From Discovery to Purchase: Improving the User …872947/FULLTEXT01.pdfFrom Discovery to Purchase...

66
From Discovery to Purchase: Improving the User Experience for Buyers in eCommerce JASPER HAHN KTH ROYAL INSTITUTE OF TECHNOLOGY INFORMATION AND COMMUNICATION TECHNOLOGY

Transcript of From Discovery to Purchase: Improving the User …872947/FULLTEXT01.pdfFrom Discovery to Purchase...

  • From Discovery to Purchase: Improving the User Experience for Buyers in eCommerce

    JASPER HAHN

    KTH ROYAL INSTITUTE OF TECHNOLOGY I N F O R M A T I O N A N D C O M M U N I C A T I O N T E C H N O L O G Y

  • From Discovery to PurchaseImproving the User Experience for Buyers in eCommerce

    JASPER HAHN

    Stockholm 2015

    Software Engineering of Distributed SystemsSchool of Information and Communication Technology

    Kungliga Tekniska Högskolan

  • AbstractThe Internet has revolutionized many areas of our lives. New forms of exchanging and retrievinginformation, making business and communication in general have been made possible with theInternet and have gone through a rapid development since its creation. In an age of nearly ubiqui-tous access to the Internet and a majority of the western world actively using social media, retailmarkets have changed, too. But compared to the rapidly changing services in other sectors, retailbusinesses have only converted an existing model to a new technology rather than coming up with anew one. Social Commerce is an approach that wants to change that. It takes into account lessonslearned from social media and shifting marketing strategies and tries to create a better shoppingexperience for customers while giving brands and fashion influencers a new platform to engagewith them. This thesis project uses literature from different fields such as interaction design, on-line marketing and fashion along with user interviews to identify the most important aspects thatwill lead towards a more social online shopping experience, particularly in fashion. It is conductedin collaboration with the local start-up Apprl (www.apprl.com) and includes an implementationpart of realizing the identified most promising features as part of the agile development processwithin the company. The field of social commerce is promising to radically change the way we buythings online and Apprl is one of many examples trying to make that happen.

    i

  • SammanfattningInternet har revolutionerat många områden av våra liv. Nya former för utbyte och hämtning avinformation, att göra affärer och kommunikation generellt har blivit möjligt med Internet och hargått igenom en snabb utveckling sedan sitt skapande. I en tid med nästan allestädes närvarandetillgång till Internet och en majoritet av människor i västvärlden som aktivt använder sociala me-dier har också detaljhandelsmarknaden ändrats. Men jämfört med de snabbt ändrande tjänsternai andra branscher har detaljhandelsmarknaden bara konverterat en redan existerande modell tillen ny teknologi istället för att hitta på en ny. Social Commerce är ett tillvägagångssätt som villändra detta. Det tar hänsyn till erfarenheter från sociala medier och skiftande marketingstrate-gier och försöker att skapa en bättre shoppingupplevelse för kunder medan det ger varumärkenoch modepåverkare en ny plattform att engagera sig i. Detta examensarbete använder litteraturfrån olika områden som interaktionsdesign, online marketing och mode tillsammans med använ-darintervjuer för att identifiera de viktigaste aspekterna som vill leda mot en mer social onlineshoppingupplevelse, speciellt för mode. Arbetet är genomfört i samverkan med det lokala startup-företaget Apprl (www.apprl.com) och inkluderar en implementeringsdel där de funktioner som harblivit identifierade som mest lovande realiseras som en del av den agila utvecklingsprocessen hosföretaget. Området för Social Commerce är lovande för att radikalt ändra sättet vi köper sakeronline och Apprl är ett av många exempel på företag som försöker att få detta att hända.

    ii

  • Contents1 Introduction 1

    1.1 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

    2 Background 32.1 eCommerce and Fashion today . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

    2.1.1 Customer Journey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.2 Personal Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.3 Fashion Blogs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.1.4 Mobile Phones and Shopping . . . . . . . . . . . . . . . . . . . . . . . . . . 52.1.5 Affiliate Marketing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

    2.2 Apprl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.3 Additional Reading: Fashion, Commerce and Online User Experiences . . . . . . . 62.4 Related Products and Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

    2.4.1 Closetspace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.4.2 Rêve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.4.3 Tipser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.4.4 Shopsense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

    3 Identifying Opportunities 93.1 Brainstorming and Inspiration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

    3.1.1 Facilitating the Shopping Process . . . . . . . . . . . . . . . . . . . . . . . . 93.1.2 Motivation and Loyalty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.1.3 Social Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113.1.4 Other Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

    3.2 Feedback from Users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133.2.1 Form of Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133.2.2 Selecting Peers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

    3.3 Analysis of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.3.1 Results from Interviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.3.2 Selecting Ideas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163.3.3 Dropped ideas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

    3.4 Planning next Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

    4 Identified Opportunities 214.1 Making content exploration more social . . . . . . . . . . . . . . . . . . . . . . . . 214.2 Visual Overhaul of eMails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224.3 Relevant eMails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224.4 Notification UI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.5 Tying the user into the ecosystem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.6 Smaller Visual Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254.7 Further Planning for Apprl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

    4.7.1 Goals, Achievements and Encouragement . . . . . . . . . . . . . . . . . . . 254.7.2 More Complex Notifications . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

    4.8 Summing Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

    iii

  • 5 Implementing New Features 315.1 Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

    5.1.1 Scrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.1.2 Specific Strategies and Arrangements . . . . . . . . . . . . . . . . . . . . . . 32

    5.2 Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325.2.1 Django . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335.2.2 Bootstrap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335.2.3 Mandrill API . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

    5.3 Theory and Reality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345.3.1 Possible Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

    5.4 Personal Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

    6 Evaluation and Outlook 366.1 Statistical Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

    6.1.1 New Tabs on Looks Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366.1.2 Redesigned and New eMails . . . . . . . . . . . . . . . . . . . . . . . . . . . 386.1.3 Only Time Will Tell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

    6.2 Feedback from Peers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386.3 Evaluation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406.4 Improving Data Gathering for Future Analysis . . . . . . . . . . . . . . . . . . . . 40

    6.4.1 Available Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406.4.2 Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416.4.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

    6.5 Next Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

    7 Conclusion 437.1 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437.2 Distinction of Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437.3 Closing Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

    A Interviews 48A.1 Interview Outline and Peer Selection . . . . . . . . . . . . . . . . . . . . . . . . . . 48A.2 Interview Task and Additional Questions . . . . . . . . . . . . . . . . . . . . . . . 49A.3 eMail Invitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50A.4 Mockups for Feature Ideas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51A.5 Quantifying Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

    B New Features 55B.1 Latest Looks and Friends Looks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55B.2 Visual Templates for eMail Notifications . . . . . . . . . . . . . . . . . . . . . . . . 56

    C Tracking User Statistics 57

    iv

  • List of Figures1 Customer Journey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Idea Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 Latest Looks Tab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 Liked Looks Tab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 eMail Template Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 Notification UI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 Evaluation: Latest and Friends Looks . . . . . . . . . . . . . . . . . . . . . . . . . 378 Evaluation: eMail Redesign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399 Mockup: Notepad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5110 Mockup: Black Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5211 Mockup: Look Slotmachine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5312 Quantified Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5413 Additional Comparison: Latest and Friends Looks . . . . . . . . . . . . . . . . . . 5514 Product Sale eMail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5615 Comparison of Tracking Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

    v

  • List of Tables1 Overview of Feature Ideas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 Detailed eMail Click Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

    vi

  • 1 IntroductionThe Internet has influenced our society and significantly changed how we perform common tasks.Digital messages exchanged on the Internet overtook traditional mail in relevance a long time agoand today fewer people even call others on their phones than just a few years ago. In fact, thosewho make calls are more and more likely to place those over the Internet instead of using a plain oldphone number. Meanwhile, students would be at a complete loss if they had to do their homeworkassignments without having access to the Internet. Even the software tools we use are starting tobe in the cloud instead of locally run applications.

