Making the most of Twitter as webcare platform
The effects of personalization and timely response on consumers’ attitudes
Marlou Peer 10165258
Master’s Thesis
Graduate School of Communication Master’s Programme Communication Science
Persuasive Communication Supervisor: dr. E.H. Maslowska
Abstract
As the Internet has changed the nature of consumers’ communication with companies from private to public, more and more companies are using webcare in their interactions with consumers. Webcare is a practice that helps consumers engage with companies through social media. Microblogging website Twitter is used a lot for webcare purposes. Because of its open and public elements, Twitter was chosen as main research platform in this study. Prior
research states that it is vital for companies to use personalization elements when engaging with consumers. The current research examined two elements that can potentially enhance a consumers’ attitude towards a brand: personalization and timely response. The following personalization elements were taken into account: signature, pronouns, name use, informal elements and non-verbal elements. As predicted, several elements of personalization are associated with a more favourable consumer attitudes: non-verbal elements and use of the consumers’ name. The results also show that timely response is positively related to consumer attitudes. There was no interaction effect of personalization and timely response on consumer attitude. The results have implications for both companies and academics. This study
concludes with some directions for future research.
Introduction
At the beginning of 2016, Ziggo, one of the largest telecommunication companies in the Netherlands made the headlines. It was not because they released a new product, had a crisis or wanted to make an announcement, but because of their online customer service. Ziggo was one of the first companies to reach traditional media with their online conversations. The media as well as the public praised Ziggo and called it ‘special,’ ‘high quality service,’ ‘the service that went too far’ and ‘scoring big game’ (AT5, RTLnieuws, AD, Metronieuws, 2016).
Consumer Michael asked Ziggo for a new splitter, since he lost his old one when he recently moved. He had another problem, however, the Ziggo subscription was not on his name, but his girlfriend’s. Michael then asked whether Ziggo could not only send a new splitter, but also send her a little love. Ziggo employee Johan responded to Michael that he could send him a new splitter, free of charge, but the love was something Michael had to fix himself. A few days later Michael received a package from Ziggo, which contained a splitter, a handwritten card and a little housewarming present. Michaels’ response? “Would it be possible to give Johan from the webcare department a raise? Or at least give him a pat on the shoulder? Besides the great service I have received, he gave a little love away as well.” Conclusion of this story is one happy consumer and loads of positive attention for Ziggo’s webcare team. The story was posted on Facebook and Twitter, and was shared and retweeted hundreds of times before the traditional news picked it up.
This example illustrates the enormous influence of social media and the power that consumers hold to either positively or negatively portray brands online. Dutch brands are aware of that power and are using Twitter more than ever before. Centraal Bureau Statistiek (CBS, 2015) states that 30% of the Dutch companies with more than 10 employees are on microblogging websites such as Twitter. This number has grown quite a lot since the last
measurement in 2012, when only 21% could be found on microblogs. CBS (2015) also states that Dutch companies are the most active on social media compared to other countries in the European Union. While in the past the emphasis lied on brands conveying stories to their consumers, today, with the Internet and social media emersion, the consumer has the
possibility to talk back to the brand (Jansen, 2010). In addition, consumers can talk about the brand with other consumers online and other consumers can observe their conversations with brands, which can have an even bigger influence.
Because of this public nature of brand-consumer conversations, brands recently started engaging in webcare. Webcare can be described as companies’ online engagement with consumers, whereby the company actively seeks and addresses consumer feedback on the Internet (Van Noort & Willemsen, 2011). Today, 59% of Dutch brands engage in webcare (Bernritter, 2017). Webcare is one of the key determinants of successful social media
activities (Schamari & Schaefers, 2015), which is why researching its effectiveness will be of great value to companies as well as for academics. While there has been a lot of research on consumer complaining behaviour in online environments and the effects of (negative) word-of-mouth, extant research on webcare is limited. Webcare is after all a relatively new element that poses both threats as well as opportunities for brands.
Webcare may be especially effective if it is personalized. In the Ziggo case, making the customer service personal was received well by the consumer himself and also other consumers. However, personalization may also result in paradoxical effects, since not every consumer may appreciate it, which has been shown for other forms of personalized
communication (e.g., email campaigns). Even though there is a lot at stake for companies, to date there has been very little research into the effects of personalization in the webcare context. That is why this study will research whether the degree of personalization in webcare responses could make a difference in consumers’ attitudes towards brands.
Furthermore, with all the technological developments, consumers become used to instant gratification. They are always live, following conversations and news in real time. They expect the same from companies they interact with, that is, a timely response. Therefore, timely responses from companies are also examined.
In summary, this study tests what roles personalization and timely responses play in consumers’ relationships with companies. Therefore, the following research question is proposed: What roles do personalization and timely responses play in affecting consumers’ attitudes in webcare?
Theoretical framework What is webcare and why is it important?
Interactions with companies have changed their nature from private to public. In the past, consumers relied on the one-to-one interaction with companies, but today they are the ones in control of the conversations. In addition, the conversations often happen in public social media environments. Social media have changed the opportunities and strategies to let brands communicate with their consumers (Mangold & Faulds, 2009). According to Mangold & Faulds (2009) social media has evolved into an important factor in influencing different aspects of consumer behaviour, including forming opinions, influencing purchase behaviour and communication, and evaluation after purchase. On social media, both happy and unhappy consumers can publically express their experiences with companies, influencing other
consumers. For example, potential buyers can see hundreds of product reviews. Moreover, (negative) word-of-mouth spreads fast and far on the Internet and once posted online, it is documented forever, because information can always be retrieved. Especially complaints can spread really fast, which can have a negative impact on brands’ reputation. An example from nearly a decade ago illustrates this well, when telecom provider T-Mobile lost between
€200.000 and €300.000 reputational damage in a ‘Twitter war’ between the company and Dutch comedian Youp van ‘t Hek (Berkeljon, 2010). To avoid such negative outcomes, companies started engaging in webcare.
Following Harrison-Walker (2001), Hong & Lee (2005) and van Noort & Willemsen (2011), webcare can be defined as the activity wherein companies engage in online
interactions with consumers by actively seeking and addressing consumer feedback on the Internet. Webcare is a quick way to aid consumers via social media channels, such as Facebook and Twitter. Webcare teams respond to questions, share information, resolve complaints and even casually chat with consumers. Webcare is thus used as a tool in support of the company-consumer relationship, but also serves as image, reputation and brand management (Van Noort & Willemsen, 2011).
