People Are Experience Goods: Improving Online Dating with Virtual Dates

PEOPLE ARE EXPERIENCE GOODS: IMPROVING ONLINE DATING WITH VIRTUAL DATES

JEANA H. FROST, ZOË CHANCE, MICHAEL I. NORTON, AND DAN ARIELY

JEANA H. FROST is a Research Scientist e suggest that online dating frequently fails to meet user expecta- with PatientsLikeMe.com; W e-mail: [email protected] tions because people, unlike many commodities available for purchase online,

are experience goods: Daters wish to screen potential romantic partners by ZOË CHANCE is a graduate student at the Harvard experiential attributes (such as sense of humor or rapport), but online dating Business School; e-mail: [email protected] Web sites force them to screen by searchable attributes (such as income or religion). We demonstrate that people spend too much time searching for MICHAEL I. NORTON options online for too little payoff in offline dates (Study 1), in part because is Assistant Professor of Business Administration at the users desire information about experiential attributes, but online dating Web Harvard Business School; e-mail: [email protected] sites contain primarily searchable attributes (Study 2). Finally, we introduce and beta test the Virtual Date, offering potential dating partners the opportu- DAN ARIELY is Professor of Marketing at nity to acquire experiential information by exploring a virtual environment in The Fuqua School of Business, Duke University; interactions analogous to real first dates (such as going to a museum), an e-mail: [email protected] online intervention that led to greater liking after offline meetings (Study 3). This work is based in part on Jeana Frost’s doctoral dissertation at the Massachusetts Institute of Technology. The authors thank Judith Donath, © 2008 Wiley Periodicals, Inc. and Direct Marketing Educational Foundation, Inc. Andrew Fiore, Shane Frederick, Amit Kumar, Leonard Lee, Jiwoong Shin, Juliana Smith, and Fernanda Viegas for JOURNAL OF INTERACTIVE MARKETING VOLUME 22 / NUMBER 1 / WINTER 2008 their advice and assistance. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/dir.20106 Correspondence concerning this article should be addressed to Jeana H. Frost.

51 Journal of Interactive Marketing DOI: 10.1002/dir

Online dating has emerged as an undoubtedly popu- experiential capability of virtual interfaces. We intro- lar way to meet potential partners: Some 11% of duce the Virtual Date, an intervention in which poten- Internet users—16 million Americans—have used an tial dating partners explore a virtual environment in online dating Web site (Madden & Lenhart, 2006). an interaction analogous to a real first date (such as The advent of online dating has both lowered the bar- going to a museum), creating an online experience rier to initiating contact (from a painful phone call to that offers an experiential preview of a real-world a click of a mouse) and simultaneously increased the interaction. number of available options. Despite this seeming promise, however, results have been decidedly mixed. People Are Experience Goods, Anecdotal evidence and market data suggest wide- Not Search Goods spread user disappointment (Egan, 2003), and gro- wth of the major online dating sites has slowed The distinction between search goods and experience (JupiterResearch, 2005). These trends are particularly goods (Nelson, 1970, 1974) is central to an under- puzzling in light of recent research demonstrating a standing of online consumer behavior. Search goods— generally positive role for the Internet in forming and detergent, dog food, and vitamins—are goods that developing platonic relationships (e.g., Amichai- vary along objective, tangible attributes, and choice Hamburger & Furnham, 2007; Kraut et al., 2002; among options can be construed as an attempt to McKenna, Green, & Gleason, 2002; Nie, 2001; Sproull, maximize expected performance along these measur- Conley, & Moon, 2005). Given the positive effects of able dimensions. Experience goods, in contrast, are the Internet on platonic social life, there appears to be judged by the feelings they evoke, rather than the great potential for the Internet to improve romantic functions they perform. Examples include movies, life as well. perfume, puppies, and restaurant meals—goods defined by attributes that are subjective, aesthetic, We suggest that the failure of online dating sites to holistic, emotive, and tied to the production of sensa- live up to user expectations is due in part to a funda- tion. Most importantly, people must be present to mental gap between the kinds of information people evaluate them; they cannot be judged secondhand both want and need to determine whether someone is (Ford, Smith, & Swasy, 1990; Holbrook & Hirschman, a good romantic match and the kind of information 1982; Li, Daugherty, & Biocca, 2001; Wright & Lynch, available on online dating profiles. Worse yet, as mar- 1995), because indirect experience can be misleading, riage continues to move from an economic exchange causing people to mispredict their satisfaction when arranged between families—based on observable they encounter that choice (Hamilton & Thompson, attributes—to a transcendental falling-in-love experi- 2007). ence—based on intangible emotional attributes—the limited information in profiles is more and more likely We propose that understanding how romantic rela- to be insufficient. Most online dating sites use a tionships are formed online can be informed by situ- “shopping” interface like that used by other commer- ating online dating on the search versus experience cial sites, in which people are classified much like any continuum—or, more specifically, by realizing that commodity, by different searchable attributes (e.g., people are the ultimate experience good. Whether height, weight, income), which can be filtered in any someone joins a dating Web site to find her soulmate way the shopper desires (see Bellman, Johnson, or a one-night stand, success is not determined solely Lohse, & Mandel, 2006). Because determining whether by her partner’s objective qualities (e.g., income and or not one likes someone romantically requires sub- height) but also by subjective qualities, based on jective knowledge about experiential attributes such moment-to-moment rapport between herself and her as rapport or sense of humor, it is perhaps not sur- potential partner. She cannot know, for instance, prising that online daters might be disappointed whether she will find a self-declared comedian funny when they are forced to screen potential partners in person, short of direct experience. Even viewing using objective searchable attributes such as income other’s impressions—via his friends’ comments on his and religion. This evident mismatch between process social networking pages—is not a sure signal that and goals makes dating a fruitful milieu in which to his sense of humor will appeal to her; although know- study online interventions designed to improve the ing others’ opinions may be helpful when pursuing

