Self-served campaign management tool for contextual advertising

1.1 Adlede Acknowledgements 1

Self-served campaign management tool for contextual advertising

Magnus Emanuelsson

Umeå University Spring 2020

Master Thesis, 30 hp

Supervisor at UmU: Leonid Freidovich External Supervisors: Mona Forsman Examiner: Thomas Mejtoft

M.Sc. Interaction Technology and Design, 300 hp

1.1 Adlede Acknowledgements 2

A

To complete this thesis, I got helped from many people and I appreciate their time.

Codemille and Adlede: For letting me do my master thesis at their company, a lot of great people there and all support given from the staff.

Mona Forsman: My supervisor at Adlede, have help me a lot and are really engage in what I have been doing. Thank you, Mona, for all help.

Leonid Freidovich: Supervisor at Umeå University, really good at feedback and have given valuable support.

Annika Bindler: Working with academic writing in English at Umeå University. Thanks for all the time you have taken to help me with my writing.

All participants of tests and form answer: Thanks for the time you took for helping in this thesis.

1.1 Adlede Abstract 3

A

Advertising is an essential way of communicating value to the audience. Adlede does this with context in focus when finding articles with right content to the right campaigns. To know if matching articles are good, advertisers need to get statistics from the campaign. Today a tool for this does not exist. This master thesis analyses, how to best develop a self-serve, user- friendly campaign management tool for publishing a campaign and shows results from the campaign.

This thesis design and create a concept of a management tool with help of frequently used UX design method. To get a solution for this concept, interviews, designs, prototypes and user testing have taken place. A final design suggestion has been determined.

1.1 Adlede Abstract 4

Sammanfattning

Annonsering är ett viktigt sätt att kommunicera värde till en specifik målgrupp. Adlede gör detta med kontext i fokus när man hittar artiklar med rätt innehåll till rätt kampanj. För att veta om matchande artiklar är bra måste annonsörer få statistik från kampanjen som körts.

Idag finns det inget verktyg för detta hos Adlede. Detta examensarbete analyserar hur man bäst kan utveckla ett självbetjänat användarvänligt kampanjhanteringsverktyg för att publicera en kampanj och visa resultat från kampanjen.

Denna avhandling utformar och skapar ett koncept för ett annonshanterings verktyg med hjälp av kända UX-designmetoder. För att komma fram till en lösning för detta koncept har intervjuer, design, prototyper och användartestning ägt rum. Ett slutligt designförslag har fastställts.