    However, while our lives are certainly heavily influenced by our online activities, not all partsof our lives have seen the same improvements by the use of technology. Certain behavioral pat-terns have almost exclusively shifted into digital - think about taking photos with your phone orstreaming music to your tablet - other common interactions have stayed true to their traditional,physical nature at least for major parts of society. Even other activities were merely adapted tonew technologies without seeing decisive changes in the way they are carried out.

    One of those adapted activities that virtually everyone engages in is shopping. Online storesbasically operate with the same model that has been used by mail order businesses for decades.The catalogs and newsletters merely moved from the physical mailbox to a web site and your eMailinbox. Shopping as a spare time activity rather than a chore is especially relevant for clothing orfashion as the corresponding industry wants consumers to refer to their products [1]. Clothing haslong since evolved from its utilitarian purpose to a statement. No matter whether you only buya new pair of pants to replace a broken pair or you have a closet overflowing with clothes someof which still have the price tags on, we all have an idea of what the specific garment should looklike to properly present ourselves the way we want to be seen. Everyone has that certain ideaabout their preferred style when they walk into a store. But wait a minute! Walk into a store?Yes, that is indeed still happening. Brick-and-mortar stores are struggling with competition fromonline stores but they are not likely to vanish any time soon.

    Living in a time of social networks and practically ubiquitous access to the Internet, our so-cial interactions might be transforming but they still have a grave influence on our behaviors asmembers of society and likewise our behaviors as customers. Harnessing the meaning of socialinteraction for business is not a staggeringly new idea, as masses of literature about social mediamarketing suggest. However, a true integration of the process of shopping for clothes with thesocial interactions we willfully engage in has yet to become a common pattern. Social commerce,identified as a trend e.g. by Wall Street Journal in 2011 [2], may not be replacing traditionalmarketing models but it is destined to become more influential in the near future; understandingit and adapting to it offers opportunities for business and customers alike. Some business modelsmay, however, become obsolete along the way as they are neither robust enough nor adapting tonew realities that value personal recommendations over direct advertising.

    Apprl is a distinguished example of a business built around the idea of social commerce. Basedin Stockholm, the young start-up has been elected the 14th-hottest start-up in Sweden by theranking on internetworld.se1 in late 2014 [3]. More recently Apprl was awarded as “Best ServiceWithin Social Shopping” in Sweden [4]. The company is still small, but constantly spreading itsreach.

    This thesis presents the existing services and explore a vision for the future based on inputdirectly from existing or prospective users of Apprl’s services. The first steps towards that future,both theoretical and practical, were taken during the course of the thesis.

    1http://internetworld.se/

    1

    http://internetworld.se/

  • 1.1 GoalBuilding on existing ideas around social commerce, the practical part of this thesis was meantto develop, or rather contribute to, a promising approach to making online shopping more social.Valuable background research and active exploration of user needs were to be used as a solid basisfor the carried out work. The resulting features would be available to the public on Apprl’s existingplatform and could thus be used by a vast number of users. Even though economic success wasnot a primary goal of the thesis, improving the experience users have on the social commerceplatform is likely to be beneficial for the company as well as their users. Along with the exemplarydevelopment for this one particular platform, the findings and considerations described throughoutthis report is supposed to give useful insights into what has been, is and might be happening alongthe the lines of a shift towards being more social in eCommerce as well as online marketing.

    In addition to the mentioned outcomes, the process applied during planning as well as devel-opment shows an example of small-team software development projects and thus contributes tothe field of software engineering by describing a realistic work environment for software projects.It can be useful as a reference for setting up software projects or serve as guidance for soon-to-begraduates of the field or others planning a career in the field.

    1.2 ScopeUsing a real-world application as a basis for this thesis is an interesting topic for observation initself. The project work carried out was a study of everyday life and work in a Swedish start-upas much as it was an analysis of and active contribution to the future eCommerce. Throughoutthis report, the supervising company plays an important role which can be seen as exemplary forthe tech start-up scene in Sweden. Practical and theoretical work lie close together in the field ofsoftware engineering which makes the intersection a rather interesting topic for observation.

    1.3 Thesis StructureThis thesis starts out with a more detailed introduction to the current state of eCommerce, fashionand their interconnection in particular in chapter 2. Furthermore the chapter introduces the objectof contemplation: the services offered by Apprl. It is rounded off with some insights into furtherreading and related projects. The first phase of the project work is detailed in chapter 3; itdescribes the process of gathering information and analyzing the findings which is essential forthe main contribution of the thesis that is covered in the subsequent chapters. The findings wereused to gather feedback in interview sessions which aree described and analyzed in chapter 3. Theconcepts for changes are outlined in chapter 4 before giving an insight into the development processin which the results were converted into actual code for the existing platform in chapter 5.

    The main contribution of this work in the mentioned chapters. The contribution consists ofseveral additions and refinements to the services of Apprl which are used to exemplify the possiblefuture of eCommerce in a real world setting. Those results are briefly evaluated and an outlookfor future development is given as an additional contribution in chapter 6 before a summary offindings and contributions is made in the concluding chapter 7.

    2

  • 2 BackgroundFor a general overview in the areas of fashion, marketing, online shopping and fashion blogging,this chapter covers the most important aspects of with an emphasis on their intersection. Overthe course of two weeks, the existing service, related research and other opportunities were studiedin detail. Additional literature helped getting more insights and the gathered knowledge fromdifferent research areas was used to generate useful input for the subsequent process of improvingthe experience on Apprl.com. Information provided here helps better understand considerationsmade throughout the rest of the thesis.

    Some initial thoughts that could evolve into feature ideas served as useful outlines and weredocumented along the way and used for the following process described in chapter 3 .

    2.1 eCommerce and Fashion todaySome key terms that are useful for the understanding of the field are explained here. Since thisthesis is carried out in the field of IT, theoretical deliberations are kept to a minimum.

    2.1.1 Customer Journey

    The basic steps of a user’s online shopping process were identified as the customer journey whichis depicted in figure 1. The customer journey describes the different steps and phases a customerpasses through during her interaction with a business in the context of shopping. On this rathercoarse level three stages can be identified: discovery, consideration and purchase. The term "cus-tomer journey" has traditionally been used to describe the actual movement of a customer in astore and various adaptions, often a lot more elaborate than the one used here, have been proposed[5]. For the given context the depicted model serves as a solid basis for discussion.

    Customers in fashion are more likely to buy again in the near future than customers for ex-ample in the housing market or people buying a car. Fashion is outdated quickly and even moreimportantly the barrier to owning more instead of simply replacing a piece is very low. Fashionbuyers want to express themselves through what they buy and wear. Cassidy et al. for exampleanalyzed the motivation to buy fashion for youth in particular [6]. The incentive to dress welland show appreciation for new trends can be based on personality traits or simply be a form ofcomplying to social norms, as explained by R. Michon et. al [7].

    The market for fashion, one can conclude, is thus broad and steadily moving. This particularobservation calls for the consideration of one very important aspect of the customer journey: thepost-purchase experience. Building a relationship with a customer means more than just sellingone product. Returning customers are what keeps a business alive and growing and only satisfiedconsumers will become loyal to a brand or business. Both retailers and brands have an interestin engaging users after a purchase is made, to build up loyalty. D. Court et. al identified theimportance of this aspect in what they refer to as the loyalty loop [8]. Loyalty of customers iscategorized into lazy and active loyalty by D. Court et. al , with only the latter ensuring a strong,lasting relationship with a customer.

    Projected onto the case of Apprl, 3 out of 4 phases of the customer journey are supportedthrough the existing services. During the course of the thesis, the importance of the differentphases are identified and corresponding strategies for improvement are presented.

    2.1.2 Personal Recommendations

    The importance of personal recommendations has always been high. Its availability, however, hasrisen dramatically over the course of the last years with the emergence of social networks and other

    3

  • Figure 1: The different phases a customer goes through during (online) shopping.

    modern communication channels. Marketers refer to personal recommendations as word-of-mouthor earned media and the industry has identified them as the most influential factor in the decision-making process of today’s consumers. The global information and measurement company Nielsen2frequently publishes reports about consumers’ trust and the influence of different channels on theirbuying behavior. Subsequent reports on trust in earned media have shown a steady increase overthe years [9, 10]. Nielsen’s analysis consistently places the influence of earned media above thatof conventional advertising, also referred to as owned media. Owned media describes all kinds ofadvertising space or air time paid for by companies such as TV commercials or banner ads onwebsites. Another marketing analyst firm, Crowdtap3, identified customers’ primary influence forbuying presents during the Christmas season of 2014 and found that two thirds of shoppers valuedrecommendations on social networks the highest [11].