52 JOURNAL OF INTERACTIVE MARKETING Journal of Interactive Marketing DOI: 10.1002/dir a search good, it is less useful for predicting prefer- is first and foremost an experiential process that is ences for experience goods (Aggarwal & Vaidyanathan, not likely to be captured in online dating profiles 2005). The Internet’s current rules-based filters are (Study 2). Finally, we introduce and beta test our generally geared toward search goods, not experience intervention—Virtual Dates—designed to move the goods, forcing daters to hunt for partners using online dating process away from relying primarily on searchable attributes even when seeking experiential searchable attributes toward providing experiential information—as though they were buying shoes information. We assign participants to view standard online by carefully filtering for brand, price, and online dating profiles or go on Virtual Dates with indi- color—when the one attribute they care most about is viduals with whom they subsequently went on speed fit (see Bhatnagar & Ghose, 2004). dates, comparing impressions formed on Virtual Dates to those created via typical online dating In short, we suggest that user disappointment with (Study 3). online dating is due in part to a crucial mismatch between the experience of online and offline dating. STUDY 1: FRUITLESS SEARCHING IN Online dating follows an information-processing consumer model of choice in which each option has a set ONLINE DATING of features (e.g., height, religion, hobbies) from which In this first study, we wanted to demonstrate the consumers must create an overall impression, analo- inefficiencies of the current Internet dating search gous to attempting to predict the flavor of a packaged process, examining both the sheer extent of searching food based on its nutritional information (grams of required to meet a partner and the dissatisfaction fat, number of calories, amount of fiber, etc.) or its con- that results when people meet others whom they stituent ingredients (vanilla, curry powder, dark have screened using the current online dating search chocolate, etc.). While one might have some sense of mechanisms. how the food will taste, only sampling it can provide an accurate answer. Dating offline, on the other hand, Method and Results involves navigating the world together and sharing ϭ ϭ ϭ experiences, providing opportunities to engage in Participants (N 132; 49 male, Mage 39.4, SD direct interaction and observation, allowing individu- 11.9) completed the survey by following a link on an als to develop an integrated, gestalt impression of oth- online dating Web site. None of the analyses below ers in order to evaluate their relationship potential varied by gender of participants (all ps Ͼ .20); we (Berger, 1979). We have two primary goals in this arti- therefore collapse across gender. cle. First, we attempt to locate the source of some of the dissatisfaction with online dating in the critical We asked participants how many hours per week they gap in information between online and offline dating. spent on the three phases of online dating: searching Second, we introduce and offer preliminary evidence profiles (to find potential matches), e-mailing users for an intervention to bridge this gap in the form of (to screen and arrange meetings with those matches), online Virtual Dates, designed to provide a glimpse and actual face-to-face encounters. Participants of the experiential information about their potential reported spending an average of 5.2 hours per week partners that daters desire. searching through profiles (using the searchable attributes available such as age, income, and educa- tion) and another 6.7 hours writing and responding to OVERVIEW e-mails, all for a payoff of just 1.8 hours of offline In the first two studies, we explore how the current interactions, significantly less than the time spent search engines on online dating Web sites lead to inef- either searching or e-mailing (ts Ն 6.7, ps Ͻ .001); in ficient and unsatisfactory dating outcomes. In Study 1, total, the ratio of search time to interaction time was we chart the general disappointment online daters nearly 7 to 1 (see Figure 1). express with current online dating search mechanisms, documenting how search time invested fails to Engaging in two weeks of research for a four-hour date pay off in a commensurate number of face-to-face might be acceptable if the searching and screening encounters. Next, we investigate our claim that dating process were itself a fun activity, but we find that

PEOPLE ARE EXPERIENCE GOODS 53 Journal of Interactive Marketing DOI: 10.1002/dir

8 STUDY 2: PEOPLE ARE EXPERIENCE GOODS 6 To support our account that part of the dissatisfaction 4 with online dating Web sites stems from this mismatch, we asked young single people—the modal online dater—what qualities they considered most Hours per week 2 important when choosing someone to date and when