Companies can engage in webcare in two ways. 1) They can respond to messages that consumers have send to the company. This is called reactive webcare (Van Noort &
Willemsem, 2011). For example, reactive webcare on Twitter are all the tweets that
consumers send mentioning (@) the company. 2) Companies can reply to messages that did not actively seek the companies’ attention, which is called proactive webcare. For example, proactive webcare on Twitter can be seen whenever a company replies to tweets that have no mention (@) of the companies’ Twitter handle. Webcare is shifting up in importance, letting it near marketing and branding. Some researchers even state that webcare is one of the key determinants of successful social media activities (Schamari & Schaefers, 2015). Brands can use their webcare to express their identity and image in interactions between consumers and companies. Companies can use consumers to do their marketing. Because after all, helping a consumer fix a problem in a public space where anybody can read along, is not only
reaching other consumers. Broekhuizen (2017) mentions that the importance of ‘care’ in webcare appears to increasingly contribute to a company’s image.
Goals of webcare
According to Van Noort, Willemse, Kerkhof & Verhoeven (2014) webcare has three main goals, which are illustrated in Figure 1. The first goal is customer care satisfaction and customer retention. Any brand that has a webcare team should build and maintain the brand-consumer relationship by keeping the brand-consumers satisfied. Problem solving and exceeding expectations are key factors in customer care satisfaction and customer retention.
The second goal covers public relations: the reputation and relationship quality of brands. Webcare is a good form of reputation management. Webcare teams have the power to detect issues early and prevent crises. A complaint can be an opportunity, if the brand
responds in the right way. Since social media has the ability to spread publicly within minutes it is important to know how to respond to them. Complaints on Twitter have the potential to become an issue quickly and this is why public relations are important in webcare. The final webcare goal has to do with marketing. Webcare can attract more consumers and enhance positive eWOM. Next to that, monitoring the Internet is easy and the insights that derive from monitoring are valuable. Brands can use these inputs to improve products and services (Van Noort, Willemse, Kerkhof, & Verhoeven, 2014). It can be stated that webcare is at the heart of the three disciplines described above (Bernritter, 2017).
Figure 1: Goals of webcare. Figure recreated based on Van Noort, Willemse, Kerkhof & Verhoeven (2014, p. 80).
Twitter as an open platform
Twitter is one of the most popular microblogging websites on the Internet. Trilling (2015) describes Twitter as a social medium where deliberative discourses show up and an open place where anyone can interact with everyone. This makes Twitter different from other social media. Facebook, for instance, is mainly closed and messages are personal. What is sharing on Facebook is called retweeting on Twitter. Anyone can retweet anyone, except when a Twitter profile is locked, which is rare. Only 2.3% of active Twitter accounts are locked (Twopblog, 2012). Twitter users can interact by replying or favouriting each other’s tweets. Next to that, to use Twitter one does not have to ‘befriend’ anyone. Each Twitter user can choose whom to follow and it does not have to be a mutual decision. This is not
something that is socially expected either (Trilling, 2015). Another important element of Twitter is hashtags (#). Anyone can choose to send out a tweet related to a certain hashtag, which is usually related to a topic. Twitter users can collectively view each and every tweet
ever tweeted within that hashtag but can also retweet, favourite or reply to the hashtag. All of these Twitter elements thus show that it is an open platform.
The role of Twitter in brand-consumer communication
Kaplan and Haenlein (2010) stated that Twitter is very effective for marketing communications. Other researchers (e.g., Edosomwan et al., 2011) found that when a company engages in Twitter, it strengthens the company’s brands. It also makes it more attractive to consumers. Brands’ presence on Twitter can also translate into financial outcomes. In later research, Kwong and Sung (2011) discovered that Dell’s twitter usage activity generated $7 million in sales. Hence, brands use Twitter. It has been estimated that every fifth tweet on Twitter refers to a brand, product or service (Jansen, Zhang, & Sobel, 2009). Out of these brand-related tweets 20% contain a consumer’s opinion. The current 500 million tweets per day provide a daily 100 million opinions about brands, products and services (Jansen, Zang, Sobel & Chowdury (2009); Sayce, 2017).
Webcare on Twitter
In research conducted by Twitter (Elrhoul, 2015) itself, four ways in which brands can build brand-consumer relationships on Twitter surfaced. The first one is empathy and helpfulness. Consumers are 76% likely to recommend a brand if the webcare team shows empathy and are offering help (Elrhoul, 2015). Being friendly towards your consumers is one of the key drivers towards a more positive brand attitude.
Second, brands should be personal. If a brand humanizes itself, including both the Twitter users’ name as well as signing off the tweet with the brand representative’s name, consumers are more likely to have a positive brand-consumer relationship, and 77% of consumers are more likely to recommend a brand (Elrhoul, 2015).
Third, Twitter discovered that brands should respond within one hour. Over 60% of the Twitter users expect to be answered by the webcare team within an hour. The quick response time then leads to positive consumer sentiment.
The fourth way to build a good brand-consumer relationship is for webcare teams to follow-up and make sure an issue has been resolved. Consumers are 31% more likely to recommend brands that do that (Elrhoul, 2015). Therefore, being accessible in webcare interactions is the final key to driving recommendations and creating and maintaining positive attitudes in relation to brand-consumer communication via Twitter.
The role of personalization in webcare
As Twitter’s (Elrhoul, 2015) study shows, webcare needs to meet certain criteria to be
successful. Recent research has studied whether the role of language use can influence how an organization is perceived. From the role of a Twitter dialogue (Colliander, Dahlèn & Modig, 2015) to the influence of formal and informal address (Lenoir, 2015; Koot, 2015), use of language is becoming more important today. Kelleher & Miller (2009) were first to introduce the concept of conversational human voice (CHV), which they define as “an engaging and natural style of organizational communication as perceived by an organization's publics based on interactions between individuals in the organization and individuals in publics” (p. 177). Earlier research tried to use CHV to explain the effects of webcare with mixed results (e.g., Van Noort & Willemsen, 2011).
Message personalization can be seen as a part of CHV. The degree of personalization in webcare responses on Twitter could make a difference in consumer’s attitudes towards a brand. Service research by Watson (2012) found that consumers respond to a service failure more favourably if the brand provides a personal service to the consumer instead of non-personal service. Other research has stated that non-personalization is one of the key determinants
in engaging with consumers online (Van Noort, Willemsen, Kerkhof, & Verhoeven, 2014; Van Noort, Antheunis, & Verlegh, 2014; Schamari & Schaefers, 2015).
The elements of personalization: Pronouns and non-verbal cues
There are different elements that make messages more personal. Pronouns (e.g., I, we, me) and non-verbal cues are two examples. Lenoir (2015) did a study on consumer responses to formal and informal address and found that most European consumers have a stronger preference for informal address, such as use of first-person singular and plural pronouns. Kwon and Sung (2011) claim that the use of first-person pronouns helps establish
relationships between consumers and brands. That is because information is stated as a belief rather than a fact, which reduces the impersonality of the message.
Also, non-verbal cues can overcome impersonality. By using non-verbal cues brands can convey emotions as well as give verbal subtleties to their consumers (Kwon and Sung, 2011). There are several groups of non-verbal cues. The first one is emoticons. Kwon and Sung (2011) found emoticons to be the second-most used non-verbal cue by brands on Twitter. In Twitter replies, emoticons were used in 35.4% of the brands’ tweets. Yet, to date there has been no research on effects of emoticons so it is important to research this.