1.1 Adlede Abbreviations 5

A

UX DSP SSP Ad

User Experience Demand Side Platform Supply Side Platform Advertise

1.1 Adlede Abbreviations 6

T

C

ACKNOWLEDGEMENTS ... 2

ABSTRACT ... 3

ABBREVIATIONS ... 5

LIST OF TABLES ... 8

LIST OF FIGURES ... 9

1 INTRODUCTION ... 1

1.1 ADLEDE ... 2

1.2 OBJECTIVE ... 2

2 THEORETICAL FRAMEWORK ... 3

2.1 ADS MARKETING ON INTERNET ... 3

2.2 CONTEXTUAL ADVERTISING ... 3

2.2.1 Keyword matching ... 5

2.2.2 Natural Language Processing ... 5

2.2.3 Mood Targeting ... 5

2.2.4 Brand Safety ... 5

2.3 UX(USER EXPERIENCE) ... 5

2.4 CATEGORIZING DATA ... 7

2.4.1 Card sorting technique ... 8

2.5 ADVERTISING DISPLAY FORMATS ... 8

2.5.1 Banners ... 9

2.5.2 Pop-ups [24] ... 10

2.5.3 other formats [24] ... 10

2.6 DIGITAL ADVERTISING PROCESS ... 11

2.7 USER TEST ... 12

2.7.1 Case study ... 12

2.8 USER STORIES ... 12

2.9 GOOGLE ADS ... 13

2.10 ANALYSIS DATA [32] ... 14

2.11 GOOGLE ANALYTIC ... 14

3 METHOD ... 16

3.1 RESEARCH ... 17

3.1.1 Interview with UX designer ... 17

3.1.2 Card sorting ... 17

3.1.3 Interviews with advertisers ... 17

3.1.4 Layout ... 18

3.2 ANALYSIS ... 18

3.3 FIRST ITERATION ... 19

3.3.1 Design ... 19

3.3.2 First prototype ... 19

3.3.3 Evaluate ... 19

3.4 SECOND ITERATION ... 20

3.4.1 Design ... 20

3.4.2 Second prototype ... 20

3.4.3 Evaluation ... 20

3.5 FINAL DESIGN SUGGESTION ... 20

3.6 CORONA VIRUS ... 20

4 RESULTS ... 22

4.1 RESEARCH ... 22

1.1 Adlede Abbreviations 7

4.1.1 Meeting with UX designer ... 22

4.1.2 Card sorting ... 23

4.1.3 Form answer from advertiser ... 25

4.1.4 Layout ... 25

4.2 FIRST ITERATION ... 30

4.2.1 Analysis ... 30

4.2.2 Design ... 32

4.2.3 Hi-Fi prototype ... 35

4.2.4 Evaluation ... 37

4.3 SECOND ITERATION ... 38

4.3.1 Design ... 38

4.3.2 Protoype ... 38

4.3.3 Evaluation ... 38

4.4 FINAL DESIGN DECISION ... 38

5 DISCUSSION ... 39

6 CONCLUSIONS ... 41

7 FUTURE WORK ... 42

8 REFERENCES ... 43

APPENDIX A ... 46

APPENDIX B ... 48

1.1 Adlede List of Tables 8

L

T

TABLE 1.USABILITY FOR USER EXPERIENCE [19] ... 6

TABLE 2.DESCRIBES THE HONEYCOMB [21]. ... 7

TABLE 3.RESULT FROM CARD SORTING FOR STATISTIC PAGE. ... 23

TABLE 4.RESULT FOR ADDING A CAMPAIGN TOOL. ... 24

TABLE 5.THE NUMBER REPRESENT MISS CLICK OR CONFUSION ... 37

1.1 Adlede List of Figures 9

L

F

FIGURE 1.CONTEXTUAL ADVERTISING PLATFORM (BASED ON GUANGDONG 2013) ... 4

FIGURE 2.HONEYCOMB DESIGNED BY PETER MOREVILLE ... 7

FIGURE 3.DIGITAL ADVERTISING PROCESS. ... 12

FIGURE 4.AN ADAPTED VERSION OF “THE WHEEL”. (R.HARTSON 2012) ... 16

FIGURE 5.AFTER THE CARD SORTING TECHNIQUE FOR GROUPING STATISTICS. ... 24

FIGURE 6.AFTER CARD SORTING TECHNIQUE FOR ADDING CAMPAIGN. ... 25

FIGURE 7.HIERARCHY OF LAYOUT. ... 26

FIGURE 8.START PAGE OF LAYOUT DESIGN. ... 26

FIGURE 9.ADDING A CAMPAIGN. ... 27

FIGURE 10.COMPARE TWO CAMPAIGNS. ... 28

FIGURE 11.STATISTICS FROM A CHOSEN CAMPAIGN. ... 29

FIGURE 12.EDIT PAGE ... 29

FIGURE 13.MOCK UP FOR SIGN IN PAGE TO ADLDEDE DASHBOARD. ... 30

FIGURE 14.MOCK UP FOR ADDING A CAMPAIGN TO ADLEDE SYSTEM ... 31

FIGURE 15.MOCK UP FOR PREVIOUS ADDED CAMPAIGNS ... 31

FIGURE 16.DESIGN FOR SIGN IN TO ADLEDE DASHBOARD ... 32

FIGURE 17.DESIGN FOR ADDING A CAMPAIGN TO ADLEDE SYSTEM ... 33

FIGURE 18.DESIGN FOR PREVIOUS ADDED CAMPAIGNS ... 34

FIGURE 19.DESIGN DASHBOARD FOR CAMPAIGNS ... 34

FIGURE 20.DESIGN STATISTIC PAGE FOR ONE CAMPAIGN ... 35

FIGURE 21.PROTOTYPE ADD CAMPAIGN. ... 36

FIGURE 22.PROTOTYPE STATISTIC PAGE ... 37

1.1 Adlede Introduction 1

1 I

Advertising is an essential way of communicating value to the audience. The value is to mediate information or convince the audience to make a purchase decision [1]. Advertising is an essential aspect of the economic growth for marketers and business [2].

The earliest form of advertising came from antiquity and campaigns were shown in the form of posters and store signs. At the end of the 1800s, advertisement appeared in papers and ads Journal [3]. New ways of advertising have begun to be used, under 1900s there was a new era of advertising. In 1900s Tv, radio, and movies make their way to become a new part of the society and opened doors for new ways of advertising [4].

