8 Virtual Environment

(Lamberton and Stephen 2016)

  • look at at changes in academic on digital, social media, and mobile (DSMM) marketing themes from 2000 2015

  • DSMM as

Theme 1: facilitator of individual expression (self-expression and communication) Theme 2: decision support tools Theme 3: market intelligence source
Era 1 (2000-2005) with online communities

recommendation agents and comparison matrices

  • (Häubl and Trifts 2000) found that with decision aids, consumers enjoy higher-quality alternatives, lower search cots, better choices.

  • (Brynjolfsson, Hu, and Smith 2003)online retailers have to respond to a stronger price competition than offline retailers.

  • Internet boom does not mean price war or choice overload for consumers

help firms predict consumer behaviors.

Era 2: 2005-2010

50% penetration

Online WOM

WOM referred customers have greater long-term value than traditionally acquired customers (Villanueva, Yoo, and Hanssens 2008) (Trusov, Bucklin, and Pauwels 2009)

Theme 2 and 3 merged together.

Who drives diffusion: audience + influencers

(Trusov, Bodapati, and Bucklin 2010) found a way to identify influential users in online social networks

(Katona and Sarvary 2008)

(Katona and Sarvary 2010)

Era 3: 2011- 2015

80% penetration

People are more expressive

UGC as a marketing tool

Consumption and production are tradeoffs (people consume more content will produce less content, and vice versa) at indiviudal level (Ghose and Han 2011)

But

  • Papers usually include methodological keywords which suggests that the new paradigm new methods to play with new data types (click stream, online chat, etc.)_

  • During this period DSMM in both worlds (academia and practice) correspond well.

Tonietto and Barasch (2020) show that generating content during an experience increases feelings of immersion and accelerate perceived time, which in turn enhances enjoyment of the experience. Moreover, consumers who are incentivized or motivated by social norms to generate content have the same experiential benefits as those who create content organically.

Lacka et al. (2021) define price impact as” the impact on the variance of stock price.” They estimate the permanent and temporary price impacts of the firm-generated Twitter content of S&P 500 IT firms: firm-generated tweets induce both permanent and temporary price impacts, depending on valence and subject matter. Tweets reflecting only valence or subject matter concerning consumer or competitor orientation onlly have temporary price impacts, while those with both attributes generate permanent price impact. Moreover, negative valence tweets about competitors generate the largest permanent price impacts.

Golmohammadi et al. (2021) show that firm responses to complaints on social media increase the potential public exposure of complaints. This negative effect can outweigh any positive customer care–signaling impact from firm responses. More specifically, a response strategy that has a high level of complaint publicization (e.g., detailed responses through multiple communication exchanges) could decrease perceived quality and firm value, diminish the positive impact of a firm’s own posts, and increase the volume of future complaints. These adverse impacts are stronger for firms that are targeted by retail investors.

Miao et al. (2021) define avatars as “digital entities with anthropomorphic appearance, controlled by a human or software, that are able to interact.” (p. 5). The authors also offer a 2x2 typology of avatars (behavioral realism and form realism) from low to high.

Bharadwaj et al. (2021) extracted emotions from salespeople using their livestream videos. In one-to-many screen-mediated communications, salespeople should not display emotions.

Jia, Kim, and Ge (2020) found that there is an inverse size-speed association (i.e., speed-based scaling effect induces people to estimate the size of an object to be smaller when it is animated to move faster) when they see digital ads (because they learned from other animate agents such as animals). To mitigate this effect, ad creator can (1) reduce the similarity between the movement of object in the ads with objects in other domains (e.g., human, animals), (2) give consumers more knowledge about the domain, (3) highlight product size information. Interestingly, the effect can be reserved when a positive size-speed association in the base domain is made accessible.