The second non-verbal cue is abbreviations. Abbreviations can be described as any form of a word that is shortened. Words like OMG (‘Oh My God’) and LOL (‘Laughing Out Loud’) are used a lot online. While abbreviations come in very handy on Twitter, they can also be seen as personalization elements. Kwong and Sung (2011) found that out of all verbal-cues brands used abbreviations most frequently. Abbreviations were used mainly in retweets (70.8%) and original tweets (67.5%), but accounted for only 44% of replies. This suggests that brands would use more formal language in their consumer-replies than in their original tweets, which may be because it is easier for brands to express their identity to all of their
followers via original tweets. Webcare wise, brands using abbreviations via replies can show they are part of the ‘in-group’ and are thus a brand who can be trusted and whose views are of interest.
Another non-verbal cue is repeated punctuation. It can be used to express emotions (Kwon & Sung, 2011). Consumers as well as brands can easily express themselves by using more than one exclamation mark or question mark. For example, ANWB uses this kind of communication towards their consumers:
1
Figure 2: Screenshot taken from ANWB’s Twitter
Kwon and Sung (2011) found that repeated punctuation is used in 16.5% of brand tweets. Dan Zarella (2013) found that even though tweets containing exclamation marks did not generate more clicks, they did generate considerably more retweets. Having more retweets as a brand means a bigger audience and more possible consumers, or as Dan Zarella (2009) states: “Retweets are spread because of some other trigger or set of triggers has been pulled in your brain. And that trigger fires the biggest gun ever seen” (p. 4).
The fourth element Kwon and Sung (2011) talk about are intentional misspellings. Just like the other elements, using intentional misspellings can be considered as informal
communication, again making the brand seem more human. Kwon and Sung (2011) also state that intentional misspellings are being used for emphasis. This way the formal ‘hello’ can become an informal ‘hellooooo’ in seconds, giving the brand a different identity when just
using formalities. They also describe this as sound mimicking, where brands can use “aaahhh” or “ooopsy” as informal elements binding themselves to consumers via
personalization. Kwon and Sung (2011) found that only 11.2% of the brands use intentional misspellings on Twitter.
The final element Kwon and Sung (2011) describe is capitalization. Capitalization was used in 9.2% of the brands’ tweets. However, since capitalization can give another dimension to a specific response, it is still an important non-verbal cue. Companies can tweet “WOW” to a consumer and it can be received as yelling, but also as enthusiasm. Using capitalization means that it is even easier for companies to convey emotions. Therefore, capitalization will be considered a meaningful variable.
There are also other elements that can be used to personalize communication. The usage of the (Twitter) name of the consumer is one way to personalize a reply. Van Noort, Atheunis and Verlegh (2014) state that if brands use the first name of the consumer it can enhance positive brand responses. That is because this type of natural communication mimics one-to-one communication and can be perceived as more human. Next to name usage, the way a brand signs their tweet may also have an influence on how a brand tweet is perceived. Talking to employee ‘Kim’ makes a brand look more human than talking to ‘KLM’.
Informal elements such as humour, references to (pop) culture, copying the writing style of the consumer and use of media and hashtags are also becoming more popular. However, previous research has not really devoted attention to them. Kaplan and Haenlein (2010) state that brands should blend in with their consumers and thereby engaging in informal communication online. Applying humour and other informal elements could make brands look more human.
Therefore, it is expected that compared to impersonal webcare, personal webcare on Twitter (i.e., containing either personal pronouns, non-verbal cues or informal elements) is
better received and leads to more positive attitudes than impersonal, generic webcare. The first hypothesis therefore states that:
H1: The use of personalization in webcare is associated with more positive attitudes than generic webcare.
The role of timely responses in webcare
Next to personalization, timely responses may play an important role in consumers’ reactions to webcare. Twitter (Elrhoul, 2015) claims the average response time of all brands engaging in webcare worldwide is 1 hour and 24 minutes, still, every company uses their own
endeavor. Some brands explicitly state their response time, while other companies just state the active times in which they reply throughout the day. For example, Coolblue
(@Coolbue_NL) promises to reply within 30 minutes and has its employees ready for tweets every day until 23.59pm (see Figure 3). In the Eptica Multichannel Customer Experience Study Retail, it was found that 64% of the Twitter users demand an answer on Twitter within 60 minutes (Eptica, 2015).
Figure 3: Coolblue’s Twitter banner
Consumers’ expectations of such fast replies can be explained by different factors. One of them is attachment to mobile phones. People are very dependent on their smartphones (Thorsteinsson & Page, 2015). According to Dscout, heavy smartphone users use their phones for about 225 minutes a day, while average users use their phone for 145 minutes a day
(Nelson, 2016). Knowing Twitter is accessible for brands and consumers anywhere and anytime, consumers expect their reply to pop up rather sooner than later.
In addition, as Broekhuizen (2017) claims, new technologies are changing the nature of webcare with many interactions getting automated. More artificial intelligence is used for the most common questions online. The use of automated responses saves quite some time and the consumer receives an answer almost immediately. However, consumers often do not know if the webcare is automatically generated or not. This causes people to use webcare instead of Googling and expect an immediate answer (Broekhuizen, 2017).
Research from over a decade ago found that when a company replies in a timely manner to an online complaint, the consumer does not have negative associations or
communicates in a negative way about the company (Hong and Lee, 2005). Also, in his study on airlines on Twitter, Huang (2015) found that the quicker the brands reply to tweets, the more revenue potentially exists as consumers were willing to spend almost $20 more for an airline that responded within one hour. If it took an airline longer than one hour to respond to the tweet, consumers were only willing to pay $2.33 more for that airline in the future.
Other research found that if a company does not respond on social media in a timely matter, an online firestorm can arise. According to Pfeffer, Zorbach and Carley (2014) online firestorms can be defined as sudden discharges of large quantities of messages that include negative eWOM and complaint behaviour in social media against a person, company or group. An online firestorm can cause serious damage to the reputation of a company, resulting in more consumers with a negative attitude towards a company.
Based on the discussion above, the second hypothesis states the following:
H2: The use of timely responses in webcare is associated with more positive attitudes than slow responses.
The joint effect of personalization and response time
According to van Belleghem (2014) consumer expectations have never been higher.
Consumers want both an excellent service and companies’ mistakes to be corrected quickly and properly. An excellent service can be reached through a personal touch, and correcting mistakes quickly and properly (Van Belleghem, 2014). Although there is no research looking at both personalization and timely responses in webcare, we can speculate that they affect each other in such a way that when both are present, consumers are especially satisfied with webcare. Therefore, it is plausible that:
H3: Personalization and timely response interact with each other resulting in synergistic effects.