The first internet advertising was made in 1994 [5]. In 2019, digital ads took over ad market and had 50.1 % of all global advertising. The total spending in 2019 was 333.25 billion USD (US dollars) on digital ads. The biggest actor in digital marketing is Google. Google had revenue at 103.73 billion USD in 2019. Today the internet is the fastest growing form of advertising [6].

In 2017 the total number of websites was over 1.75 billion [7]. One major issue of placing ads on the internet is to find a convenient advertising spot. However, a solution to target suitable ad spots is contextual advertising. Contextual advertising is about matching ad with the right content, by right keyword, to the right audience [8]. An advertising that suits well with context of page is more likely to assemble more clicks. To find right content, there are several techniques that can be applied, such as Keyword extraction, Natural language processing, Mood targeting and Brand safety [8]. Contextual advertising does not need to collect personal data and avoids the use of cookies.

A company working towards making contextual advertising better and with higher precision is Adlede. Adlede has taken contextual advertising to a step beyond keywords. In an AI-based software can Adlede find the right content, not just in text but also in video and images. The right content found via keyword extraction, natural language processing, brand safety and mood targeting [8].

A benefit of advertising on the internet is the possibility to provide feedback to company instantly. A company can see how many clicks and how many interactions they collect in real- time. Right now, there is an issue with Adlede, and it isn't easy to receive good user-friendly statistics under and after a campaign. Adlede has no tool for an advertiser to add a new campaign to the system. The purpose of this paper is to study with the help of research, interviews, inspiration from similar tools and user tests on how to make a self-served management tool for Adlede. This tool will be designed, tested and implemented.

1.1 Introduction 2

1.1 ADLEDE

This thesis is written and carried out with Adlede. Adlede is an affiliated company to Codemill.

Codemill was created in 2007 and has its main focus on technical video solutions. International clients, such as, The Guardian, ITV and ProSiebenSat 1 make Codemill one of the big actors in the video solution market [9]. Codemill has three affiliated companies. One is called Accurate Player, which is a media player for broadcast, post-production and media professionals [10].

Another one is called Accurate Video, which is a video software company [11]. And the third one is Adlede, where the main focus is contextual advertising, which includes keyword extractions, natural language processing, brand safety, and mood targeting. Adlede running EU’s biggest project in AI-driven contextual understanding, this makes Adlede too one of the major actors on AI driven contextual advertisement [8].

1.2 OBJECTIVE

This master thesis analyses, how to best develop a self-serve, user-friendly campaign management tool for publishing a campaign and shows results from the campaign. The analysis serves as the foundation for making a design and later implementing the tool. The focus lays on advertisers’ opinions, on which statistic to show from a campaign, and how they decide to add a campaign.

Research Questions

This paper’s goal is to determine how a self-served management tool can be designed and should be designed to fulfil advertisers’ expectations. To determine this design, this thesis aims to gain a more comprehensive understanding of how advertising works, how a tool can be designed to make a satisfactory UX and investigate which results the advertiser would like to have from a campaign.

The research questions for this thesis are:

I. What statistics do advertisers require from a campaign?

II. What function is requested when adding a campaign?

III. How should statistics be categorized?

2.1 Contextual advertising Theoretical Framework 3

2 T

F

This thesis will be focused on how online advertising works today and also on UX design.

Therefore, the theoretical framework is based on essential facts about different types of advertising. The focus will lay on contextual advertising and techniques to sort and filtering in a big amount of data. The theoretical framework explains some concept of UX design. The concepts of UX are needed for design the advertising/reporting tool.

2.1 ADS MARKETING ON INTERNET

1990 www (world wide web) emerged as the new way of communications for consumers. In 1994 the first ad was created for websites. In the beginning, advertising was displeasing for users. After a while advertising were accepted from users instead of letting them pay for programs/websites [5].

Internet has made it easier for companies to globally sell their products . When there was no internet, most of the companies made local advertising such as posters or ads in journal. Now a day’s companies have a global display window for their products via the www. Global display window makes it easy for companies to reach clients through the whole world. Advertising to consumers was prior to introducing products and producers, today most of the company’s advertising to satisfying consumers’ needs and demands [12].

For now, the fastest growing marketing component is digital forum [13]. Digital forum is where a user can post or read digital posts, digital forum is also the marketing place where companies put the most money on. The major market strategy via digital forum is direct marketing. Direct marketing is when an ad is aiming directly to a selected customer. If the selected customer is not interested in the advertisement, it could disturb and make a negative effect on consumer [13].