    In short, consumers are interested in other people’s opinions about products and actively eval-uate choices based on that input.

    On a related note, a lot of research has been conducted on automated recommendation systems([12, 13, 14, 15, 16]). Some even focused on fashion [17, 18]. However, this thesis and the philosophyof Apprl value the personal aspect a lot higher than automated "pseudo-personalization". Recom-mendations are part of the experience, but the social connections were deemed more relevant thanthe recommendation of individual products for the time being. The entire field of recommendationsystems is thus not analyzed in more detail.

    Traditionally, fashion has been a field in which few leaders or innovators inspire other people andcause a trickle-down effect from fashion enthusiasts down to the mainstream markets as explainedby Barnes [19].

    2.1.3 Fashion Blogs

    Taking one step back from recommendations by personal peers, the role of certain influencers canbe identified as highly relevant for consumers. Barnes take on social media’s relevance for thefashion industry identifies a change in who those influencers are in the field of fashion [19]; whileit used to be a domain of fashion journalists, it has shifted towards fashion bloggers as the primarysource of input for consumers with a special interest in fashion.

    Along with the rise of fashion bloggers, online communities have been built around fashionand shown a significant impact on consumer behavior. Studies have adapted the community ideatowards the blogosphere and its readership. For example in 2014 Pihl identified the question "[...]how communities forming around sets of brands, rather than a brand, can be explained." [20]. His

    2http://www.nielsen.com/3http://corp.crowdtap.com/

    4

    http://www.nielsen.com/http://corp.crowdtap.com/

  • results state that beyond "appreciation" and "loyalty", the readers of fashion blogs also developa sense of "close relation" to their favorite bloggers. Pihl’s findings prove how influential fashionbloggers are in the region in which he carried out his research. That region happens to be Sweden,the same country this thesis and Apprl’s biggest part of operations are carried out in. Takinginto account recent viewer statistics, the influence of Swedish fashion bloggers remains high. As ofMarch 2015, the top 10 professional blogs are all beyond 100.000 readers per week and even 8/10private blogs are exceeding that number, according to Bloggportalen[21], Sweden’s biggest servicefor blog rankings.

    2.1.4 Mobile Phones and Shopping

    Retailers across all sectors have noticed a rise in mobile phone usage inside their stores in recentyears. Mobile users tend to compare prices and buy online with their always-connected phones.But that doesn’t necessarily mean the death of brick-and-mortar stores according to recent studiesby McKinsey [22] and UPS [23]. A recent study found that 56% of shoppers are using their phonesto research products online, yet 58% end up buying at the brick-and-mortar store eventually [22];the negative impact might thus be less severe than store owners may intuitively guess. McKinsey’sreport does, however, point out that including the mobile segment into one’s strategy is not straight-forward and has cost a lot of retailers big amounts of money for very disappointing results. Theyprovide some positive examples as well and one of the main take-aways is this: buyers don’t wantmore apps, they want a convenient experience which is likely best delivered by the means of awell-optimized mobile website.

    On- and offline shopping go hand-in-hand. Giving consumers tools that can combine bothpractices will be important in the future and one should at least be aware of this co-existence whenconsidering fashion consumers and their experience.

    2.1.5 Affiliate Marketing

    The term affiliate marketing described the marketing practice of paying per generated sale insteadof a fixed rate or a rate of impressions or clicks. Back in 2006 Duffy analyzed its benefits overthe other mentioned payment models [24]. He describes the concept as a "win-win relationship"because all involved parties benefit from the same triggered user behavior. Affiliate links can beplaced on different places of a website or within eMail marketing. Although technically possible,the payment model is not usually paired with banner ads or other purely commercial content.Successful uses of affiliate market often rely on more personal and/or relevant combinations ofprimary content and linked products or services. Examples include tech bloggers linking to asmart phone they reviewed or a fashion blogger linking to the dress she is wearing in the pictureshe posted.

    2.2 ApprlAs briefly introduced in chapter 1 Apprl specializes in making eCommerce social, also referred toas social commerce. The business is built around the idea that content created by peers to whicha user has a personal relation is many times more meaningful that conventional advertising in theform of banner ads.The service is built around two main principles:

    • Engaging fashion bloggers with the service and supporting their creation of high qualitycontent, providing income for both the bloggers and Apprl iself through affiliate marketing.Bloggers can sign up on Apprl to become publishers in order to start earning money.

    5

  • • Providing a social platform for users and bloggers to interact with products and each otherthrough creation of and communication about content.

    As of the start of this thesis, the first principle had been the focus of operations. There are,however, intersections between these two goals in terms of features of the service, which is subjectto the investigation throughout this thesis. Several ways of curating and creating content whichcan then be displayed on other channels are the core features of Apprl. Those main principles areoutlined briefly.

    Content Creation Users can use product images to create content called looks. The availablekinds of looks are tagged photos or collages of product photos. For either one the user can interactwith looks in the form of clicking shown products to buy them or finding similar items through anoverlay that appears when hovering over the products (or touching them on mobile devices).

    Content Curation In addition to actively creating looks, all products liked by a user are collectedwithin her personal store page. Additionally, curated stores can be created manually to onlyshowcase a certain selection of products. (Remark: the curation feature had been scheduled andwas introduced shortly after work on this thesis started)

    Embedding Content on Blogs Looks or entire shops, based on products a user has liked (see be-low), can be embedded into other websites such as blogs. The required HTML code is automaticallygenerated and does not require technical skills from the user.

    Shopping Users can browse the vast amount of products that is aggregated from different onlineretailers who are part of the Apprl network. Search and filter options help the user to navigateand find products. The same applies for content a user finds embedded on third party websites.

    Social Interaction Users (both regular and publishers) can interact on apprl.com. Looks can bepublished for other users to see and their individual stores can be accessed. Users can follow otherusers and like products or looks. Liked products automatically show up in a user’s personal store.

    Sharing Content on Other Channels Other social networks can be used to share content orproducts created on apprl.com. This way, fashion-minded users on Apprl can communicate theirstyle further beyond the scope of Apprl’s user base.

    2.3 Additional Reading: Fashion, Commerce and Online User ExperiencesThe background reading and preparation covered a vast amount of different fields and while someliterature was mainly used to get a better understanding of the required fields, a lot of resourcesprovide information that has been used throughout the work and writing of this thesis.

    An inspiring read that definitely challenged the way certain problems were thought about was"Seductive Interaction Design" by S. P. Anderson [25]. As will become clear in chapter 4, fewchanges were actually made concerning the interaction. The insights from this book have playedtheir fair part in re-considering certain drastic changes other literature might have enthusiasticallyrecommended because it laid out psychological reasoning for design aspects that went well beyond"looking pretty". A very thorough analysis of fashion blogging, especially in Sweden, has beenconducted by C. Pihl from the School of Business at Stockholm University [20]. His insights arevaluable to better classify the role of fashion in our society. A few inspirations for feature ideas,especially about motivation, loyalty and gamification were based on understandings gained by

    6

  • reading Loyalty 3.0 by R. Paharia[26] and Y. Chou’s Actionable Gamification [27] respectively.Both books take a rather analytic approach to the topic and explain the why instead of only givingsuggestions about how to apply their principles. The reports and analysis articles by both McKin-sey and Nielsen were helpful resources and often inspired further reading beyond the scope of thisthesis. Topics as the mentioned importance of personal recommendations are best understood andmost valuable when looking at statistics they provide rather than predictions without supportingdata as referenced in section 2.1.2. For the general importance of social media it is important torecall how many people are actively using it today. Numbers from The Swedish Internet Infras-tructure Foundation show that 47% of Sweden’s population is using Facebook daily and 95% inthe age group 16-25 is at least using it sometimes [28].