Figure 4: Model of the research question
Method Design
In order to research the role of personalization and timely responses in webcare, a content analysis is conducted. BrandAsset Valuator 2014 created a list of the top 100 strongest brands in the Netherlands. Four different brands that operate in the Netherlands, in different branches were picked from the top 20. The brand had to a) be active on Twitter daily, b) have a
10.000 and 100.000 followers (see Appendix A). This resulted in the following selected (n = 4) Dutch brands: ANWB, Bol.com, Efteling and Schiphol.
These companies all have different goals. ANWB is a Dutch service company who are specialized in the organization of traffic and tourism. On Twitter they have 65.3k followers and they have integrated their webcare on their main account, @ANWB. Bol.com is one of the biggest Dutch webshops, selling almost everything – from books to make-up and from condoms to washing machines. On Twitter they have 53.5k followers and have integrated their brand with its webcare at @Bol_com. Schiphol is the biggest airport in the Netherlands and an important airport in Europe. Almost 64 million passengers passed Schiphol in 2016 (NOS, 2017). Their Twitter account, @Schiphol, has more than 70k followers. Finally, amusement park Efteling was selected. With nearly 61k followers on their account
(@Efteling), it is the biggest active entertainment park on Twitter. With these four companies, all located in different industries, a more objective image of the companies on Twitter is created. In the end the selected companies are from different sectors: We have one retail company (Bol.com), one company providing entertainment (Efteling) and two service companies (ANWB and Schiphol). All set criteria are met.
Sample
For these four brands, over 17,000 user tweets and over 95,000 brand tweets have been extracted from Twitter over the period of May 26, 2016 to May 2, 2017. From this sample, a random sample has been drawn. The random sample consists of a total of 1,150 analyzed tweets (see Table 1). From the brands Bol.com, Efteling and Schiphol 66 (n=66) users have been randomly selected and analyzed for the sample. From ANWB 68 (n=68) users have been randomly selected and analyzed.
Table 1
Division of users among the brands
Brand Number of tweets Percentages
ANWB 247 21.5% Bol.com 356 30.9% Efteling 270 23.5% Schiphol 277 24.1% Total: 1,150 100% Content analysis
A codebook was created to measure personalization, timely response and consumers’ attitude (see Appendix B). The first seven questions can be considered general questions and contain variables such as the date of the tweet, conversation language and type of interaction. The second category covers reaction speed of the brand in minutes. In the third category
personalization is measured through 1) personal pronouns, 2) signature usage, 3) name use, 4) non-verbal cues and 5) informal elements. The last category of the codebook measured the consumers’ attitude towards the brands’ reply. It is measured by favourite count and a coded consumer sentiment.
The tweets were coded by one coder. To ensure validity a second coder coded 29 users, which makes up for 13.7% (n = 158) of the total sample. Intercoder reliability is reported in the results section. The results of the analyzed tweets were inserted in IBM SPSS 22 where statistical analyses were carried out.
Variables Personalization
Personalization is operationalized in terms of several different elements.
1) Pronouns. For example, “I”, “me”, “we” and “our”. When the brand responses in first personal (singular or plural) it is coded as personal. Tweets written using other pronouns are considered impersonal.
2) Signature. When the brand used the name or initials of the employee (either in
combination with the brand name or without), it is considered personal. When the brand signed the tweet using the name of the brand, or the response was not signed, it is considered impersonal.
3) Use of consumers’ name. If the name of the consumer is mentioned, the tweet is coded as personal. If the name of the consumer is missing, it is impersonal.
An example of the first three dimensions is shown in the following tweet by @AlbertHeijn to a consumer:
Figure 5: A screenshot showing Albert Heijn answering user @Catherine_Pijls2
This example contains three elements of personalization. It is written in first person (“I”), is ended with the employee’s initials (“AdB”) and it uses the name of the consumer
(“Catherine”).
2
4) Non-verbal cues. This measure is based on Kwon and Sung’s (2011) and consists of several elements: I) Emoticons are divided in two groups: emoji’s and icons. The Japanese word emoji literally means image (e) and character (moji). Emoji’s can be defined as smileys and objects that are being used in electronic messages and on web pages. They are similar to icons except for the fact that icons use expressions using numbers, letters and punctuation marks, while emoji’s are a pictorial representation. Both can express a person’s mood and feelings. Emoji’s however can also express places, objects, types of weather, food, sports, animals and way more. II) Abbreviations can be described as any form of a word that is shortened. Use of words like OMG (‘Oh my God’) and LOL (‘Laughing Out Loud’) is measured. III) Repeated punctuation (e.g., !!!). IV) Capitalization (e.g., WOW or COOL) and V) Intentional misspellings (e.g., ca$h or hiiiiya).
5) Informal elements. Finally, personalization is measured through the use of informal elements. In this category use of I) humour, II) references to (pop) culture, III) words that express friendship, IV) copying the writing style of the consumer, V) hashtags and VI) media use (e.g., gifs), are considered elements of personalization.
Next to measuring personalization in five levels, representing the different elements, there was also one measure created by combining all of the elements. This variable was created to make it possible to test for interaction effects. All elements are measured on a yes/no basis. Each ‘yes’ corresponds to 1 and no corresponds to 0. Taking these scores
altogether, the variable named ‘personalization’ is created with a 5-point scale ranging from 1 (no personalization) to 5 (high personalization). All elements weigh the same and a total of 14 points is the maximum personalization amount. A high personalization score is coded
between 14 and 12 points. A medium personalization score is coded between 11 and 9 points. A neutral personalization score is coded between 8 and 6 points. A low personalization score
is coded between 5 and 3 points. And the no personalization category scored between 2 and 0 points (1 and 2 points were absent in the data set).
Timely response
Timely response is operationalized in minutes. A new variable is created for timely response dividing the response time in five categories, ranging from very slow (more than 3 hours), slow (up to 3 hours), neutral (up to 2 hours), fast (up to 1 hour) to very fast (up to 30 minutes). This division is made based on the four companies claiming to respond within 1 hour or less and the research showing consumers expect response within 1 hour. Outliers will be included in the analysis because social media has grown into an on-going process day and night. People want instant gratification. If someone lands at Schiphol at 03:00am and has a question, he/she should be able to get the answer in a timely matter, even though it is after midnight or on the weekend.
Consumers’ attitude towards the brands’ reply
The consumers’ attitude is measured with several elements.
1) Favourite count: Favourites have the function in which anyone can ‘favourite’ a specific tweet. The amount of people who have favourited a brands’ tweet could be an indication of how well the tweet was received. More favourites usually result in a more positive attitude.
2) Retweet count: Another indicator of the attitude is the retweet count, which could serve as an indicator of how popular and well or bad received a brands’ tweet was. Consumer do not retweet tweets at random, but only the tweets they want to show to their followers. This variable is important because depending on the context and reply of the brand,
retweeting a brands’ tweet could therefore serve as positive- as well as a negative- electronic word of mouth.