2.2 CONTEXTUAL ADVERTISING

To find right content for an advertisement and to reach out to the right audience there is an advertising method named contextual advertising. Contextual advertising aims at advertise spots relative to the context. However, it will increase the chance to more ad-clicks from users if the ad corresponding to the content. More ad-clicks will increase selling for advertisers and also higher revenue for the publisher. From advertiser to user four segments are included in a contextual advertising platform [14] as shown in Figure 1. The following concepts are important.

2.2 Contextual advertising Theoretical Framework 4

• Advertiser have a campaign or ad they aim to reach out with. Advertiser pays for the ads.

• Publisher holds webpage with space for a campaign/ad. Publisher provide a good user-friendly spot to induce many clicks. Publisher earns money from number of clicks, more traffic more money.

• Ad platform is where Adlede software works. Adledes software system is finding matching pages or matching articles. The articles and the ads match with the highest possible similarity to maximize publisher’s and advertiser’s revenue.

• Users are the potential buyers of the ads [14].

FIGURE 1.CONTEXTUAL ADVERTISING PLATFORM (BASED ON GUANGDONG 2013)

In 2003 Google provided the first contextual advertising platform. Since then almost every big search engine has a similar platform.

There are several techniques to finding articles/ web places with a content-based approach.

They can be combined to make a better ad match and are described next.

2.2.1 Keyword matching

2.2.2 Natural Language Processing 2.2.3 Mood Targeting

2.2.4 Brand Safety

2.3 UX (User Experience) Theoretical Framework 5 2.2.1 KEYWORD MATCHING

One technique is to have a keyword matching algorithm. Let us say a user reads an article about scuba diving in Australia and the article has an ad about a flight trip to Australia. Then, the match is well fitted. Several problems can appear when making keyword matching [15].

• The words can have a substantial meaning, like Jaguar. There are a car named jaguar and also an animal. Then, the article can be about a fast animal, which is a jaguar and an advertisement about the car jaguar could appear.

• If the word is an equivalent like New York and the big apple. The big apple is a common way to express New York but it can be hard for a keyword matcher to understand it.

• Mismatch in the semantics. An article about scuba diving drowning accident and the keyword matching with an ad about scuba diving trip. The ad miss fit with the article.

2.2.2 NATURAL LANGUAGE PROCESSING

Using a keyword is just looking at the specific word, while using natural language processing is looking up language around that word. Natural language processing gives an understanding if the content is about an animal (jaguar) or a car (jaguar). This increases the precision of the ad.

2.2.3 MOOD TARGETING

A content about a funeral has most likely a sad context (mood). The content about a national team winning the world cup, is most likely a happy or exiting context. To match the right mood can be helpful when designing and placing an ad [8].

2.2.4 BRAND SAFETY

When publishing online and not knowing the content there could be a brand placed next to undesirable content. To avoid these kinds of inconvenient placement brand safety will help.

The goal for brand safety is to protect a brand's reputation and to keep them out of content that can be negative and damaging their influence [16].

2.3 UX(USER EXPERIENCE)

In 1993 Don Norman coined the term UX (User Experience) and he has written the famous book about UX “The design of Everyday Things” [17].

2.3 UX (User Experience) Theoretical Framework 6 As of today, UX is a term that has several different definitions, In the last years there have been several conferences, workshops, forums and similar meetings to come up with a unified guideline on what UX is about. Despite all this gathering, a consensual definition of UX could not be determined [18]. Therefore, the UX is a widespread term.

Why is it so hard to be unified with this definition about UX? UX is much about fuzzy and dynamic concepts. Many of the parameters for UX are based on people’s thoughts, experience, and emotions. The wildly spread opinion of what making the user feel good, educe to a concept that is difficult to determine [18]. However, some authors have come up with established concepts that can work as guidelines when designing UX designs. One of the most famous UX designers is Jakob Nielsen. In 1994 Nielsen came up with 10 usability heuristics.

These heuristics have been used wildly and are well acknowledged when designing user- friendly designs [17].

Nielsen defines the concept of UX as all that makes the user feel good. A user-friendly design is not noticed by the user, everything working fine, and the user don’t feel bad about the design. One user can feel good about a design while another is feeling the opposite, the UX is a personal opinion but a good UX fulfils most users need. According to Nielsen a useful design includes two terms to consider: Utility and Usability. Utility is all functionality in a design, such as tools and if they fulfil the user need. Usability is defined in five terms and refers to how easy and pleasant the design is for user [19]. The terms are described in Table 1.