    2.4 Related Products and ServicesBy now the reader should have an overview of the underlying motivations for creating productsthat focus on fashion, online shopping, social commerce or related topics. In fact, the marketplacefor online services and mobile applications has seen huge interest in the area and identifying allrelevant and competing concepts would be far beyond the scope of this thesis. However, this sectionpresents a few hand-picked examples of tools that use the dynamics identified earlier on. The listis not supposed to be complete but tries to be as representative as possible for the existing marketand relevance for Apprl’s own target audience.

    2.4.1 Closetspace

    A very large and successful service is Closetspace [29]. Their service is based on offering ready-made "looks" that include links for buying the displayed products but they can also be saved forlater once the user signed up. Closetspace is the main product of a company called Stylitics, basedin New York City. Their main business strategy is selling insights about customer behavior tobrands and fashion outlets. The current offering on Closetspace includes a wider range of featuresand is supposed to serve as a "central hub" for fashion-oriented people. Users can catalog productsthey own in their digital wardrobe and get recommendations for looks based on what they alreadyown; recommendations are even taking into account their local weather forecast. As an additionalpaid feature a personal stylist can be purchased on a per-month basis.

    While the stated intent is to provide looks from fashion experts such as bloggers, the selectionpresented on the website is curated by brands or shops themselves and makes the impression ofvery generic collections of products whose sole unifying characteristic is the prominent placementof a certain product by that brand. Closetspace is not offering social interaction features such asfriending or following other users. Revenue is solely generated for the company itself by means ofselling items and users purchasing access to premium features.

    2.4.2 Rêve

    Solely focused around brands and products, the Rêve mobile apps for iOS and Android are builtaround a community principle similar to Apprl’s [30]. The start-up company is based in Stockholmand their product went live in May 2014. Rêve’s focus clearly lies on mobile but the company alsooffers a web-based version of their service. People are encouraged to comment on products and"reve" them, which can be considered their version of the common "like". A clear differentiatingfactor is that the service unifies the concept of brands and shops. Nearly all brands are in factconnected to their own web shop with one notable exception being Zalando. The brand or shoplogo is prominently placed next to each product unlike on Apprl’s service where the shop namesare only mentioned in subtle text hints on product detail pages.

    7

  • While the general concept of buying products is comparable to that on Apprl, some shops havedirect integration for the mobile applications. One encountered example was that H&M productscould be ordered straight from the application. Other shops as well as all products clicked onRêve lead to the shop websites exactly as they do on Apprl. All users are treated as customersonly without any opportunities for creating content or even earning money. Content creation orcuration features are not part of the service.

    A Notable mention is that one of team member at Rêve, Xavi Gratal, is combining his work atthe company with his PhD studies at KTH in the fields of computer vision and artificial intelligence.

    2.4.3 Tipser

    Also based in Stockholm, Tipser could be Apprl’s closest competitor [31]. Their service appearsto replicate all features Apprl offers in an equally minimalist, yet different design. Based on anintroduction video the user learns about the possibilities the service has to offer. The entire websiteis in Swedish and there does not seem to be any way to access information in English. Unlike theother services mentioned before Tipser is designed as an affiliate marketing platform for contentcreators such as bloggers (even though the term blogger is not specifically stated). The presentedoptions for setting up one’s own store page are vast and suggest that putting up a shop should bethe main priority for Tipser’s users. Embedding looks, in the form of tagged photos, is an optionmentioned on the site but was not actually discovered during a brief investigation of the service.

    The most notable differentiating feature is that Tipser offers a streamlined checkout processthrough their own site. Products are ordered directly on their site and Tipser handles all furthertransaction with the underlying retailers.

    Trying to sign up for the service unfortunately resulted in an error message, written in Swedish,whenever trying to log in4. However, signing up through Facebook worked. The interface isdesigned around the same actions of exploring and liking as found on Apprl. However, it puts theemphasis on products instead of users. A feed of recently created shops exists but no looks or waysto follow other users were to be seen. There are no customization options and no pointers as to howone creates the actual shop. Once the button for generating the embed code was found and thecode tried in a local file it could not load the store in a browser. According to the debug console inChrome a certain "userid" was not found. Some of the additional features such as tagging productsin photos seem to be exclusively supported in blogg.se. (Remark: One press release, discoveredlater, actually states that Tipser is geared towards blogg.se users. The actual web site does notstate that at all.)

    2.4.4 Shopsense

    Shopsense [32] serves its’ users with access to a vast selection of products that can be used to create"widgets" which can be embedded on other websites. The widgets can have logical layouts suchas different grids and are highly configurable, yet not comparable to the collages Apprl offers as"looks" due to the lack of creative freedom. Shopsense works like a traditional affiliate network butadds a lot of possibilities for dynamically creating different sizes and variations of their widgets. Inaddition to a highly polished and feature-rich interface and well-designed yet minimalist sharingcontent, Shopsense offers detailed statistics and simple analysis tools to track user earnings. Whilefavorite products can be saved for a user’s account similar to the other presented services, no socialinteraction is possible. In fact, other users and their information can not be accessed at all.

    Shopsense is the affiliate program to Shopstyle [33], which helps users create sleek looking storesbased on products and brands they like but does not generate any income in itself.

    4Tested on May 15th

    8

  • 3 Identifying OpportunitiesIn order to produce relevant additions to the existing service later on it was necessary to applya structured approach to the exploration and subsequent narrowing down of new ideas. After aninitial broadening of the considered opportunities, they were grouped into promising areas andsummarized into more meaningful ideas. Additional insight was gathered through interviews withexisting and prospective users and eventually led to a final assessment of the initial ideas.

    Throughout the rest of this chapter the background and fundamental opportunities are ex-plained and the ideation process is described. Steps were taken according to their order of oc-currence, but details were added for clarification when necessary rather than in order of theirinception.

    3.1 Brainstorming and InspirationThe process started out with a few ideas that had already come up during a meeting before theactual start of the thesis work. Those ideas were very general in nature and consisted of a broadrange of possibilities from very small interface optimizations to actively aiding the user during thecheckout process. Extensive reading for a better understanding of the field and the interactionwith related services, both covered in the previous chapter, helped in the process of generatingnew ideas. Those ideas were collected and described very briefly to serve as basis for discussionbetween the author of this thesis and representatives of Apprl.

    Considerations about which areas were starting points for several of the ideas serve as the guidethrough this section. It introduces the concepts that were identified as possible improvements foreach individual area.

    3.1.1 Facilitating the Shopping Process

    One concern that came up during the initial conversations with Apprl was that the checkout processis known in the industry to be a very critical part of the shopping process. A study by KISSmetricsfor example showed that in many cases more than 40% of the customers drop out after the first stepof a checkout process and concluded that anything that makes checkout easier and/or faster has asignificant impact on completion rates [34]. While Apprl is not an online store itself and thus doesnot control the checkout process, the business model is still based around finished purchases. Oneof the biggest hurdles for users that had been identified was the requirement to create an accountor logging in to an existing one. Based on that a few ideas were designed around facilitating thatprocess for users. These ideas included:

    • Auto-Login: Store account information or build connections to stores and automaticallylog in users when they want to purchase.

    • Registration/Checkout Aid: Keep certain information for the users such as shippingand billing information and offer a way to conveniently insert that information into thecorresponding form without manual effort.

    • Navigation Bar: Keep the familiar top bar from Apprl.com when users are sent to anexternal site, that could offer above features as well as navigation, liking the product orsimilar.

    Along the way, other subtle adjustments to the actual purchase were considered that dealt withreducing steps required for the user. Only one parameter proved so essential for nearly all productsthat it was added to the list: the size for a garment. Not only does the user have to select a size

    9

  • once on the product page on retailer’s website, certain sizes might also be sold out, so that a usercould filter out wrong sizes beforehand, when using the Apprl shop (on Apprl.com or embedded).That information should then be passed on to the retailer to eliminate the need for size selection.