0 choosing someone to marry, and classified responses Screening E-Mailing Actual as either searchable or experiential traits. Were daters profiles interaction to list primarily searchable attributes, then a tool like Figure 1 an Internet search agent or a rules-based filter would Online Daters Spend Seven Times as Many Hours Screening be appropriate for identifying partners. Should they Profiles and E-Mailing Potential Partners Than in Actual Face- list a significant number of experiential attributes, on To-Face Interactions (Study 1) the other hand, then online dating interfaces likely fall short of providing the information their users desire. We expected, of course, to confirm the intuitive online daters find the search process aversive. We notion that people would report desiring much more asked participants to rate how much they liked three experiential than searchable information about activities: online dating, “offline” or face-to-face dating, potential partners. and, as a comparison, how much they enjoyed watching movies, all on 10-point scales (1 ϭ not at all to 10 ϭ very much). On average, participants were less sat- Method and Results ϭ ϭ ϭ ϭ ϭ isfied with online dating (M 5.5, SD 2.3) than Participants (N 47; 37 male, Mage 24.2, SD 7.7) offline dating (M ϭ 7.0, SD ϭ 2.2), paired t(131) ϭ 6.2, were unmarried individuals who completed this sur- p Ͻ .001. Sadly, though offline dating beat online dat- vey online as part of an unrelated series of experi- ing, most participants were so dissatisfied with both ments. As in Study 1, none of the analyses varied by they would have preferred to stay home and watch a participant gender of ( ps Ͼ .20); we therefore collapse movie (M ϭ 7.8, SD ϭ 2.2), paired ts Ն 3.1, ps Ͻ .01. across gender. Nor was this dissatisfaction with online dating merely an artifact of our having unsuccessful daters in our We asked participants two open-ended questions: sample; in fact, our participants reported having When choosing someone to date, what qualities do you arranged an average of 7.6 dates via an online dating consider most important? and When choosing someone site, and this number was positively correlated with to marry, what qualities do you consider most impor- satisfaction with online dating, r(125) ϭ .21, p Ͻ .03. tant? Participants listed similar numbers of qualities for dates (M ϭ 3.5) and spouses (M ϭ 3.6), t Ͻ 1. Two While we suggest that the lack of satisfaction with independent coders, blind to the hypotheses, then online dating is due in part to the amount of search coded these lists into searchable and experiential time involved, we do not have data on time spent traits. They were instructed to define a trait as arranging offline dates, which likely also involve searchable if one could judge the presence of this char- much more search time than payoff. However, our acteristic from an online dating profile and as experi- account suggests that it is another factor—the mis- ential if some personal interaction would be required match between the kinds of information people wish to judge the presence of this trait. Intercoder agree- to know and the information available online—that ment was 86%. drives dissatisfaction with online dating. Even if the amount of time and effort required to arrange a date Not surprisingly, participants listed more experiential online and offline are the same, the lack of experien- traits for potential marriage partners than potential tial information available may still lead to greater dating partners (Ms ϭ 3.0 and 2.6), t(46) ϭ 1.97, p Ͻ disappointment even in those cases when efforts to .06: the more serious they were about living with find a date pay off in a cup of coffee. someone, the more they wanted to know what that

54 JOURNAL OF INTERACTIVE MARKETING Journal of Interactive Marketing DOI: 10.1002/dir

4 (Viegas & Donath, 1999), to create Virtual Dates, an Searchable interface in which pairs of potential daters navigate a Experiential 3 visual environment together, exchanging real-time messages with images as triggers for conversation. In particular, we hoped to show that the images would 2 prompt partners to discover shared interests and sim- ilarities, an important predictor of liking (Byrne, 1

