the launch stage, they found the existing sellers bid higher than new sellers and have a stronger impact on click-through rates. The reverse is true during the mature stage. As for buyers, the new buyers have a greater impact on the click-through rates and price during the launch stage and this impact is even more prominent during the mature stage. Additionally, the impact of new buyers lasts three times longer than
that of existing buyers. Godes, Ofek, and Sarvary (2009) examined the impact of competition on two-sided platforms in both duopoly and monopoly settings with analytical models. They found in a duopoly setting the media firms tend to charge more for their content than what they would charge in a monopoly case where no competition exists. This contradicts the common belief in the negative relationship between competition and price that the price is lower when competition is more intense. As a result of the network effects of a two-sided market, the profits from advertising may decrease at a higher level of competition, but the content profits could still increase. Jiang, Jerath, and Srinivasan (2011) examined the role of Amazon as a platform provider in linking small sellers with customers and the strategies it adopts in observing the demand for sellers’ products and offering the high-demand products themselves and examined, in a game-setting, the firm strategies. Chakravarty, Kumar, and Grewal (2014) focused on business-tobusiness platforms and examined their total customer orientation as a function of buyer-side versus seller-side concentration on the platforms and found that total customer orientation increased with buyer-side concentration.

Crowdsourcing platforms are another type of platform that connect firms to their customers (the crowd) and help generate ideas for new products and services. Such innovation platforms allow firms to repeatedly collect ideas from a dispersed crowd of customers and choose the best ideas to develop further. Bayus (2013) researching Dell’s Ideastorm platform found that customers who repeatedly submitted ideas were more likely to provide good ideas but once they won their success rate dropped. Luo and Toubia (2015) focusing on online idea generation platforms suggested that the platforms should customize the task structure of the idea contests on the basis of each customer’s (those who submit ideas) domain-specific knowledge in order to increase the idea quality. As innovation platforms become increasingly popular, there is more research attention on how to increase ideation quality in such platforms. In the context of firm-sponsored community platforms, Manchanda, Packard, and Pattabhiramaiah (2015) focused on “social dollars” (the economic benefit to a firm generated from customers being engaged members of the community) and found stronger social than informational source of economic benefits accrued for the firm from customers in the platform, highlighting the benefits of running such communities for the firm. A more detailed treatment of platforms can be found in
Sriram et al. (2015), where the authors identified opportunities to advance the empirical literature in platform research.

There is still a significant gap in our understanding of the processes within the platforms that can lead to more efficient and effective interactions and outcomes (for both firms and customers/crowd). For example, how can platforms maintain the engagement of customers/crowd in e-commerce interactions or in new product/service development ideation processes? How can social processes and commercial processes co-exist and complement each other on platforms? As the social networks facilitate commerce on their platforms – Facebook runs a virtual Marketplace to let users trade within a local community and Chat apps like WeChat and Line can be used to order groceries delivered to the door, the research on the relationship between social and commercial features of a platform assumes greater importance.

3.4. Search engines