3) Consumer sentiment: This variable measures consumers’ sentiment in their final reply. Presence of: exclamation marks, a sarcastic tone, positive and negative emoticons, positive and negative words (e.g., that express gratitude, happiness, joy, anger or
confusion). All these elements taken together will form consumer sentiment, from which the coder decides if it is positive, mixed, negative or neutral.
Results Intercoder reliability
Intercoder reliability has been measured through Krippendorf Alpha (KALPHA). Four variables have been randomly for an intercoder reliability check. On the first variable (consumer sentiment - use of positive emoticons) α = .875 (n = 19) was scored with 2000 bootstrap samples. On the second variable (interaction type) α = 1.00 (n = 145). On the third variable (humour) α = .746 (n = 65). The final variable (name of the consumer) is α = .939 (n = 65). Reliability of the data is substantial to almost perfect therefore the reliability can be considered high.
Descriptives
The sample consists of 85.6% tweets in Dutch and 14.4% tweets in English. The average length of one conversation (exchange of tweets) is shown in Table 2. The average time it took a company to respond is calculated at M = 86.77 minutes (SD = 254,1, Mdn = 14).
Out of all the tweets, 47.2% are written in first-person plural pronoun (n = 222). Respectively 4.5% and 6.2% are written in third-person or in first-person singular while 42.1% of the tweets contained other pronouns. The tweet was signed with the name of
employee in 48.2% of the cases (n = 226), while 31.6% of the sample did not sign the tweet at all (n= 148). The other 20.3% signed their tweet using the initials of the employee (n = 95). The name of the consumer was used in 42.8% of the cases (n = 202).
Table 2
Descriptive statistics of conversation length
n Min. length Max length M SD
ANWB 68 1 9 3.63 1.525
Bol.com 66 2 28 5.39 4.285
Efteling 66 2 19 4.08 2.598
Schiphol 66 2 11 4.20 1.939
Type of interaction
From the entire sample of 1,150 tweets, n = 475 tweets are brand replies, n = 545 are consumers replies and n = 130 are tweets solely served to analyze consumer sentiment. Out of the n = 545 consumer replies, 28% tweeted a statement (“Yes the issue concerns an iPhone 6”) to a brand (n = 286). Questions make up for 13.7% of type of interaction (n = 140), followed by 7.7% complaints (n = 79). Compliments (2.9%), suggestions (0.7%) and other (0.3%) are tweeted the least.
Personalization
To test the first hypothesis, a multinomial logistic regression was conducted to analyze the relationship between personalization of webcare and consumers’ attitude as expressed in the sentiment of the final response (positive, neutral or negative). Favourite count was not used in
not interpretable. Also retweets were not used to measure attitude, as their meaning is difficult to interpret without additional information from the consumer. Pronouns, signature, use of consumers’ name, non-verbal cues and informal elements were included as predictors. A multinomial logistic regression was used because the dependent variable is categorical with three levels. It was chosen to compare the categories against the last category (negative attitude) because this category represents no effect: Personalization did not have the desired effect on consumers’ attitude (Field, 2013). That is why there is no negative attitude in the table. A test of the full model against a constant model was statically significant (see Table 3), indicating that the predictors reliably distinguished between positive, neutral or negative attitude (χ2(24) = 61.80, p < .001). Nagelkerke’s R2 of .25 indicates that 25% of the variation in attitude is explained by the model. The Wald criterion demonstrated that non-verbal cues (p < .05) and name use of consumer (p < .01) are statistically significant. The use of two non-verbal cues are positively associated with attitude. Therefore, the assumption that the use of personalization in webcare is associated with more positive attitudes than generic webcare is partly supported.
Timely response
To test the second hypothesis, a cross-tabulation was conducted to investigate the
relationship between timely response and attitudes. The relation between these variables is
significant, χ2(8, n = 270) = 15.35, p < 0.05. Therefore, the assumption that use of timely responses in webcare is associated with more positive attitudes is supported.
Table 3
Output of the multinomial logistic regression for personalization
95% CI for Odds Ratio
b (SE) Lower Odds Ratio Upper
Positive attitude Intercept 3.6 (1.2)** Non-verbal cues 2.4 (0.81)** 2.22 10.87 53.93 Informal elements 0.86 (0.86) 0.44 2.35 12.74 Pronouns -1.81 (0.84) 0.03 0.16 0.85 Signature -0.41 (0.66) 0.18 0.66 2.4 Name of consumer -2.19 (0.75)** 0.03 0.11 0.48 Neutral attitude Intercept 4.17 (1.13)*** Non-verbal cues 1.87 (0.79)* 1.37 6.45 30.50 Informal elements 0.57 (0.83) 0.35 1.77 9.05 Pronouns -0.19 (1.36) 0.06 0.83 12.2 Signature -0.27 (0.62) 0.23 0.77 2.59 Name of consumer -1.71 (0.72)* 0.05 0.18 0.74
Note. R2 = . .20 (Cox & Snell), .25 (Nagelkerke). Model χ 2
(24) = 61.80, p < .001. * p <
.05, ** p < .01, *** p < .001
Interaction of personalization and timely response
A multinomial logistic regression was conducted to predict consumers’ attitude using personalization, timely response as predictors and an interaction of the two. In the analysis, negative attitude is chosen as reference category. The model shows main effects but no interaction effect (see Table 4). In the main effects model, a fast response was significantly
related to positive attitudes χ2(1)= 5.685, p < 0.05. A test of the full model against a constant model was statically significant, indicating that the predictors reliably distinguished between positive, neutral or negative attitude (χ2(2)= 9.737, p < .01). Both Cox & Snell’s R2 and Nagelkerke’s R2 of 0.04 indicate a weak relationship. Variation in attitude is explained for 4% by the model. The Wald criterion demonstrated that there is no interaction effect between personalization and timely response in both positive (p > .05) and neutral (p > .05) consumer attitudes. Therefore, the assumption that the combination of personalization and timely response results in synergistic effects, which has a bigger effect on consumer’s attitudes than either one of the predictors alone is not supported.
Table 4
Output of the multinomial logistic regression for personalization ✕ timely response
95% CI for Odds Ratio
b (SE) Lower Odds Ratio Upper
Positive attitude Intercept 1.40 (0.59)* Personalization ✕ timely response 0.01 (0.06) 0.90 1.01 1.14 Neutral attitude Intercept 2.79 (0.55)* Personalization ✕ timely response -0.08 (0.06) 0.83 0.92 1.03
Discussion
Consumers increasingly use social media to interact with brands and more brands are implementing webcare every day. If brands use the right kind of elements in webcare, consumers’ attitudes can be increased. This study examined the most effective means for brands on Twitter to gain a positive consumer attitude through webcare. The results of this content analysis partly confirmed H1. Personalization is associated with a more favourable consumer attitude. Making the webcare personal by using the name of the consumer had a positive influence on the consumer. As predicted by Kwon and Sung (2011), non-verbal cues are useful elements in webcare. Non-verbal elements were most effective on a consumers’ attitude when two of the elements were combined. Despite the expectations, pronouns, signature of the brand and informal elements had no significant effect on the consumers’ attitude.