KOUKOVA, KANNAN, and RATCHFORD (2008) found evidence from experiments that consumers’ content consumption preferences (e-book, books, etc) vary across platforms

(Yadav and Pavlou 2014) reviews marketing in computer-mediated environments

(Grewal, Gupta, and Hamilton 2021) Multimedia Medium (Twitter, Facebook, Instagram) and Modality (auditory, tone, visual, olfactory, gustatory, haptic systems, visual)

  1. Data Gathering
  • Consumer privacy

  • Legal and ethical concerns regarding web scraping

  1. Data Analysis
  • Feature extraction

  • Dimensionality reduction

  1. Interpretation

Substantive Challenges

  • Sentiment-based targeting

  • Location-sensitive targeting

  • Computers as Consumers

  • Privacy Risk

8.1 Social Media

(Pelletier et al. 2020)

  • Consumer usage motivation for Facebook, Twitter, and Instagram and understanding which platforms consumers prefer to use to co-create with brands
  • Types of customer response:
    • Social purpose: interact with others and share content with others

    • Information purpose: seeking out info, news, events, professional business purposes

    • Entertainment purposes: have fun or follow their favorite celeb and pop culture events

    • Convenience purposes just to browse and observe when they are bored

  • Uses and gratification theory: benefits extracted from a media source vary based on the different purposes for which a person chooses to consume media. (Severin, Tankard, et al. 1997)
    • Social motivation: people use platforms with more broad ways of communications for social motivation (e.g., Facebook)

    • Informational motivation: users with informational purpose usage will come to the platform with an emphasis on real-time information (Twitter)

    • Entertainment motivation: Hedonic purpose is to escape and enjoy experiences (Instagram). Users with entertainment purposes will go to a platform with an emphasis on hedonic experiences (Instagram)

    • convenience motivation: a source of convenience or distribution (to pass the time). Easy browsing to scroll content (e.g., Instagram). Users will go to the platform with visual content and minimal text (Instagram). People find all platforms (Instagram, Facebook, and Twitter) convenient.

  • Users who desire to engage in co-creation will go to a platform with a broader array of communications mechanisms.
  • With social media, consumers co-create brand stories and share personal experiences, which can enhance or destroy brand equity
  • Findings:
    • For informational purpose, consumer use twitter

    • For social purposes, people like to use Twitter and Instagram

    • Instagram is a primary platform for entertainment motivation and co-creating with brands

    • Facebook has the lowest usage intentions and co-creation (despite being the largest platform and network most widely used by marketers) Even though practitioners have a strong bias toward Facebook as a social media promotional platform, but then one size does not fit all.


(appel2019a?)

Social media:

  • “a collection of software-based digital technologies—usually presented as apps and websites—that provide users with digital environments in which they can send and receive digital content or information over some type of online social network” (p. 80)

  • “social media to be a technology-centric—but not entirely technological—ecosystem in which a diverse and complex set of behaviors, interactions, and exchanges involving various kinds of interconnected actors (individuals and firms, organizations, and institutions) can occur” (p. 80)

Social media at present

  1. Platform: major and minor, established and emerging the at providing the underlying tech and business model

  2. Use cases: How various kinds of people and orgs are using this tech and for what purpose

This paper categorizes social media as

  1. Communication between known others (e.g., family friends)
  2. Communication between unknown but share common interests
  3. Access and contribute to digital content (e.g., news, gossip, and product reviews)

Future of social media in marketing

  1. Immediate future:
    1. Omni-social presence: sharing is embedded on every social platform, embedded in every digital tool. Platforms broaden their platform to encompass the Omni-social world. Affect consumer decision-making process (beginning to end). Social media is shaping the cultural world.

    2. The rise of influencers: increase accessibility and appeal of social media. Micro influencer (more authentic and credible), virtual influencer (i.e., non-human, CGI).

    3. Privacy concerns on social media: no hard privacy definition., more and more people delete social media.

  2. Near future
    1. Combating loneliness and isolation: on the rise

    2. Integrated customer care: social-media-based customer care., reduced customer service using humans.

    3. Social media as a political tool:

  3. Far future
    1. Increased sensory richness: AR, VR

    2. Online, offline integration and complete convergence: omni channel approach (e.g., mobile app to do AR) online and offline self (self presentation)

    3. Social media by non-humans: AI, social bots, hard to do attribution model, buy likes and shares, erode consumer trust,


(kietzmann2011?)