    3.1.2 Motivation and Loyalty

    A look at existing user statistics tells a clear story: usage drops a few months after users create theiraccounts. But how could that be? New features have constantly been added, the tools becamemore polished and the number of available products was not only steadily increasing, but alsokeeping up to date with upcoming trends. So how would we keep users engaged with the servicefor more than a few months? In addition to keeping up the initial motivation it seemed importantto create loyalty towards the service. Section 2.1.1 identified a customer’s loyalty towards fashionbrands as an important factor for marketers’ considerations. Here, however, we were looking atthe loyalty towards Apprl.com as a product.

    Initially, it is important to give the user a motivation to start using your service at all. There-after, the experience must appeal to the user. Just because you measured that a peer user canfinish a given task in "under x seconds" you do not know whether given the opportunity the userwould want to experience the interaction again. Giving users a positive experience is what reallymakes an online service successful.

    Previous efforts by Apprl included competitions about look creation according to a certain topicor based on a brand. Those were mainly meant to attract new users to the platform and actuallyyielded positive results in that regard. However, each resulting spike in activity faded out withinweeks after those special events had taken place.

    The reason users were initially engaged when competitions were introduced is easy to under-stand. In efforts to create more pleasant experiences a lot of attention has been paid to the topicof gamification within recent years. Competitions are just one manifestation of applied gamifica-tion. After both successful and failed attempts at it have been made, approaches such as Chou’sOctalysis [35] are trying to outline the essence of what actually makes gamification useful. Beingdelighted by an experience is generally what it boils down to, not conventional game elements.Anderson also touches on the topic and makes a valuable point in his chapter about "The Gameof Seduction" when he writes:

    A game first has to be fun and engaging - without the points and badges that get somuch attention; a simple reward schedule - however addictive - leads to frustration ifpeople don’t enjoy the activity being reinforced. In contrast, good games bring joy topeople’s lives.

    Why is it possible to spend hours transfixed on a screen with no awareness of thepassing time? Why do people enjoy games? And can we create the same emotions withthings that are not games? (S. P. Anderson, Seductive Interaction Design, p. 148 [25])

    One needs to understand that a game in these terms does not necessarily have to be something weintuitively identify as such. It can be anything we enjoy doing. Anderson encourages designers anddevelopers to find game-like characteristics inherent in an application before adding new featuresbased on the gamification principles. Looking into what already made Apprl enjoyable to use aswell as what might make it even more enjoyable were two very interesting questions for both thebrainstorming and the interviews.

    Improving the subjective joy a user experiences can be done in various ways. It is essential,however, that it fits into its context to achieve the desired goal. That might for example meanmaking a common task easier and more convenient to perform or sometimes simply putting a smile

    10

  • on someone’s face. Anderson refers to MailChimp’s5 slightly rude little maskot and cites how only0.007% of the users actually activated the "party pooper mode” that eliminated the seeminglyunprofessional but certainly delightful remarks made by the maskot.

    These principles had to be considered in regard to Apprl’s existing service and supported byfindings detailed in the previous chapter, content creation, social interaction and shopping weredeemed pleasant activities in themselves. These core features are of utmost importance for theservice. It is however not trivial to add more exciting layers to the existing interaction withoutdistracting from those main parts of the experience. Two important pieces to the puzzle areidentified by Paharia as mastery and progress, the subjective feeling of a user that she improvesand becomes better at something by means of achieving certain goals. Adding meaningful goalsand honoring those in the right way is a delicate topic and has been considered with great caution.It is not without reason that Chou’s book Actionable Gamification carries the subtitle BeyondPoints, Badges, and Leaderboards. Thus, meaningful concepts are required in this area and thecreated ideas were meant to have exploratory character rather than being final concepts.

    In addition to direct interaction, staying up to date with what other users are doing is veryinteresting for users. Reminding users about recent activities could increase their loyalty if it piquestheir interest. Prior to this thesis some very straight forward notifications were being sent out viaeMail. Those notifications were solely text-based and thus carried little visual appeal. Notifyingusers could also serve as reminders to users who have started to forget the service and would thusbenefit the experience in more than one way. In order for notifications to more accepted, though,their visual appeal had to be improved.

    Concluding these deliberations, once a general motivation for using your product is ensured, itis important to keep that momentum up and keep your users excited over the long term; createloyalty. When measuring success of a website, sign-up rates are definitely an important indicatorbut long-term engagement is what keeps a business running. Looking at the activity rates of usersand identifying what users seem to find motivating in the short and long term were identified asimportant areas for investigation during the feedback sessions. This excerpt describes some of theideas that could serve as improvements in the areas of short- and long-term motivation:

    • Unlocking features as reward for certain behavior (number of likes, amount of content created,etc.)

    • Improved aesthetics of notifications/alerts (someone likes your content, an item is on sale,etc.)

    • Bring notifications/alerts into the application in addition to eMails

    • Useful and/or playful new features (see designs in Appendix A.4)

    • Customizable shops6

    • More features for look creation

    • Notify when items are sold out, so that a look could not be completely purchased any more

    3.1.3 Social Interaction

    Apprl likes to refer to their services with the term social commerce. Unfortunately, at the time ofstarting this thesis Apprl had rather little to offer in terms of social interaction. The clear business

    5http://mailchimp.com/6Curated Shops and some customizability for embedded shops had already been scheduled without the author

    knowing about it at this stage. The concept was included in the subsequent feedback analysis nevertheless.

    11

    http://mailchimp.com/

  • goal had been to support publishers with their work and generate sales. Those publishers builttheir own fashion communities, as described in 2.1.3, and are using Apprl mainly as a tool and lessas a platform. Including "regular" users in the interaction had hitherto been a lower priority butwas supposed to be tackled, which was part of the motivation behind this thesis. As described inthe previous section, user motivation is an important parameter for success. And interacting withother users and comparing yourself with others has a big impact on motivation as Anderson notes.Establishing Apprl.com as a community platform is one of the future goals of the company andwill properly further exploit the benefits of social commerce on a more personal level.

    Besides the plain possibility of interaction, its value often lies in the subjective impact a userhas. This closely ties in with findings presented in the previous section about motivation (3.1.2).Paharia identifies purpose as one of the five key intrinsic motivators in his introductory chapterabout motivation [26]. He defines purpose as the feeling of "I make a difference", an aspect that ismost notably observed by the culture of "likes" and "re-posts" ("-tweets", "-pins", ...) on varioussocial media platforms. People who are used to social media are accustomed to receiving likesand relevance is determined by what gets likes more than anything else. Nilforoush provided justone of many posts about the craze surrounding the "Like" in 2014 [36]; the number of other rantsabout the "like culture" is immense. Entire businesses are built around suggesting content thatwill generate likes, as demonstrated by Content Captain7, a German start-up. Giving every userthe chance to participate in this culture and giving him the feeling of those small victories, doingsomething that matters is an important piece of the social interactions on modern websites. It is,however, important to give everyone a chance to be recognized for the content she shares and notto make the mistake of establishing only all time rankings in which an average user will never showup. Both Anderson and Chou make this point about not being too short-sighted about certainelements of gamification.

    There was a huge spectrum of ideas surrounding this topic and thus, no full list is given aboutthe possibilities that were considered during this stage. Furthermore, the insights won in thesubsequent interviews (section 3.2) were deemed very important to pick the most promising ideasfrom the pool and refine them accordingly.

    One area identified as deserving improvements was visibility : who can see your content? Vis-ibility of content could be modified in various ways and directions and only further input couldsteer in which. On one hand, exposing more users to your content, thus increasing your visibility,would satisfy a user’s drive for purpose. On the other hand, certain content might only be meantfor certain people’s eyes when privacy is concerned. Facebook’s "close friends" list comes to mindin this regard.

    Another key issue is figuring out who to "follow" in the first place. Online communities mayhave a certain level of intersection with real life acquaintances but as the likes of Twitter andInstagram showcase not everyone you follow needs to be someone you have met personally in orderto be of particular interest to you. It seems important to give new users better ways of findingpeers to follow to make their experience more social. Instead of artificial predictions, the ideahere was to use actual activities, such as "this user liked the same products as you" to displaysimilarities and encourage the forming of links within the network.Among the more elaborate ideas extracted from these considerations were:

    • Show "latest" and "friends" looks, to have a better chance of having your looks show upfor other people. The existing "popular" feed is commonly dominated by users with biggerreach.