Confirming H2 consumers have a more positive attitude when brands respond in a timely manner compared to when companies respond late. This result is in line with previous research by Huang (2015). It can be stated that a timely response is therefore an important element of webcare.
Finally, the interaction effect between personalization and timely response was examined. Effects similar to H1 and H2 were expected for the interaction effect of
personalization and timely response. However, contrary to the expectations, the results did not indicate a significant interaction effect of personalization and timely response. Therefore, H3 was not supported. We need more research before we can dismiss the interaction effect. The combination of personalization and timely response in relation to webcare must be studied more in the future.
Limitations and directions for future research
First it must be noted that only four brands were investigated in this study. Seeing as all of the brands have fans tweeting them, it might have been an implication that these brands were among the most popular Dutch brands. There are many people who already like Efteling and are tweeting the company as a fan. This also occurred for Bol.com, where consumers tweeted the company for fun instead of making inquiries or complaints.
Another implication regarding the chosen brands is that the image of the company can set the tone for how funny or personal a certain reply can be. For example, Efteling has a young, fun image where notably younger people send tweets to whereas ANWB is a more serious company where most of the tweeters are older and have more serious (car) issues. However, this study would need to be extended with a survey to find out if that is the case. Next to that, Twitter has a 140-character limit at the time, meaning that brands should respond in a short and clear manner. However, Twitters’ direct messages (DM’s) do not have the 140-character limit. One limitation is therefore that brands often ask their consumers to continue their conversation via DM. When having a conversation via DM, it is no longer public. Therefore, this study did not have the opportunity to perfectly code all consumer attitudes, which can explain the predomination of neutral brand attitudes.
One direction for future research is to extend to the study over Facebook and observe the differences of consumer and brand behaviour on both Twitter and Facebook. Seeing that Facebook and Twitter have different brand communities, it might be interesting to find out if there are differences between use of personalization and reaction speed as well as on brand attitudes. Next to that, interactions outside of Twitter are also interesting to study. If different
interactions are studied combined with other methods it could generate interesting results. Also, we need to study it over time to grasp changes in attitudes. In addition, brands
Conclusion
Despite the media attention webcare is getting, there has been little research on how brands can improve consumers’ attitude via webcare. This research attempted to bring in more clarity and contribute to the knowledge of webcare elements in the context of Twitter. The results bring useful outcomes for companies, as well as for academics, knowing that certain
personalization elements and timely responses have a positive influence on the attitude of the consumer as this may help influence consumer behaviour.
References
Amsterdammer wil opslag voor webcare-medewerker Ziggo na bijzondere actie (2016, February 28). AT5. Retrieved from
http://www.at5.nl/artikelen/153277/amsterdammer_wil_opslag_voor_webcare-medewerker_ziggo_na_bijzondere_actie
van Bellegem, S. (2014). The future of the customer relationship: A researched-based vision [Powerpoint slides]. Retrieved from
https://www.slideshare.net/stevenvanbelleghem/the-future-of-the-customer-relationship-35681784/2
Berkeljon, S. (2010, October 27). Imageschade T-Mobile ‘uit de lucht gegrepen’. Volkskrant. Retrieved from http://www.volkskrant.nl/media/imagoschade-t-mobile-uit-de-lucht-gegrepen~a1040967/
Bernritter, S. (2017). Lecture 9: Participating for engagement [PowerPoint slides]. Retrieved from University of Amsterdam, Brands and Organization in Social Media,
Blackboard: https://blackboard.uva.nl/
Retrieved from
http://nos.nl/artikel/2152140-bijna-64-miljoen-zo-veel-passagiers-zag-schiphol-nog-nooit.html
Broekhuijzen, D. (2017). Bedrijven en Influencers op social media: De 10 belangrijkste trends. De Beste Social Media. Retrieved from
http://debestesocialmedia.nl/bedrijven-en-influencers-op-social-media-de-10-belangrijkste-trends/
CBS (2015). Gebruik sociale netwerken sterk toegenomen [Web article]. Retrieved from https://www.cbs.nl/nl-nl/nieuws/2015/27/gebruik-sociale-netwerken-sterk-toegenomen
CBS (2016). Gebruik professionele sociale netwerken stijgt [Web article]. Retrieved from https://www.cbs.nl/nl-nl/nieuws/2016/04/gebruik-professionele-sociale-netwerken-stijgt
Colliander, J., Dahlén, M. & Modig, E. (2015). Twitter for two: investigating the effects of dialogue with customers in social media. International Journal of Advertising, 34(2), 181-194. http://dx.doi.org/10.1080/02650487.2014.996197
Edosomwan, S., Praksan, S.K., Kouame, D., Watson, J. & Seymour, T. (2011). The history of social media and its impact on business. The Journal of Applied Management and Entrepreneurship. 16(3). 79-91.
Elrhoul, M. (2015). Research: Four ways brands can build customer service relationships on Twitter [Blog post]. Retrieved from
https://blog.twitter.com/marketing/en_us/a/2015/research-four-ways-brands-can-build-customer-service-relationships-on-twitter.html
Eptica (2015). Retail 2015 Eptica Multichannel Customer Experience Study. Retrieved from https://www.eptica.com/500_retail_study_2015
SAGE Publications.
Harrison-Walker, L. J. (2001). E-Complaining: A Content Analysis of an Internet Complaint Forum. Journal of Services Marketing, 15(5), 397–412.
Hong, Y. & Lee, W. (2005). Consumer Complaint Behavior in the Online
Environment. In Gao Y. (Eds.), Web System Design and Online Consumer Behavior (90-105). Hershey, PA: Idea Group Publishing.
Huang, W. (2015). New research: consumers willing to spend more after a positive customer service interaction on Twitter [Blog post]. Retrieved from
https://blog.twitter.com/marketing/en_us/a/2015/new-research-consumers-willing-to-spend-more-after-a-positive-customer-service-interaction-on.html
Jansen, L. (2010). Merken in Conversatie op Social Media - Het effect van het aangaan van conversaties op social media op de merkperceptie [Master thesis]. Retrieved from http://www.scriptiesonline.uba.uva.nl/360469
Jansen, B. J., Zang, M., Sobel, K. & Chowdury, A. (2009). Twitter Power: Tweets as Electronic Word of Mouth. Journal of the Association for Information Science and Technology. 60(11). 2169-2188. 10.1002/asi.21149
Kaplan, A., M. & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business Horizons. 53(1). 59-68.