Number of Attributes 1971). Previous research has demonstrated the benefits of adding image content (e.g., Churchill, Snowdon, 0 & Munro, 2001; Kraut, Fussell, & Siegel, 2003; Dating Marriage Whittaker, 2003) with such artifacts serving as social Figure 2 catalysts to stimulate conversation (Karahalios, Single People Prefer Experiential Attributes to Searchable 2004). Virtual Dates are therefore designed to bridge Attributes by a Ratio of Three to One When Searching for Dating the gap between offline and online dating by increas- Partners and Five to One for Marriage Partners (Study 2) ing social presence in online interaction (see Walther, 1996), thus simulating a first date in the real world. experience might be like. Most importantly for our account, participants listed more experiential than Our design of Virtual Dates as a tool for helping peo- searchable traits in both dating and marriage ple locate better matches for the most experiential domains: nearly three times more for dating partners good of all—people—was informed by the efforts of (experiential M ϭ 2.6, SD ϭ 1.4; searchable M ϭ .9, other Internet retailers who more generally have SD ϭ .8), t(46) ϭ 7.1, p Ͻ .001, and five times more for attempted to improve the odds of finding a good potential spouses (experiential M ϭ 3.0, SD ϭ 1.9; match with experiential goods by simulating the searchable M ϭ .6, SD ϭ 1.9), t(46) ϭ 8.8, p Ͻ .001 offline shopping experience using a more experiential (see Figure 2) . While searchable traits included items interface (see Alba et al., 1997; Steckel et al., 2005). such as “religious,” “athletic,” and “physically attrac- With My Virtual Model, for example, consumers cre- tive”—attributes that one could use to filter options ate an online model of their body that they then use online—the vast majority of traits listed were experi- to test how garments may look when they try them on ential and required more interaction than merely at home (Nantel, 2004, and see Klein, 2003). Indeed, viewing someone’s dating profile, such as “makes me such interactivity has been shown to increase favora- laugh,” “understands me,” “will get along with my bility toward both products and Web sites (Sicilia, other friends,” “affectionate,” “loyal,” and “fun.” Ruiz, & Munuera, 2005). Virtual Dates constitute an effort to bring these strategies to online dating, allow- STUDY 3: VIRTUAL DATES ing people to try each other on to see whether they fit, simulating their lives together before they ever meet. Given results from Studies 1 and 2, how might we After participants met on a Virtual Date, we hosted a improve the online dating process? People seem to speed-dating session during which they met potential spend too much time conducting searches based on partners face-to-face. Each participant met his or her attributes they do not value highly, failing to connect Virtual Date partner and one individual for whom with people for offline dates, and finding the few they had read an online profile, allowing us to compare offline dates they do schedule to be unsatisfying. the success of Virtual Dates to standard online dating. Study 1 suggests that while people find both offline and online dating difficult, offline dating is preferred, while Study 2 demonstrates people’s desire for more METHOD information about what their partner will be like to ϭ ϭ ϭ “try on” in person. Can we bridge the gap between Participants (N 24; 12 male, Mage 27.1, SD 5.1) current online dating practices and the experience of were recruited on an online dating Web site we created meeting someone in person, before daters go to the for the Massachusetts Institute of Technology commu- trouble of actually scheduling an offline date? We nity, thematchup.net. Participants completed the first used an evolving enriched chat platform, Chatcircles two sections of the study—reading one profile and

PEOPLE ARE EXPERIENCE GOODS 55 Journal of Interactive Marketing DOI: 10.1002/dir

Pre-event session

Virtual Date (with one participant)

and Speed-dating event Recruitment View profile Meet for speed-dates Sign up on Website (of another participant) (with both participants)

Day 1 Day 5–6 Day 7 Figure 3 Timeline of Events (Study 3) going on one Virtual Date—at their home or office by dating sites. They contained responses to multiple- logging on to a preassigned Web site. One to two days choice questions regarding occupation and school sta- later, all participants met face-to-face during a four- tus, religion, relationship goals, and desire for children, minute speed date. After the speed date, participants as well as a username, headline, and an open-ended answered four questions about their partner: How personal essay. much do you like this person? How similar to you is this person? How excited are you about this person, and how Speed Dating. One to two days after reading pro- comfortable do you feel with this person? All questions files and going on Virtual Dates, participants then were answered on a 10-point scale (1 ϭ not at all to took part in a speed-dating event. Participants were 10 ϭ very.) (See Figure 3 for a timeline of events.) seated in a room with tables arranged in a horseshoe shape; women sat on the outside of the horseshoe and men in the center, with each woman facing a man. Types of Contact Partners had four-minute unstructured conversations Virtual Dates. Participants logged onto the appli- with each of the 12 members of the opposite sex, cation using a preassigned URL, which randomly including both the person whose profile they had assigned them to an opposite-sex partner. Participants viewed and the person they had Virtual Dated. entered a virtual environment in which they were represented by a simple avatar—a circle in a color of Results and Discussion their choice. The circle expanded to fit participants’ utterances, typed in a text box below, then shrank as Interactions with Virtual Dates. Not surprisingly the quotes disappeared, leaving a subtle trace. given that these individuals had little information Participants could navigate the environment by mov- about each other, most conversations (75%) contained ing their circles together or separately, chatting about some questions about the other person’s demograph- the images displayed in the environment (e.g., Lisa ics, or searchable traits (age, occupation, religion, etc.). Simpson and Jessica Simpson, George Bush and John We were most interested, however, in how shared Kerry). Each Virtual Date lasted for 15 minutes (see experience—in people’s acquisition of knowledge Figure 4 for two sample interactions). The environ- about experiential traits—developed in the real-time ment visually modeled shared experience by includ- Virtual Dates interface. In order to understand user ing the rule that participants could only “hear” what experiences, one research assistant and one of the their partners were saying (i.e., read the text bubble) authors coded the text from these interactions on a if they were near one another (i.e., if their avatars number of dimensions described below—agreement were close together on screen.) across these classifications was high (91%).

Personal Profiles. We used the personal profiles par- First, we coded for any instance in which the features ticipants had created on our online dating site, which of the interface itself created conversation, including were similar in structure to the profiles on major comments about ease of use, design features (e.g., the

56 JOURNAL OF INTERACTIVE MARKETING Journal of Interactive Marketing DOI: 10.1002/dir

M: do you like to dance? D: yes, waltz? M: does that mean you like also disco freestyle? D: haha....i don't know how to disco freestyle M: the big easy is a fun place to go to dance....you been there? D: yes, i liked it. would be fine too! what would be important to you, before you would go there with a chatroom acquaintance?