TABLE 1.USABILITY FOR USER EXPERIENCE [19]

Usability Description

Learnability First time a user enters the design, how easy is it to achieve basic tasks.

Efficiency How quick can a user perform tasks after they learned the design.

Memorability How much does the user remember and re-establish proficiency of the design if they had a pause from it for a while.

Errors How often does a user make errors, if they happen how severe they are, and how quickly can a user recover if an error occurs.

Satisfaction How satisfied is the user about the design.

This is a guideline for one process when designing a user-friendly design. However, a good UX design is how the user feels about the design. A well-designed application that looks terrible can make it a bad UX design. The design also needs to be aesthetical appealing.

Before starting to design for users, three considerations should be evaluated. What and How as normally what there in products but there is also one part, Why.

What: What can a user do with these designs, for example, send a message.

How: How to interact with the design.

Why: Why are we creating these designs? Needs user have. Before designing what and how, design why and figure out the need of the user. [20]

There are many different opinions of what UX is. One acknowledges way of representing UX is via the honeycomb. The honeycomb is developed by Peter Moreville in 2004 as shown in Figure 2 and explained in Table 2.

2.4 Categorizing data Theoretical Framework 7

FIGURE 2.HONEYCOMB DESIGNED BY PETER MOREVILLE [21].

TABLE 2.DESCRIBES THE HONEYCOMB [21].

Factor Description

Useful Who is going to use this product? Is this product or system useful for anyone?

Desirable Why should the user choose this product? The power and value of the brand, identity, image, and other elements make a user choose a more expensive less user-friendly design instead of a cheap good user-friendly design.

Usable Is the design easy to use, is it intuitive?

Valuable The product must fulfil the value for the user but also the value for the business.

Findable The navigation of the product, user finds what they are looking for.

Accessible Consider all kinds of a user when designing the products, prioritize the one with disabilities.

Credible What makes the users trust in the product, for this the information architecture, design and content are important bricks.

2.4 CATEGORIZING DATA

There are several different ways to figure out how to categorize big amounts of data and represent then intuitively. One of the most common method is the card sorting technique.

Card sorting technique is going to be used in this work.

2.5 Advertising display formats Theoretical Framework 8 2.4.1 CARD SORTING TECHNIQUE

The method of card sorting can be applied to grouping, ordering and/or giving a name to objects or concepts. Card sorting is often applied in user-centred design and is a popular technique to understand the mental factor behind the structure of a website [22]. An example of this is the following. How should the menu for a car sale site look? Should the menu be categorized in type of car (cab, SUV, family, etc.)? Should it be categorized in car brand or should it be categorized in carbon dioxide emissions? Also, what should be the sub menu for these? Card sorting can also be used to find out what sorting and filtering should include. The concepts behind card sorting are to make a large number of cards with statements or products written on them. Then, it’s up to user/testing person to categorize the cards in a way that fits their thoughts.

There are two mayor techniques used when doing user tests.

• Closed card sorting. With this technique, the user is shown predefined categorize to put the card in.

• Open card sorting. This technique mixes all cards together and let the user make their own categories [23].

After applying one of the techniques, data can be collected in two different ways.

• Qualitative collecting of data is looking for some kind of pattern from user tests.

• The other way to collect data is quantitative. Quantitative collecting uses different tools to analyze the users sorting. When using quantitative method, a clustering analysis, and a similarity matrix can be executed for computing the categories and the structure for website/sorting functions/filtering functions. From quantitative data, a result can also be found by looking at the distance users and placing the cards from each together. Also, how many steps it takes to sort the cards the same way as other users [22]?

Card sorting is a noble way to design a structure of websites and figure out categorists. Card sorting method is not always working; it can be the choice of technique or users have a big disagreement on categories for the cards. Another issue could be cultural disagreement. All these issues make it difficult a for designer to just take the result and do website/sorting/filtering from that [22].

2.5 ADVERTISING DISPLAY FORMATS

One reason for companies to make an ad is to have an economic growth. This could be done by two approaches; one is to try to persuade clients to buy their products/ services. The other one is to make an informational ad. Informational ad lets companies create a bound to their consumers [24].

2.5 Advertising display formats Theoretical Framework 9 Publishing ads through internet can be done in several ways. New ways are coming over time and bellow there are some examples of existing today.

• Banners

• Pop-ups

• E-mails

• Search engine results

• Video streams

• Social media posts Let us describe these examples.