Kelleher, T. & Miller, B. M. (2009). Conversational Voice, Communicated
Commitment, and Public Relations Outcomes in Interactive Online Communication. Journal of Communication, 59. 172–88.
Klant krijgt briefje met liefde van webcare-medewerker (2016, February 28). RTLNieuws. Retrieved from https://www.rtlnieuws.nl/nieuws/binnenland/klant-krijgt-briefje-met-liefde-van-webcare-medewerker
Koot, F. (2013). Organizations with a Human Voice:Antecedents and Consequences of a Conversational Human Voice in the Webcare Responses of Organizations [master thesis]. Retrieved from: http://dare.uva.nl/cgi/arno/show.cgi?fid=503537
Van Kronenburg, E. (2016, February 28). Ziggo-medewerker stuurt verrassing op na vraag van klant. Lindanieuws.nl. Retrieved from: http://www.lindanieuws.nl/nieuws/ziggo-medewerker-stuurt-verrassing-op-na-vraag-van-klant/
Kurt, J., Bakker, F. & Kremers, S. (2014). Brand Asset Valuator 2014 [PDF file]. Retrieved from
https://www.rankingthebrands.com/PDF/BrandAsset%20Valuator%202014,%20Dutc h%20strongest%20brands.pdf
Kwon, E. S., & Sung, Y. (2011). Follow me! Global marketers' Twitter use. Journal of Interactive Advertising, 12(1), 4-16.
Lenoir, A. (2015). Are You Talking to Me? Addressing Consumers in a Globalised World [master thesis]. Retrieved from https://repub.eur.nl/pub/79036
Mangold, W. G. & Faulds, D. J. (2009). Social media: The New hybrid element of the promotion mix. Business Horizons, 52(4). 357-365.
http://dx.doi.org/10.1016/j.bushor.2009.03.002
Nelson, P. (2016, July 7). We touch our phones 2,617 times a day, says study. Network World. Retrieved from
http://www.networkworld.com/article/3092446/smartphones/we-touch-our-phones-2617-times-a-day-says-study.html
van Noort, G., Antheunis, M.L. & Verlegh, P.W.J. (2014). Enhancing the effects of social network site marketing campaigns: if you want consumers to like you, ask them about themselves. International Journal of Advertising, 33(2), 235-252. doi: 10.2501/IJA-33-2-235-252
van Noort, G., Willemsen, L.M., Kerkhof, P. & Verhoeven, J.W.N. (2014). Webcare as an integrative tool for customer care, reputation management, and online marketing: a literature review. In Kitchen, P. J. & Uzunoğlu , E.(Eds.), Integrated communications in the postmodern era (77-99). Basingstoke, Hampshire: Palgrave Macmillan.
Van Noort, G. & Willemsem, L. M. (2011). Online Damage Control: The Effects of Proactive Versus Reactive Webcare Interventions in Consumer-generated and Brand-generated Platforms. Journal of Interactive Marketing, 26. 131–140.
Poelmans, K. (2011). Webcare via social media: hoe snel moet je reageren? [Web article]. Retrieved from https://www.frankwatching.com/archive/2011/12/22/webcare-via-social-media-hoe-snel-moet-je-reageren/
Pfeffer, J., Zorbach, T. & Carley, K. M. (2014). Understanding online firestorms: Negative word-of-mouth dynamics in social media networks. Journal of Marketing
Communications, 20(1-2), 117-128. DOI:
http://dx.doi.org/10.1080/13527266.2013.797778
Sayce, D. (2017). [Blog]. Retrieved from http://www.dsayce.com/social-media/tweets-day/ Schamari, J. & Schaefers, T. (2015). Leaving the Home Turf: How Brands Can Use Webcare
on Consumer-generated Platforms to Increase Positive Consumer Engagement. Journal of Interactive Marketing, 30, 20–33. DOI:
http://dx.doi.org/10.1016/j.intmar.2014.12.001
[Screenshot from Albert Heijn’s Twitter profile]. Retrieved from https://twitter.com/albertheijn/status/847804941976707073 [Screenshot from ANWB’s Twitter profile]. Retrieved from
https://twitter.com/ANWB/status/831057702197542912
[Screenshot from Coolblue’s Twitter banner]. Retrieved from https://twitter.com/coolblue Van Son, E. (2016, February 28). Ziggo-medewerker geeft een beetje liefde aan klant.
Metro. Retrieved from
https://www.metronieuws.nl/nieuws/koffiepauze/2016/02/ziggo-medewerker-geeft-een-beetje-liefde-aan-klant
Thorsteinsson, G & Page, T. (2015). User attachment to smartphones and design guidelines. International Journal of Mobile Learning and Organization, 8(3/4), 201-215. DOI: 10.1504/IJMLO.2014.067020
Trilling, D. (2015). Two Different Debates? Investigating the Relationship Between a Political Debate on TV and Simultaneous Comments on Twitter. Social Science Computer Review. 33(3). 259-276
Twopblog. (2012, May 24). The last 100 million Twitter accounts [Blog post]. Retrieved from http://www.twopblog.com/2012/05/last-100-million-twitter-accounts.html
Watson, S. (2012). Consumer Responses to Service Situations: Tests for
Main and Interaction Effects. Journal of Retailing and Consumer Services, 19(3), 287–96. DOI: https://doi.org/10.1016/j.jretconser.2012.02.003
Zarella, D. (2009). The Science of ReTweets [Informally published work]. Retrieved from http://danzarrella.com/science-of-retweets.pdf
Van Zwienen, S. (2016, February 28). Ziggo-medewerker scoort punten met beetje liefde. AD. Retrieved from http://www.ad.nl/home/ziggo-medewerker-scoort-punten-met-beetje- liefde~a8339c3
A
strongest brands in The Netherlands according to BrandAsset Valuator 2014
Appendix B
CODEBOOK CONTENT ANALYSIS Introduction
The goal of this study is to find out whether personalization and reaction time on brands’ webcare accounts on Twitter have an influence on consumer’s attitude towards the brand. In order to achieve this, the following codebook can be used to measure personalization, reaction time and attitude. Taken together, these variables can be used to analyze possible synergistic effects in regards to webcare strategies on Twitter. Reaction time regarding the tweet is measured in item 8. Personalization is measured in item 9. Finally, consumers’ attitude towards the brands’ reply is measured in item 10. The codebook ends with an emoticon chart with most popular icons and emoji’s used online.
1. Coder ID [__________] 2. Brand 0 – ANWB 1 – Bol.com 2 – de Efteling 3 – Schiphol
3. What was the date of the tweet? (DD/MM/YY) [___][___][____]
4. Conversation language 0 – Dutch
1 – English
2 – Other _____________________________________* *If other, remove from sample. Tweet does not fit the sample criteria.