Seven functional building blocks:

  1. Identity: self-disclosure
  2. Conversation
  3. Sharing
  4. Presence
  5. Relationships
  6. Reputation: not only people but also content
  7. Groups

Differences Matter: the 4 Cs the 4 Cs: cognize, congruity, curate, and chase - relate how firms should develop strategies for monitoring, understanding, and responding to different social media activities

  • Cognize: firms should recognize and understand their social media landscape

  • Congruity: develop strategies that are congruent with different social media functionality

  • Curate: firms should be the curator of social media interactions and content.

  • Chase: firms should understand the conversation and info flow of social media content.



8.2 Who-Generated Content

Observational and experimental work on Negative WOM (Sen and Lerman 2007)

Differences between Firm and User-generated content can be found in Colicev, Kumar, and O’Connor (2019)

Colicev, Kumar, and O’Connor (2019) propose a framework of how FGC and UGC (based on information processing and source credibility) match on marketing funnel stages and related studies. UGC has a strong correlation with awareness and satisfaction while FGC (specifically vividness) are more effective for consideration and purchase intent. While UGC valence dominates UGC volume for these stages. Brands with higher corporate reputation have stronger relationships between dimensions of FGC and the marketing funnel stages.


(Hewett et al. 2016)

  • Brand Buzz in the Echoverse

  • Echoverse: “the entire communication environment in which a brand/firm operates, with actors contributing and being influenced by each other’s actions.” (p. 1)

    • Actors

      • Firms: advertising, press releases

      • Consumers: online WOM, attitudes, and behaviors

    • Sources of actions: firms, news media, online WOM, consumer sentiment, and business outcomes

  • Data: longitudinal US financial services

  • Use: Vector Autoregressive Model with Granger causality

  • Can view table 1 for review on echoverse components

  • Symmetric echo between traditional media and online WOM.

  • Firm communications influence the traditional media, consumer sentiment, business outcomes (e.g., deposits)

  • Negative news spiral: Bad news can spread fast

    • Solution: Twitter and press releases can save the day
  • Advertising do not directly affect traditional media, online WOM, or consumer sentiment, but directly affect firm outcome.

  • Firms care more about online WOM as compared to consumer sentiment (measured by survey)

  • Online WOM can affect consumer sentiment, but not the other way around.

  • Online WOM affects business outcome., but not the other way around.

  • Twitter gone from positive spiral to negative spiral overtime.

  • Even though firms counteract negative WOM with positive tweets, the volume of their tweets stay limited.

  • At first, Twitter has no impact on business outcome, but later they do have more significant impact on firm performance (p. 15)

  • Wells Fargo extremely positive tone backfire in negative firestorm, while consistent (i.e., large volume) of moderate tweets are better.

8.3 Images

Potential Variables:

  • Structural Complexity: complexity due to visual features (Pieters, Wedel, and Batra 2010). Measured by the Canny Edge Detection technique (Forsythe, Sheehy, and Sawey 2003), because edge information is highly correlated with observed complexity

  • Color Complexity: complexity due to variety of colors within an image (Reinecke et al. 2013). Measured by the color variety algorithm (Bing Zhou, Shuang Xu, and Xin-xin Yang 2015)

    • \(C = \sum_{i = 0}^m n_i \log(\frac{n_i}{N})\) where

      • \(m\) is the number of distinct colors

      • \(n_i\) the number of pixels of the i-th color

      • \(N\) is the total number of pixels

  • Image Permanence: how adept images are at remaining in users’ memory (after users are no longer looking at them) (Khosla et al. 2012, 2015). Measured based on AMNet memorability index (Fajtl et al. 2018), which has been verified by (Leyva and Sanchez 2021) (achieved near human consistency in predicting image memorability).