A: Do you like dancing? S: I like dancing even though I’m not good at it. I also really like techno music. What music do you like? A: From classical to dancing S: Cool. I like classical ...

Figure 4 Two Samples of Virtual Date Interactions Spurred by Image Content (Study 3) shape and movement of the avatars), or the text box. to signify closeness—one user wrote, “You lead, and I For instance, one user, on finding that both she and her will follow” to which her partner replied “:-)”. In all, date had selected the color green for their avatar, 42% of conversations included specific reference to remarked “interesting that we both picked green.” moving through the environment. Another user, referring to the movement of the avatars, said, “you are dancing circles around me,” to which her Finally, we explored whether these factors combined to partner responded, “like Fred Astaire—if he were a red help people to develop a sense of one another’s person- circle.” In all, 58% of conversations included exchanges alities while they discovered shared interests: we coded centered on aspects of the chat environment. for any conversation in which one person asked the other a question about a preference (e.g., his or her Second, we coded for instances in which users chatted opinion about a piece of art) or hobby (e.g., whether he about the specific setting we chose for their Virtual or she liked some activity), in which the partner res- Date—an interface designed to resemble an art ponded positively. Some 58% of conversations specifi- gallery—which we hoped would spur conversation. cally built off the interface to uncover common likes This seemed to be the case. One user commented, and dislikes. One pair, for example, after wandering “I see we stumbled into an art gallery . . . ,” to which together to the picture of the tangoing couple, dis- her partner replied, “Yesterday I was at the MFA cussed their favorite styles of dance (waltz, jitterbug [Museum of Fine Arts]; never thought this would be stroll, jazz, hip-hop, disco freestyle) then discussed analogous.” Most importantly, the interface allowed their recent experiences at their favorite club, at which partners to coordinate movement through this space point one user wrote, “Yes, I liked it. What would be in real time, as though on a real date, and we there- important to you [to know about someone], before you fore coded for mentions of shared decision making. would go there with a chatroom acquaintance?” This One person wrote “Care to wander over to the couple moved all the way from a comment spurred by a Simpsons?” while another pair, when one person sug- picture to the intimation of a possible date. gested going toward Lisa Simpson and the other toward a picture of two people tangoing, wrote, “We can Overall, participants enjoyed the interface, making do both—Lisa then tango?” They even used movement comments such as “this is strangely addictive,” “this is

PEOPLE ARE EXPERIENCE GOODS 57 Journal of Interactive Marketing DOI: 10.1002/dir

hilarious,” “I’m having fun,” “it’s pretty cool,” and “I love highly correlated with this measure of interest, r ϭ .50, it.” Most importantly, their comments indicated that the p Ͻ .001. Thus, the increases in liking we observed on interface was helping them to form connections: even Virtual Dates appear to be increases in romantic lik- though each participant knew he or she would meet this ing, not merely in platonic liking. Virtual Date shortly at the speed-dating session, three out of the twelve couples were so eager to talk more that they exchanged e-mail addresses—and one even But would the positive impressions formed during exchanged telephone numbers—as their Virtual Dates these brief interactions persist after participants left ended. the speed-dating session? While Sunnafrank and Ramirez (2004) showed that a brief interaction with a Liking. Did the impact of these enriched interac- same-sex stranger predicted liking months later, only tions and discoveries of common ground carry for- following our opposite-sex Virtual Dates couples fur- ward through these couples’ face-to-face interactions ther forward through time would allow us to state during speed dating? We calculated a composite mea- conclusively that Virtual Dates lead to relationship sure of the four items that participants completed formation. Indeed, while participants were able to after speed dating—liking, similarity, excitement, and uncover more experiential information about their comfort (Cronbach’s a ϭ .85)—to create an overall partners during their Virtual Dates, and this led them measure of liking for their partner. Participants liked to like each other more after meeting, we cannot say their Virtual Dates partner significantly more that these more positive impressions are more accu- (M ϭ 5.7, SD ϭ 1.7) than they liked the individual rate or lead to greater long-term compatibility with- whose profile they had read (M ϭ 4.5, SD ϭ 1.5), out similar follow-up data. It is possible, for example, t(30) ϭ 2.1, p Ͻ .05.1 These findings support previous that people merely perceive themselves to be more results that people who engage in standard online compatible as a result of using the interface, but are dating (i.e., profile searches) show decreases in liking incorrect in this assumption; future testing can after face-to-face encounters (Norton, Frost, & Ariely, explore whether such interfaces lead to more accurate 2007); our results suggest that Virtual Dates may impressions and greater compatibility in addition to decrease this trend. leading to more positive impressions. Finally, we did not test how Virtual Dates would compare to other Follow-up Study. We were primarily interested in existing types of online interactions (e.g., chat rooms forming successful dating relationships, not just friend- or e-mail exchanges). We therefore cannot with cer- ships. E-mail address exchanges, along with some flir- tainty say they are superior to these other formats— tatious dialogues (“Didn’t know that’s what would turn only that participants enjoy both on computer scientists at MIT”/“you’d better not know the interface and their interactions in that interface. what turns people on here . . . what’s a turn-on for Future research is needed to explore which combina- Harvardians?”) might suggest that interest was more tion of technologies is most successful for improving than platonic, but to test more rigorously whether the online dating. differences in liking between reading profiles and virtual dating were associated with increased romantic We have discussed Virtual Dates as a means of uncov- interest, we conducted another speed-dating study ering the kinds of experiential information not avail- (N ϭ 24) in which we both asked people to rate how able in traditional online dating. A different stream of much they liked their partners (1 ϭ not at all to research, however, has suggested that one of the issues 10 ϭ very much), and also asked them to characterize with computer-mediated communication (CMC) in their overall impression of their partner by selecting general is that it allows precisely what we are claiming among three options: not interested in seeing the person it helps to correct: careful construction of one’s online again, interested in this person as a possible friend or persona, leading to distorted impressions (see Walther, professional contact, or interested in going on a date 1996). Of course, many of the attributes that are with this person. As expected, ratings of liking were untestable online are similarly untestable in person (Wood, 2001)—loyalty, for example, is a highly desirable 1 Due to the time constraints of the speed-dating session, not all characteristic in a mate, but it can only be revealed participants completed the dependent measures. over time, and even real offline first dates offer an