5. Type of interaction from the consumer This should only fall into one category
1 – Complaint 2 – Compliment 3 – Suggestion 4 – Statement 5 – Other _____________________________________ 99 – Brand reply
6. Is the tweet the beginning of an interaction? 0 – yes
1 – no
Hereby is meant, has there been an on-going conversation before this tweet. It is only a beginning of an interaction if a new subject is introduced and if it is a first response from the company about the subject. In those cases tick “yes”, if there already was a conversation going on, tick “no”.
7. Who is the tweet from? 0 – Brand
1 – Consumer à Go to question 10 8. Reaction time regarding the tweet
Reaction time is measured in whole minutes from the minute the consumers’ tweet was send until the minute the brand has replied to the first, initial tweet of the consumer.
8.1 How many minutes (exact) are between the original tweet and the reply from the brand:
[_______] minutes
Example: If the consumers’ tweet to the brand was send at 9:21 A.M and the company replied at 9:57 A.M, the exact amount of minutes that needs to be filled in is 36 minutes.
9. Personalization
The following questions are about the response of the brand in response to the customer.
9.1 Does the brand write its tweets:
1 – In first-person plural pronoun 2 – In third-person
3 – Other _____________________________________ Examples:
0: “I”, “me” and “my” in English and “ik”, “mijzelf” and “mijn” in Dutch.
“I will look into this problem very quickly. I am sorry for the inconvenience.” 1: “We”, “us” and “our” and “wij”, “ons” and “onze” in Dutch.
“We are looking into this problem very quickly. We are sorry for the inconvenience.” 2: “it” and “Bol.com”
“Bol.com is looking into this problem very quickly. Sorry for the inconvenience.”
9.2 How is the branded tweet signed: 0 – With the name of the employee 1 – With the initials of the employee 2 – With the name of the brand
3 – With the name of the employee and the brand 4 – With the initials of the employee and the brand 5 – The response is not signed
6 – Other _____________________________________ Examples:
0: Chloe or Kevin 1: AvD or PD 2: ANWB
3: Chloe from ANWB 4: AvD from ANWB
9.3 Did the brand use the consumers’ name or Twitter handle 0 – Yes
1 – No Examples:
9.4 Does the brand use any of the following non-verbal cues3 The next variables will be coded as yes/no variables.
9.4.1 … emoticons? 0 – Yes 1 – No 9.4.2 … repeated punctuation? 0 – Yes 1 – No 9.4.3 … capitalization? 0 – Yes 1 – No 9.4.4 … abbreviations? 0 – Yes 1 – No 9.4.5 … intentional misspellings? 0 – Yes 1 – No Examples:
9.4.1: :-) or :( (see emoticons chart for more) 9.4.2: Hey!!! or What??
9.4.3: WOW or COOL (Responses like ‘OMG’ and ‘WTF’ are considered abbreviations, not capitalization)
9.4.4: OMG or LOL 9.4.5: Ca$h or hiiiiyaaa
9.5 Does the brand use any of the following informal elements?4 The next variables will be coded as yes/no variables.
9.5.1 … humour? 0 – Yes
1 – No
3
9.5.2 … references to (pop) culture? 0 – Yes
1 – No
9.5.3 … words that expresses friendship: Mate, bro, homie, (girl)friend 0 – Yes
1 – No
9.5.4 … copy the writing style of the consumer? 0 – Yes 1 – No 9.5.5 … hashtags? 0 – Yes 1 – No 9.5.6 … media? 0 – Yes 1 – No Examples:
9.5.1: Consumer: This morning I gave birth to a food baby and I think @tacobell is the father
Taco Bell (Brand): I want a DNA test.
9.5.2: Consumer: I am thinking of switching to @t_mobile T-Mobile (brand): We welcome you to the Dark Side.
9.5.3: “If you had bought a Whopper today, the second one was free, girlfriend” 9.5.4: Consumer: Y r u neverrrr respnding to me?
Brand: Bc we r veryyyy busy. Wht can we do fr u today? 9.5.5: #WeAreAwake or #Happytohelp
9.5.6: Gifs, images or other media inserted in tweet.
10. Consumers’ attitude towards the brands’ reply
Consumers’ attitude is measured after the brand has offered the consumer a solution to their initial tweet. These are thus the replies from the consumer after the company has replied.
10.1 Favourite count
The amount of people that favourited the tweet. [____________]
10.2 Retweet count
The amount of people that retweeted the tweet. [____________]
Answer the following questions only for the consumers’ reply to the brand after the initial tweet has been answered. If tweet was a brand reply the next questions do not need to be answered.
10.3 Does the consumers’ reply contain any of the elements below: The next variables will be coded as yes/no variables.
10.3.1 …. exclamation marks? 0 – Yes
1 – No
10.3.2 … words that express gratitude? 0 – Yes
1 – No
10.3.3 … words that express happiness, joy (or synonyms to either of them)? 0 – Yes
1 – No
10.3.4 … words that express anger, confusion (or synonyms to either of them)? 0 – Yes 1 – No 10.3.5 … sarcastic intent? 0 – Yes 1 – No 10.3.6 … positive emoticons? 0 – Yes 1 – No 10.3.7 … negative emoticons? 0 – Yes 1 – No
Examples 10.3.1: !!!
10.3.2: Thank you very much
10.3.3: “Perfect, good, nice, happy, great, amazing, whoo”
10.3.4: “Mad, angry, don’t understand, confused, annoyed, frustrated, unhappy” 10.3.5: You’ve been really good at not helping me, thanks!
10.3.6: Emoticons such as :), :D, or <3 (see emoticons chart) 10.3.7: Emoticons such as :(, :s,, :| or :o (see emoticons chart)
10.4 What is the consumers’ sentiment in regards to the brands’ reply
A response can be coded positive (0) when positive emoticons and/or positive words are included as well as exclamation marks. A response can be coded as mixed (1) if there is some positivity included in the response but the consumer is still not happy. A response can be coded as negative (2) whenever there are negative words or emoticons included as well as exclamation marks. A response can be coded as neutral (3) if there was no reply from the consumer, if the reply stated a simple “thank you” or “okay” or if the consumer simply favourited the tweet.
0. Positive 1. Mixed 2. Negative 3. Neutral 99. Cannot decide Examples:
0: This answer helped me a lot – thanks!
1: This helps me a little but I do not have sufficient information to fix my problem. 2: This reply did not help me at all.
Table 1
Emoticons Chart: Used emoticons online. Most icons work backwards too.
Icon Emoji Meaning
:) :] :-) smiley or happy face
:D XD :)) laughing, big grin
:( :c frown, sad
D: D= horror, disgust, angry
;) wink
<3 love (heart)
:P :p tongue sticking out, teasing
:O :-O surprise, shock
:/ :\ :-/ sceptical, annoyed, uneasy
:X :# sealed lips, embarrassed
0:) O:-) innocent, angel
:'( crying
:| straight face
:-S :s worried, confused