  • Number of faces: based on Google Vision API

  • Aspect ratio: height divided by width

  • Average HSV value: Average HSV values across all pixels within an image

  • Themes: LDA based on keywords (labels) from Google Vision API


Pieters, Wedel, and Zhang (2007) found that optimal feature design of ads (e.g., brand, text, pictorial, price, and promotion) can be achieved without increased costs. Moreover, the authors also propose two entropy-based measures of clutter effects, where they characterize the salience of feature ads based on Attention Engagement Theory.

Wedel and Kannan (2016) is a review on marketing analytics for data-rich environments. And competition for attention in these environments are more fierce.

Y. Li and Xie (2019) found a significant and robust positive mere presence effect of image content on user engagement in both product categories on Twitter, and high-quality and professionally shot pictures consistently lead to higher engagement on both platforms (e.g., Twitter and Instagram) for both product categories. However, the effect of colorfulness varies by product category, while the presence of human face and image–text fit can induce higher user engagement on Twitter but not on Instagram. And the fit between the image and the accompanying message matters

Jalali and Papatla (2016) study visual UGC. Where color compositions were operationalized as combinations level of hue, chroma, and brightness. Consumer engagement (i.e., click rate) is higher for photos that have higher proportions of green and lower proportions of red and cyan, as well as higher chroma of red and blue.


8.4 Mobile and Smartphone

(Rangone and Renga 2006) offers a framework in the domain mobile advertising

Melumad and Pham (2020) found that consumers derive not only function benefits, but also emotional benefits from their smartphones. Under stress, consumers want their smartphones, because smartphones give users greater stress relief. Moreover, this pacifying effect is greater for one’s own smartphone than other’s smartphones.


8.5 Review

Nguyen et al. (2020) found that more expertise leads to less extreme evaluations. Reviewing experts have less impact on a brand valence metric (which affects page rank and consumer evaluation). And experts do benefit and harm service providers with their ratings. Hence, excellent experiences may lead to lower ratings from experts (than from novices)

Observational data show that experts are more likely to post negative reviews.

Finkelstein and Fishbach (2012) found that as people gain more expertise, they are more likely to respond to negative reviews (i.e., feedback), which is robust for many domains (e.g., language acquisition, environment, marketing). novices are more likely to respond to positive feedback while experts seek and respond to negative feedback

Folse et al. (2016) defines negatively valenced emotional expressions as having intense language, all caps, exclamation points, emoticons in online reviews. Reviews with negatively valenced emotions are viewed as more helpful and damage attitude toward the product when used by experts, but when sued by novices, a negative self-reflection is observed and attitude towards the product is unchanged. Language complexity moderates the adverse effect of expertise on trustworthiness.

Dai, Chan, and Mogilner (2019)

  • Based on Amazon reviews and experiments, Consumer reviews are less likely to be trusted for experiential purchases than they are for material ones.

  • This impact stems from the idea that ratings of experiences are less representative of the purchase’s objective quality than reviews of actual objects. These findings reveal not only how word of mouth influences different types of purchases, but also the psychological mechanisms that underpin customers’ reliance on consumer reviews. Furthermore, these findings imply that people are less responsive to being instructed what to do than what to have, as one of the first examinations into how people choose among numerous experience and material buying possibilities. (X. (Shane). Wang et al. 2021)

  • extract and monitor products and attributes info from consumer reviews using machine learning and NLP (to create embedded representation)

    • In customer evaluations, embedded representation characterizes (represents) textual data like particular product features by employing the words that surround such textual data (i.e., the contextual information). Neural networks are used to measure how similar different product attributes are based on what people say about them (i.e., contextual information), which shows similarities and differences in how people use the attributes. From this embedded representation, the model then picks out multi-level clusters of product attributes that show how abstract the product benefits are at different levels.
  • Close the gap between engineered attributes (i.e., concrete attributes) and meta-attributes (abstract attributes)

  • Survey can be inconsistent and time consuming.