58 JOURNAL OF INTERACTIVE MARKETING Journal of Interactive Marketing DOI: 10.1002/dir opportunity to present one’s “best self” by dressing life encourages people to exist solely online, having carefully and avoiding certain topics of conversation. brief and meaningless acquaintanceships and then While it is true that Virtual Dates allow for more con- moving on, a potentially disastrous breakdown in trolled self-presentation than face-to-face interac- social capital (Putnam, 2001). Online relationships tions, they are at minimum less controllable than a are not necessarily fleeting, however, and online life static online dating profile, which can be carefully has enormous potential to foster social connections, honed and revised over weeks and months; on Virtual creating entire communities of social support (see Dates, users have only seconds to construct responses Rheingold, 1993). Each year, more people are drawn to their partners’ quips and queries. Compared to to the virtual communities teeming with life online, face-to-face interaction, then, Virtual Dates may per- and some even “emigrate” there: the current popula- mit some distortion, but compared with standard tion of virtual worlds in games such as Second Life, online dating, they permit much less. World of Warcraft, and EverQuest is estimated at more than 20 million people, and 20% of these gamers claim the online realm as their “primary place of res- GENERAL DISCUSSION idence” (Castronova, 2005). The online interactions that players have with one another often lead to real- Online dating facilitates introductions that would life friendships and romance, with a reported 29% of otherwise be nearly impossible to generate, in both female gamers dating someone they met in a game the sheer number of options available and the relative (Yee, 2003). Many players, however, do not seek real- ease with which contact can be initiated with those world interaction with one another because they find options. Thus, online dating Web sites—and online com- the online experiences so satisfying in themselves; the munities more generally—are highly successful in average online role-playing game player spends 22 allowing people to connect with others outside of their hours a week in his or her virtual world (Yee, 2003). existing social circles. The present research suggests, These data suggest that whether the impact of the however, that when people are searching for romantic Internet on social life is positive or negative depends relationships, as with other experience goods, online on the specific goals of the user in a specific domain. information seeking can leave much to be desired. Earlier, we suggested that people are the ultimate Study 1 demonstrated that people do not enjoy online experience good, and that when finding a soulmate, dating as much as offline dating, or even watching only face-to-face interaction will do. The requirements movies—in part because the time and effort invested for successful relationships can vary widely, however; in searching is not rewarded with positive real-world while for some a greater degree of direct experience is payoffs. Next, we confirmed that the qualities people essential—as with online dating—for others, such as desire most in a partner are qualities that are diffi- online gamers, current interfaces may be sufficient.2 cult to identify using an Internet search engine but must be experienced in an interaction (Study 2). In Managerial Implications Study 3, we introduced an intervention designed to improve the online dating experience, Virtual Dates, Virtual interaction interfaces will become an increas- which simulated a first date, allowing couples to ingly important marketing device over the next interact in real time. Participants’ reactions were decade, particularly for goods and services in which more positive toward individuals whom they had experiential attributes are important (see Deighton & Virtual Dated than toward those whose profiles they Kornfeld, 2007). Videos, virtual tours, and customer had seen, effects that carried forward through an ini- testimonials are improving consumers’ ability to judge tial face-to-face meeting in a speed date. their potential purchases, and new digital scent technology can already heighten the sense of being there

Life, Virtual and Otherwise 2 We should also point out that in some cases, the removal of face- Is inserting online interfaces into social life helpful or to-face interaction that online life engenders is the very aspect that offers benefits (see Fogg, 2003), from simulating risky sexual deci- harmful, on balance? We have focused on ways in sion making (Downs et al., 2004) to disrupting normal group inter- which online dating can be beneficial, but some actions in an effort to improve performance (Norton, DiMicco, researchers have expressed the concern that online Caneel, & Ariely, 2004; see Kiesler & Sproull, 1992).