    • More relevant notifications/alerts (who else liked the same things, summaries of how manylikes your content received this week, etc.)

    7http://contentcaptain.de/

    12

    http://contentcaptain.de/

  • • Allow users to specify who gets to see their content based on groups or lists.

    • Allow comments on products/looks (A version of this had existed before but didn’t fit intothe product at the time)

    3.1.4 Other Improvements

    On a more general level many subtle tweaks to design and experience could be made. It is acommon issue that you grow biased towards quirks of a system you helped build yourself. If ithas been that way for a while, you stop questioning it. This effect may have also brushed offon the author, so that further considerations were postponed until more external feedback couldbe gathered. However, a few inconsistencies were identified and noted that had not previouslybeen noticed by other members of Apprl. Some of those probably arose out of the simple factthat different parts of the application were added at different times and replicated similar featuresin slightly different ways instead of properly reusing existing code. Others were simply designdecisions that seemed questionable and could not be traced back to any apparent reasoning.

    Thinking beyond the current reach of Apprl, integrating Apprl with offline media such asmagazines, flyers or billboards was another idea. Existing short-links for products or looks couldbe used for this purpose, but further solutions such as offering QR-codes directly on the site couldbe valuable additions.

    For those using the platform mainly to drive sales through external channels, adding productsto Apprl’s catalog themselves or making changes to wrongly parsed products seemed like usefulfeatures. Those particular ideas were conflicting with Apprl’s views but were still taken intoconsiderations throughout the analysis and decision making phases.

    3.2 Feedback from UsersApprl is still a small start-up and does not have its own PR/marketing department. They some-times get feedback from their bigger publishers and stores in their network but those groups canbe considered business partners rather than regular users. Measuring success this far was mostlybased on actual earnings being made. Feedback from regular users had not been gathered for along time prior to this thesis project. While a lot of areas worth improving had been identified asdescribed in the previous chapter, they were not the only topics of interest for gathering user feed-back. As mentioned before, a certain bias towards your own product might make you insusceptibletowards certain issues. Hence, all parts of the existing service were subject of investigation.

    Interviews with existing or prospective users were conducted to gain these insights. It wasimportant to learn how they react to and how they feel about Apprl’s services. The input fromthese interviews was of high importance for the decisions made later on and helped to clear outbiased ideas that had been enthusiastically proposed by either the author or employees of Apprland turned out to be of little or no meaning for the actual users.

    3.2.1 Form of Feedback

    For the specific goal identified prior to the interviews it was clear that the user experience, whatmotivates users in the short and long term had to be investigated first and foremost. Based onthe fact that Apprl had already been operating for a considerable amount of time and had activeusers at the time of writing, the existing service could be used for the gathering of feedback. Theexisting interface was thus demonstrated as part of interactive interview sessions that led to moremeaningful results than abstract questionnaires or other indirect forms of investigation. Duringthe interview sessions, the peers were given a set of tasks to perform on the current platform.

    13

  • One part even involved one of the most prominent partner sites that was already using Apprl’stool extensively: the fashion portal at www.metromode.se. They were asked to explain to theinterviewer what they were doing and what they expected to be the outcome of certain actionsthey performed.

    Interviewees were asked whether it would be okay to record audio of the interviews and theyall agreed to that. Each interview’s audio recording was revisited at least once, but transcriptionswere omitted due to time constraints. The interviews were not fully scripted but only preparedbased on a general outline in order to make the peers feel more comfortable which was supportedby the chosen environment: cafés in the center of Stockholm, selected by the interviewee. A fullversion of the outline for the interviews can be found in Appendix A. Watching the interactionclosely, the interviewer was able to understand more subtle things such as the time it took a userto find the correct place to click and where they started looking first on the page. These insightscoupled with the actively uttered feedback from users gave a much clearer picture of how the usersfelt rather than just measuring the time taken to perform a task under pressure. In fact, dependingon the user’s speed, tasks were added, refined or removed to stay within the promised 30 minutetime frame. When given the option to continue after 30 minutes however, all except one participantwere willing to finish some more tasks and talk about other issues and the possible future of Apprlor social commerce in general.

    The second, voluntary, part of the interviews was more theoretical and had been preparedunder the assumption that the peers already had some time to understand Apprl’ service and thatthey were actively willing to contribute more. Feedback given bluntly without pressure from theinterviewer is known to be less biased and thus more valuable. Predicting this behavior turned outto be correct. Once having completed the tasks most of the interviewed peers had some questionsabout certain functionalities or Apprl/social commerce in general which occasionally led to livelydiscussions about current and future developments. Thereafter, unless previously touched upon, aset of very direct questions were woven into the emerging discussions and lastly a few possible newfeatures were presented to the interviewees, including some visual prototypes for the more complexfeature ideas (see Appendix A and A.4). Those complex features were mainly geared to increaseuser’s motivation through usefulness or playfulness as outlined in 3.1.2 and were unlikely to bechosen for further pursuit due to the required time for developing them and the general preferencefor minimalism on Apprl’s side.

    3.2.2 Selecting Peers

    Over the course of 2.5 weeks, 10 peers were interviewed. Two of them were publishers on Apprl,the others either existing users or young consumers with different levels of interest in fashion.

    Peers were chosen based on the target audiences currently or prospectively aimed at by Apprl.In its existing form the service had been mostly used by fashion bloggers, or publishers, to createand reuse content. Thus, efforts in maintaining and improving the service had been mainly targetedtowards that particular group of users. Other people with an extraordinary interest in fashion hadbeen attracted by the service, too, but remained a minority. Expanding the user base wouldmean reaching out to new audiences, meaning people with less interest in fashion. As describedin 2.1.2 the majority of people rely heavily on personal recommendations and since we also knowthat fashion matters to nearly everyone to a certain extent (2.1.1), bringing those who have lessexpertise in the field together with the influencers of the fashion scene is likely to yield desirableresults for all parties involved. It was thus crucial to find peers with different attitudes towardsfashion. A certain predisposition towards online services and social media was desirable but couldgenerally be considered a given based on the fact that 72% of the Swedish society are using socialnetworks and 85% are engaging in online shopping [28].

    14

    www.metromode.se

  • Peers were invited through Facebook (prospective users) or personal eMails (existing users).Facebook groups around student life and other activities were chosen to post the short invitations.A friendly portrait of the author was used to give it a personal touch. Existing users were contactedbased on their activity on the platform. Some of the selected ones used to be active but seemed tohave stopped using the platform, their answers were expected to help understand what had beenlacking in terms of long-term motivation even though they seemed to have enjoyed the service atsome point.

    As mentioned before, the focus of Apprl had been to improve their tools for publishers and fewof them had been giving feedback actively unless something was not working at all. Furthermore,publishers are also making use of all the features for regular users; they simply have access to someadditional options and statistics to monetize their content. In order to let publishers be a part ofthe study, some publishers were contacted based on business relations and the likeliness of thembeing available for an interview. A pledge from metromode.se (mentioned earlier) to have a fewbloggers provide feedback for was never followed up on unfortunately.

    3.3 Analysis of ResultsBefore moving on to the conceptualization of the features which were implemented during thecourse of this thesis, the results of the prior background research combined with findings of thesubsequent interviews is discussed. This section mentions which opportunities turned out to bepromising and also talks about the ones that could not be considered for further pursuit due tothe companies’ plans or the feasibility of their implementation based on time and infrastructureconstraints.