(Schoenmueller, Netzer, and Stahl 2020) found Polarity self-selection of online reviews and it reduces the informativeness of online reviews

(Banerjee, Dellarocas, and Zervas 2021) Q&A section (answered by other consumers and sellers) improve fit and match between products and consumers and reduce negative reviews regarding mismatch.


(Ordabayeva, Cavanaugh, and Dahl 2022)

  • Negative internet reviews from socially distant (but not socially close) individuals may not be as harmful to identity-relevant brands. Because a negative review of an identity-relevant brand can threaten a client’s identity, the consumer will seek to strengthen their relationship with the brand.

  • They show that this effect does not appear when the review is positive or when the brand is irrelevant.


(Nishijima, Rodrigues, and Souza 2021) interestingly found that Rotten Tomatoes ratings has no affect on box office performance using the categorization of “fresh” or “rotten.” They obtain box office performance data from boxofficemojo


8.6 Slacktivism

Supporting a political or social cause through social media or online petitions, with minimal work or commitment.

8.7 e-Marketplace

8.8 Social Listening Platforms

According to Forrester Wave’s report by Jessica Liu and Sara Dawson (2020), the top 10 social listening platforms (SLPs) are

  1. Brandwatch
  2. Digimind
  3. Linkfluence
  4. ListenFirst
  5. Meltwater
  6. NetBase Quid
  7. Sprinklr
  8. Synthesio
  9. Talkwalker
  10. Zignal

8.9 Live Streaming

(Lin, Yao, and Chen 2021): found that happy streamers can make audience happier and tip more. This effect is bigger for broadcaster with more experience, receive more tips, and popular.

8.10 Augmented Reality

Yim, Chu, and Sauer (2017)

  • When compared to web-based product presentations, AR provides effective communication benefits by providing better novelty, immersion, enjoyment, and utility, leading to better attitudes regarding medium and purchase intention.

  • In the AR condition, immersion mediates the relationship between interactivity/vividness and two performance outcomes: usefulness and enjoyment, whereas on the web, no significant routes between interactivity and immersion, nor between previous media exposure and media novelty are found.

  • Two schools define Interactivity (p. 91)

    • Technological outcome: focus on speed, mapping, range (i.e., the extent to which one can manipulate content).

    • Users’ subjective perceptions: very much related to motivation (then what is the contribution of this school of thought?)

  • Vividness (i.e., realness, realism, or richness) is “the ability of a technology to produce a sensorially rich mediated environment” \[@steuer1992, p. 80\] (which is very similar to the mapping aspect of Interactivity, then wouldn’t it be hard to operationalize these two constructs).

    • Technological perspective: vividness increases depth and breadth of the usage experience
  • Model (p.94): it makes sense that media usefulness and enjoyment directly affect attitude toward medium. However, immersion could potentially directly affect purchase intention via the path of attitude toward the product.


8.11 Artificial Intelligence

(Garvey, Kim, and Duhachek 2021)

  • When engaging with an AI agent, consumers respond better in terms of increased purchase likelihood and satisfaction when a product or service offer is poorer than expected.

  • Consumers are more likely to respond positively to a human representative if the offer is better than expected.

  • When administering offers, AI agents are regarded to have weaker intentions than human agents, which explains for this result

  • Marketers may anthropomorphize AI agents to reinforce their perceived intentions, giving them a way to get credit from customers when they make a better offer and avoid blame when they make a bad one.


8.12 Search Engine

(Guan and Cutrell 2007) found the Google golden triangle where people pay attention the the top 3 organic results and spill over to the fourth one (paid).

8.13 Customer Engagement

(Harmeling et al. 2016) visualize the mechanism of how customer engagement affects firm performance

(V. Kumar et al. 2010) categorize 4 components of a customer’s engagement value:

  • Customer lifetime value (the customer’s purchase behavior)

  • Customer referral value (incentivized referral of new customers)

  • Customer influencer value (customer’s behavior to influence other customers)

  • Customer knowledge value (value added to the firm by feedback from the customer)