PEOPLE ARE EXPERIENCE GOODS 59 Journal of Interactive Marketing DOI: 10.1002/dir

(see digiscents.com). Thus, great strides have already kinds of information they desired. For instance, if been made in matching people with products using daters express a greater desire to know about potential virtual interfaces. In the domain of relationships— partners’ sense of humor than their intelligence, a suc- from dating relationships to friendships—virtual cessful interface would incorporate technology that worlds are increasingly central in the formation and would more easily allow humor to emerge than intelli- development of social relationships. Thus, more gen- gence (for example, logging in to watch and comment erally, matching people with people will be one of the on clips from The Office). In addition, gathering infor- critical areas in which virtual interfaces will become mation from consumers on those attributes they increasingly common (see also Nambisan & Baron, consider most important allows for customization. If a 2007). As just one example, companies can use such particular dater cares more about intelligence than virtual interfaces to give interviewers the opportu- sense of humor, offering an interface that allows this to nity to test interviewees’ skills in a simulated work be revealed (e.g., a jointly completed crossword puzzle) situation. would be more effective. Thus, in designing any interface, starting by asking consumers which kinds of We suggest that crucial to designing successful virtual information they desire—and focusing on what their interfaces is careful consideration of (a) the different end goal might be—should not just inform but deter- goals of the many kinds of relationships online and mine what information is most easily accessible. (b) the different kinds of information people value most highly or find most diagnostic when forming these different kinds of relationships. REFERENCES Aggarwal, P., & Vaidyanathan, R. (2005). Perceived effec- Different Relationship Goals. For those relation- tiveness of information agent routines: Search vs. expe- ships formed and maintained solely online (as in rience goods. International Journal of Internet Marketing and Advertising, 2, 38–55. Second Life), situations and contexts need not be Alba, J., Lynch, J., Weitz, B., Janiszewski, C., Lutz, R., reality-bound; in online dating, or in contexts in which Sawyer, A., et al. (1997). Interactive home shopping: business relationships may move from conference Consumer, retailer, and manufacturer incentives to par- calls to shared workgroups, however, getting a sense ticipate in electronic marketplaces. Journal of Market- of what it might be like to meet a potential partner ing, 61, 38–53. face-to-face may be paramount, making interfaces Amichai-Hamburger, Y., & Furnham, A. (2007). The positive that model the real world more desirable. Indeed, the net. Computers in Human Behavior, 23, 1033–1045. prototype design of Virtual Dates that we introduced Bellman, S., Johnson, E. J., Lohse, G. L., & Mandel, N. (2006). in Study 3 was specifically designed to overcome the Designing marketplaces of the artificial with consumers in mind: Four approaches to understanding consumer behav- ways in which current search mechanisms thwart ior in electronic environments. Journal of Interactive daters from achieving their goals that we identified in Marketing, 20, 21–33. Study 1, by mapping the online dating experience Berger, C. R. (1979). Beyond initial interaction: Uncertainty, more closely to the goals that daters have. For man- understanding, and the development of interpersonal agers, a better understanding of the specific goals of relationships. In H. Giles & R. St. Clair (Eds.), Language consumers in different online social domains—from and social psychology (pp. 122–144). Oxford, UK: Basil finding a bridge partner to meeting one’s true love— Blackwell. can inform the design and implementation of search Bhatnagar, A., & Ghose, S. (2004). Online information mechanisms to better map the different search needs search termination patterns across product categories and consumer demographics. Journal of Retailing, 80, in these different domains. 221–228. Byrne, D. (1971). The attraction paradigm. New York: Different Kinds of Information. In addition to Academic Press. mapping closely to the different goals that different Castronova, E. (2005). Synthetic worlds: The business and relationships elicit, successful interfaces should also be culture of online games. Chicago: University of Chicago designed to provide the different kinds of information Press. that these different relationships require. Our Study 2 Churchill, E. F., Snowdon, D., & Munro, A. (Eds.). (2001). revealed not just an overall view of the information Collaborative virtual environments: Digital places and consumers want about partners, but also what specific spaces for interaction. London: Springer Verlag.