    3.3.1 Results from Interviews

    One thing was certain a few minutes into each of the interviews: using Apprl.com was "fun",as most users put it. Apprl’s service definitely had a certain appeal and peers said they werelikely to use it again. That was an important milestone, but looking at the numbers of previoususers, the initial excitement seems to have faded after a few weeks or months. The biggest pointsthat would give the users a greater incentive to keep using the service for a prolonged time couldbe identified based on their comments during the interviews and especially during the followingdiscussion. Not all statements and personal remarks could be considered individually. Therefore,recurring patterns and similar remarks by the interviewed peers were analyzed and captured ina spreadsheet trying to quantify the feedback as ground for discussion between the author andApprl. Appendix A.5 shows the result of this process.These are some exemplary quotes (paraphrased for better readability):

    • "That content looks really polished. It’s great when I am looking at something nice andautomatically get the option to buy it!"

    • "I want to find everything I am looking for on Apprl. I am especially missing [brandname]"

    • "Where do I find my stuff?"

    • "I would really like to see [person of interest] on here."

    • "I would want to work on a look together with friends, for example for coordinating our stylefor a big event"

    • "Where does this come from? Why is it showing me dresses? I’m a guy!" [looking at thestart page]

    15

  • Based on initial statements by the interviewed peers they could be considered quite diverse whenit came to interest towards "fashion". The assumption for the project was that to a certainextent personal recommendations about fashion could be interesting to everyone. And indeed, allinterviewees including those who had initially considered themselves "not interested in fashion"helped underpin this assumption. During the course of the interviews each subject agreed, oftento their own surprise, to sometimes base decisions about clothing purchases on input from peoplethat are important to them: friends were named by nearly everyone and other influential figures intheir lives, such as athletes, actors or even successful entrepreneurs played a role for more than halfof those interviewed. This was valuable feedback and supported the general concept behind Apprlas well as the plan to become more social. Who is on the platform is more of a business goal andan important clue for the social media and marketing agenda but cannot be directly approachedfrom a technological stand-point.

    Some of the ideas that had been previously made up turned out to be rather unimportant forthe users when confronted with them. A brief overview of those is given at the end of this section.Other findings led to the selection of ideas to pursue further, which is described in the followingsection.

    3.3.2 Selecting Ideas

    Many of the identified areas for improvement clearly hit a nerve. Some new ideas only came upbased on what was said during the interviews or were directly suggested by peers. The morepromising topics are mentioned here and covered in more detail throughout the following chapterthat works out concepts for actual implementation. To facilitate the process of choosing the mostrelevant ideas, the set of ideas that had been found to be worthy of further consideration after theinterviews was mapped out on a table using small paper cards. A matrix layout was chosen tovisualize the parts of Apprl’s service and the business interests covered by the various ideas. Anannotated version of this step that clarifies the underlying structure is depicted in figure 2.

    Figure 2: The gathered ideas structured by key interest areas for Apprl and part of the service.

    16

  • Visualizing the ideas in this way helped immensely in finding overlaps in the focus areas andpartially confirmed some assumptions about what is important to users while other ideas were noteven used any more at this point because they had proven to be of little interest. The selection ofideas to be implemented as part of the thesis was made based on three criteria:

    1. Importance based on user feedback

    2. Technical similarity to maximize efficiency during development

    3. Value for the company

    Without writing about the ideas of each individual card the general direction of results and thedecisions towards which they led are explained here.

    Social Interaction is Key Several of the interviewed peers expected that their interaction withthe service would be more "fun" once they had some of their personal friends on the networkand could interact with them. This finding correlates with an important insight from previousresearch: social interaction is very important for users; feeling as if their actions are perceived byothers they care about is one of the key factors to keep them engaged in whatever communitythey have chosen to be a part of. The existing creation tools have proven to be fun for users butsocial features offer room for improvements. Comments were made that hinted to the importanceof friends seeing your creations. From the set of initial ideas the idea of adding two new tabs tothe list of existing looks that show "latest" and "friend’s" looks was promising to improve just thatcompared to the existing service. Another idea that appeared based on the interest of seeing morecontent related to people’s friends is to not only show looks they created but also ones the likedon their profile pages. There was a general tendency towards more fine-grained visibility scopesthat can be understood to be about ensuring that friends see your content and you see theirs.Some peers were interested in prohibiting access to certain content to the public altogether. Thiswill surely be an issue for social commerce and seems an important concern to the author. Thecompany’s stance on the topic, however, is to make everything public in general.

    Furthermore, the re-introduction of a comment function has been identified as an importantitem to be put back onto Apprl’s roadmap and was scheduled to be carried out with more planningat a later point this year. An interesting question that could not be answered adequately withinterviews was whether users would stay excited about their interaction with the content in thelong term. However, it seemed that the interacting with relevant people is a key ingredient forlong-term engagement with the service.

    Notifications While the visual appeal of notifications was not tested during the interviews (adecision to redesign them had been made already), their importance for the users could be veri-fied. Some users, however, did not want to be "annoyed" by notifications. They only wanted toget an eMail about "important" things. What is important to an individual could not be unifiedinto a single solution. Nevertheless, the desire for more relevant notifications, but possibly fewernotifications for those who preferred it, could clearly be seen. This led to further concept develop-ment regarding new types of eMail notifications and/or finer grained settings, such as "only sendnotification after at least 5 likes" as was suggested by one interviewed peer.

    Additionally, bringing notifications onto the website itself proved to be a worthy addition forall users. For those users who simply did not want more eMails in their in-box this would stillkeep them updated about what is happening in Apprl, while others simply get an easier overviewin addition to the eMail notifications.In summary, three improvements in this area were selected for improvement during the next phase:

    17

  • • Visual templates for more appealing look of eMails

    • New kinds of notification eMails such as activity summaries and/or finer grained settings.

    • Including notifications on apprl.com for all users, whether they are receiving eMails or not.

    Being Relevant Previously only considered within the context of being more socially relevant,some of the interviewed peers saw a certain value in finding things more quickly. The "similarproducts" panel on product pages was heavily used and found to be useful during the interviews. Itsrecommendations were based on a rather simple algorithm, and yet the need for a recommendationsystem running in the background did not seem warranted. One interviewee pointed out that shecould just click a few more times when it is "just there anyways", so that the quality of therecommendations was not a limiting factor considering the convenience the feature added.

    One identified issue, however, was the content users saw on their start page. Once they un-derstood where it originated from (the users and brands you follow) they were less concerned butstill felt like it could be more personalized, e.g. through also giving them a glimpse at what theyliked recently. An actual problem was that the implementation had a bug that caused the gendersetting on the start page to always default to "all". Coupled with the vast majority of female usersand products this made the content seem much less relevant for male users. Accordingly, the newbehavior was chosen to be using the user’s gender as default which could be considered a "bug fix"rather than a new feature.

    3.3.3 Dropped ideas

    Of course not every idea could be followed up on within the given time. Other features simplydidn’t comply with Apprl’s strategy, had been on the roadmap for later implementation already orsimply were beyond the scope of what Apprl’s platform is supposed to offer. This section focuseson ideas that were dropped for those reasons.

    The following chapter 4 also mentions concepts for some of the features that were identified aspromising but put on hold for the time being.

    Beyond Scope Unfortunately, some of the more complex features had never been expected tobecome reality during the thesis. While the feedback for them was positive in general, theirimplementation would have consumed more time than was available for development. In additionto that, Apprl’s philosophy has always been to strive for simplicity and less useful features havebeen removed before to streamline the experience. There exists, however, a plan to improve theAPIs for third party developers. This might be accompanied by prototypical examples of whatcould be done based on the existing catalog of products, vendors and tools that Apprl alreadyoffers or eventually result in other developers realizing similar ideas. The "Look Slot Machine"and "Notepad" concepts are promising candidates for external services built on top of those APIs.

    One feature that was used by two of the peers only but turned out to be of limited usefulnesswas the search feature. Depending on the search term the results ranged from very close matchesto completely unrelated content. It was unclear to the author what the exact algorithm did butit seemed to mostly rely on description text and product names. There was no direct mapping ofkeywords to categories so that for example using terms including colors might also yield resultsthat went along with or had details in that color. A short discussion with (only the business partof) Apprl revealed that the area had been identified as low-priority before and was expected torequire a bigger development effort to be re-built. A technical analysis was thus not carried out.