60 JOURNAL OF INTERACTIVE MARKETING Journal of Interactive Marketing DOI: 10.1002/dir

Deighton, J., & Kornfeld, L. (2007). Digital interactivity: Nantel, J. (2004). My virtual model: Virtual reality comes Unanticipated consequences for markets, marketing, into fashion. Journal of Interactive Marketing, 18, 73–86. and consumers. Working Paper, Harvard Business Nelson, P. (1970). Information and consumer behavior. School, Cambridge, MA. Journal of Political Economy, 82, 4, 729–754. Downs, J. S., Murray, P.J., Bruine de Bruin, W., White, J. P., Nelson, P. (1974). Economic value of advertising. In Palmgren, C., & Fischhoff, B. (2004). Interactive video Y. Brozed (Ed.), Advertising and society (pp. 109–141). behavioral intervention to reduce adolescent females’ New York: New York University Press. STD risk: A randomized controlled trial. Social Science Nie, N. H. (2001). Sociability, interpersonal relations, and and Medicine, 59, 1561–1572. the Internet: Reconciling conflicting findings. American Egan, J. (2003, November 23). Love in the time of no time. Behavioral Scientist, 45, 420–435. New York Times. Norton, M. I., DiMicco, J. M., Caneel, R., & Ariely, D. (2004). Fogg, B. J. (2003). Persuasive technology: Using computers AntiGroupWare and Second Messenger: Simple systems to change what we think and do. San Francisco: Elsevier. for improving (and eliminating) meetings. BT Ford, G. T., Smith, D. B., & Swasy, J. L. (1990). Consumer Technology Journal, 22, 83–88. skepticism of advertising claims: Testing hypotheses Norton, M. I., Frost, J. H., & Ariely, D. (2007). Less is more: from economics of information. Journal of Consumer When and why familiarity breeds contempt. Journal of Research, 16, 433–441. Personality and Social Psychology, 92, 97–105. Hamilton, R. W., & Thompson, D. V. (2007). Is there a sub- Putnam, R. D. (2001). Bowling alone: The collapse and revival stitute for direct experience? Comparing consumers’ of American community, New York: Simon & Schuster. preferences after direct and indirect product experi- Rheingold, H. (1993). The virtual community: Homesteading ences. Journal of Consumer Research, 34, 546–555. on the electronic frontier. Reading, MA: Addison-Wesley. Holbrook, M. B., & Hirschman, E. C. (1982). The experiential Sicilia, M., Ruiz, S., & Munuera, J. L. (2005). Effects of inter- aspects of consumption: Consumer fantasies, feelings, and activity in a web site—The moderating effect of need for fun. Journal of Consumer Research, 9(2), 132–140. cognition. Journal of Advertising, 34(3), 31–44. JupiterResearch. (2005). Consumer online dating behaviors: Sproull, L., Conley, C. A., & Moon, J. Y. (2005). Prosocial 2003–2005. Jupiter Media. behavior on the net. In Y. Amichai-Hamburger (Ed.), The Karahalios, K. G. (2004). Social catalysts: Enhancing com- social net: Understanding human behavior in cyberspace munication in mediated spaces. Unpublished doctoral (pp. 139–161). New York: Oxford University Press. dissertation, Massachusetts Institute of Technology, Steckel, J. H., Winer, R. S., Bucklin, R. E., Dellaert, B. G. C., Cambridge, MA. Drèze, X., Häubl, et al. (2005). Choice in interactive envi- Kiesler, S., & Sproull, L. (1992). Group decision-making and ronments. Marketing Letters, 16, 309–320. communication technology. Organizational Behavior and Sunnafrank, M., & Ramirez, A. (2004). At first sight: Human Decision Processes, 52, 96–123. Persistent relational effects of get-acquainted conversa- Klein, L. R. (2003). Creating virtual product experiences: tions. Journal of Social and Personal Relationships, 21, The role of telepresence. Journal of Interactive 361–379. Marketing, 17, 41–55. Viegas, F., & Donath, J. (1999). Chat circles. Paper present- Kraut, R. E., Fussell, S. R., & Siegel, J. (2003). Visual infor- ed at the Conference on Human Factors in Computing mation as a conversational resource in collaborative Systems. physical tasks. Human Computer Interaction, 18, 13–49. Walther, J. B. (1996). Computer-mediated communication: Kraut, R., Kiesler, S., Boneva, B., Cummings, J., Helgeson, Impersonal, interpersonal, and hyperpersonal interac- V., & Crawford, A. (2002). Internet paradox revisited. tion. Communication Research, 23, 3–43. Journal of Social Issues, 58, 49–74. Whittaker, S. (2003). Things to talk about when talking about Li, H., Daugherty, T., & Biocca, F. (2001). Characteristics of things. Human Computer Interaction, 18, 149–170. virtual experience in electronic commerce: A protocol Wood, S. (2001). Remote purchase environments: The analysis. Journal of Interactive Marketing, 15(3), 13–30. influence of return policy leniency on two-stage deci- Madden, M., & Lenhart, A. (2006). Online dating. sion processes, Journal of Marketing Research, 38, Washington, DC: Pew Internet and American Life Project. 157–169. McKenna, K. Y. A., Green, A. S., & Gleason, M. (2002). Wright, A. A., & Lynch, J. G. (1995). Communication effects Relationship formation on the Internet: What’s the big of advertising versus direct experience when both search attraction? Journal of Social Issues, 58, 9–31. and experience attributes are present. Journal of Nambisan, S., & Baron, R. A. (2007). Interactions in virtual Consumer Research, 21, 708–718. customer environments: Implications for product sup- Yee, N. (2003). Inside out. The Daedalus Project 1–2. port and customer relationship management. Journal of Retrieved March 27, 2008, from http://www.nickyee.com/ Interactive Marketing, 21, 42–62. daedalus/archives/000523.php

PEOPLE ARE EXPERIENCE GOODS 61