    An interesting point that had been noticed before and was confirmed by one interviewed pub-lisher was that fashion bloggers tend to have very high standards for their content. That particular

    18

  • publisher was not satisfied with the image quality of the collages created on Apprl.com since over-lapping images were not cut out correctly. She also disliked the intrusive optical appeal of the"bubbles" placed on looks using an uploaded photo. While the first concern was simply deemeda special case, considering the alternative tool named was Adobe Photoshop, the second aspectmight be taken into consideration later. Since no such feedback had arisen prior to the interviews,Apprl neglected the idea of adding an on/off toggle for the buttons so viewers would get the chanceto see the original picture or the one with overlaid links to the products for the time being.

    User Interface When it came to certain slight design changes, the mentioned bias towards theirown product drove Apprl to reject changes to things that were "not broken". They might entrusta professional user interface (UI) designer with that topic and would like to deploy A/B (or split)testing for future design changes. A/B testing was briefly considered to be included for the thesisbut a lack of resources to actually integrate the needed infrastructure for split testing resulted inpostponing most design changes altogether. There were several slightly confusing elements of thesite that needed some explanation. For example nearly all interviewees had trouble retrieving a lookthey had just created because neither the placement of link to one’s own profile was obvious nor didthey understand that an unpublished look did not show up in the counter on the corresponding tabon their profiles. One conclusion that was drawn from this and was taken more seriously, however,was that new users should get a better "onboarding" in the form of an interactive tour introducingthe main features of the service. A very general introduction had been in place for a while, but alot of users never even read it. The new implementation was supposed to be harder to skip/missand be interactive rather than a purely static description of available features.

    A barrier that pops up even before a user takes her first steps with the service is the landingpage you see on Apprl.com when you first visit the page. None of the interviewed peers who hadn’tused the service before understood what they could do on Apprl. As a result they all stated thatif it weren’t for the subsequent interview and the interviewer giving them tasks to perform theywould have most likely left the page at this point. Again, the company was hesitant to makechanges to this page without insights from an expert.

    One issue that was noted several times during the interviews was the setting for location. Mostinterview partners were not Swedish and preferred an English interface. Finding that setting wasnot very easy since it was placed at the bottom of the page. Some of the peers did, however, thinkthat prices in SEK were easier to grasp for them than the USD connected with the internationalsetting. This was not possible. A decision to convert the entire service to English-only had beenmade prior to the writing of this thesis. Due to more pressing topics for the responsible developersit had to be postponed to an undisclosed time. At least the location settings default selection hasbeen improved by the use of a new engine for determining the user’s location. Once the languageis always set to English this issue will thus automatically be solved; only the currency and listedstores will then be adapted to a user’s location automatically.

    Product Information Both publishers and regular users had trouble with wrongly categorizedproducts; the interviewed publishers were interested in having a chance to both add productsthat they could not find on Apprl.com and suggest or make corrections to wrong information.Technical considerations as well as the management and maintenance of granting those editingrights were deemed too costly for that feature to be implemented. Manually categorizing theamount of products coming in from different vendor’s feeds would require an immense amount ofwork and thus the focus in this area lies on finding a good balance between little overhead andminimum amount of errors in the information. The information scraping for products has beenimproved shortly after the interviews and should now yield slightly better results. Making theautomated algorithm smarter was considered a more promising way of minimizing the problem.

    19

  • Adding external products with own photos was added to the development roadmap and will bedeveloped outside the thesis.

    Due to the lack of consistent information from all vendors, the concept about adding size filterson Apprl’s shops had to be omitted as well. Several peers actually mentioned the demand beforethe feature had actively been suggested. Apprl acknowledged the usefulness of this feature andwould like to make it possible in the future.

    Another vital source for information are product photos. Nearly all interviewees wanted tosee shots from more than one angle and when it came to creating looks most of them would havepreferred to use pure product photos instead of pictures containing a model. An inquiry with Apprlrevealed that this feature was indeed very interesting for the service but could not be realized dueto the limited amount of information provided by different vendors. However, the aforementionedoption to upload photos to add extra products was deemed to remedy the problem to a certainextent. Additionally, during the interviews, it turned out to be a minor problem for users to simplygo to a store’s product page to look at more photos there instead of seeing them directly on Apprl’swebsite.

    Long Term Engagement through Gamification The Topic of long-term engagement can bethought of as a very broad topic and is closely related to other topics identified earlier and se-lected for implementation. The ideas of goals and achievements, however, did not make it intothe implementation even though they earned very positive responses during the interviews. Thisdecision was based on the fact that the outcome of further investigation into which particularminimum set of achievements would help to improve the experience was uncertain and there wasno prior experience to rely on for making design decisions. A lot of thought has gone into this andan internal gathering of ideas surrounding the topic was started and contributed too without anydefinite results at the time of writing. Section 4.7.1 in the next chapter gives some insight into thelatest state of said discussion.

    3.4 Planning next StepsDrawing conclusions from section 3.3.2, there is one obvious ingredient to long-term engagement ofa user that was chosen as the main area for improvements: interaction with people who matter toher. This means friends for some, role models and other influencers for others or maybe both forsome people. What we could not do was forcing those people to also join the network. What wecould do on the other hand was improving the way in which people can interact with one anotherand keeping them updated about what is happening in their personal network. More detailedconcepts derived from the findings described in this chapter is presented in chapter 4.

    20

  • 4 Identified OpportunitiesIn the previous chapter the process that led to the identification of key areas of interest wasdescribed in detail. Descriptions of the ideas created and rejected at different points in time,however, were kept to a minimum. This chapter re-examines the reasoning behind the selection offeature ideas and provides a more detailed concept of what was supposed to be achieved duringthe implementation phase that followed.

    4.1 Making content exploration more socialBackground Why a more social form of interaction is important for the success of a platform inthe age of social media as the primary medium for a majority of people has been analyzed andexplained extensively in previous chapters. After some initial ideas for a more social experiencehad been developed, more variations came up based on feedback given by the interviewed peers.The general motivation was to give users more ways to find content that was created by people theyare interested in or what those users liked. This means that the content that is most relevant toyou personally will not be drowned in the content that is popular among the rest of the communitybut rather be seen by those who are more important to you. The approach is closely related tothe topic of one of the new eMails presented in 4.3.

    Concept/ Design In order to maintain a consistent look across the website, similar areas of thesite were chosen as examples for the look and feel and according front-end code could be re-used.Bringing more looks to the users’ attention is achieved by adding two additional tabs to the listof looks. The list has always been completely public and showed looks based on a pre-computedpopularity ranking. The new design includes one "latest" tab that only sorts based on creationdate as well as a "friends" tab on which only looks from people you are following are displayed.Figures 3 and 4 depict the new designs for said pages. Prior versions were omitted because theybasically just showed empty white space where the new tabs were placed.

    Figure 3: The "latest" tab on the redesigned lookspage.

    Figure 4: "Liked" looks on a users profile.Counter on first line shows an aggregate.

    21

  • 4.2 Visual Overhaul of eMailsBackground Apprl had been sending out eMail notifications for certain events for a long time.The content of those messages was limited to simple one-liners without any layout or graphicalelements. eMail marketing has come a long way since its beginnings and in order to get a user’sattention it is important not only to have information that matters to her but also be visuallyappealing and easy to understand.To date, the following types of notifications had been used:

    • Someone starts following you

    • Someone liked a look you created

    • A product you like is on sale (different text for further reduction)

    • One of your Facebook friends joined Apprl

    Concept/ Design What makes any kind of information easier to understand is the use of graphicelements. Bringing your point across quickly is what will drive conversion rates in the contextof eMail. Hence, the new templates for eMails put an emphasis on images while using very littletext. One subtle trick used here is borrowed from the field of psychology and was also describedby Anderson [25]. Psychology tells us that the simple act of using the word "you" makes a pieceof text a lot more likely to grab the reader’s attention.

    Two of the eMail notifications are shown in figure 5 visualizing the immense difference betweenthe previously used and newly designed variants. The remaining notification types were "an itemis on sale" and "a Facebook friend joined Apprl", an example of the first is shown in AppendixB.2. Unfortunately, neither the