2.5.1 BANNERS

The first form of ads on internet was appearing as simple rectangles and squares also called banners [13]. A banner is a rectangular box and its function is to be a placeholder, for example to an advertising campaign. The goal with a banner is to lead visitors to press on the banner and to redirect user to advertiser’s destination (E.g. website, web shop, etc.).

For advertising on internet, banners are the most common display format. There are three types of placing banners on web.

• Keyword banners are used for companies that looking to have a keyword to their ads, users enter a specific word in the search engine and this banner will appear according to the keyword.

• Random banners have no target destination. The banner appears randomly on predefined places on websites. This is often used to introduce a new product in the market. It doesn’t follow any specific pattern.

• Fixed banners cannot be moved and are fixed to specific places on a website.

Banners can easily be customized. It is often a video or some rich coloured image. This will lead the user to press at the banner. The banner is often visible for a long period of time. The most advantageous with a banner is the ability it has to redirect the user to the company’s website.

Banners are expensive and advertising budget might deplete. The limitation of space on a host website can make it hard to place the banner in a good place for a low cost. There are a couple of methods to place a banner on a publisher’s website to a high benefit. One method to place banners cost efficiently is to exchange them between websites. Company A can let company B have their banner on A’s website, if B lets A have it on their. This method is cost-effective, but it has a problem. If website A has much higher traffic then B, there is no benefit trade for company A. Solution to this is to give to each company credit depending on how much traffic they have also and how on long a banner is up on the website [24].

There is no standard for the size a banner should have. But Google ads have come up with five sizes that they call “top-performing ad sizes” and say that those perform best.

2.5 Advertising display formats Theoretical Framework 10

• “Medium rectangle” is 300x250 pixels. This one performs best in embedded text areas also at the end of articles.

• “Large rectangle” is 336x280 pixels. This one performs best in embedded text areas also at the end of articles.

• “Leaderboard” is 728x90 pixels. This one performs best above the main content or in forum sites.

• “Half-page” is 300x600 pixels. Says to be the fastest growing and are the most impactful size to brand advertising.

• “Large mobile banner” is 320x100 pixels. This one is just used in mobile phones. It has twice the size of a standard mobile banner [25].

2.5.2 POP-UPS [24]

A common expression for pop up is ad spawning. A pop up appears when a user enters a site it can also appear if the user does some action. This method is annoying, substantial number of users do not like it.

2.5.3 OTHER FORMATS [24]

Email advertising: sending out an email with advertising.

Search engine advertising: Make ads depending on a search word.

Video advertising: Make ads before under or after a video.

Social media advertising: Put ads on social media like Facebook, Instagram, Snapchat, linked in and so on.

There are three common navigation systems to push user through the web into an advertised website.

• Display advertising, this method put ads on good visible parts of a website, like buttons, skyscraper, and banners.

• Networking and affiliation advertising find a group of similar websites in contents and links them together. The ads are shown in the form of hypertexts and are placed inside the content of the host website.

• The last technique is called Searched-Based Advertising. This one places advertisers on web sites that are matched with a search word query.

2.6 Digital advertising process Theoretical Framework 11

2.6 DIGITAL ADVERTISING PROCESS

WWW (World Wide Webb) had over 1.75 billion websites at 2017 [7]. For an advertiser to find right websites and put ad on those sites is time-consuming. To assist advertisers in finding a large collection of websites with similar content, ad networks was created. The ad network is an intermediary between advertisers and publishers. The price for a publisher’s ad spot is determined by ad exchange. The ad exchange is similar to the stock exchange where the highest bidding advertiser gets the spot on the publisher's site [26]. Contextual advertising platforms is a simplified version of the digital advertising process. Figure 3 is a visualization of a digital advertising process. Let us define the terms used on the visualization.

RTB (real-time bidding) [26]

The ad exchange works with RTB (real time bidding) system. While a user is waiting on the browser to load, RTB checks for advertisers willing to pay for ad spot on that site. The advertiser with the highest bid gets the ad spot. The biding is done rapidly.

Programmatic [26]

Making a regular ad like putting up a poster on a store. There is a seller for the ad space and a buyer for this ad space. Selling and buying on ad space on a technical platform is named programmatic. Programmatic is a programming way to choose which advertising spot should be bought depending on several criteria. These criteria can be, the user's geographic place, gender, age, interest, etc. Programmatic is often working with an RTB system to buy ad spots, programmatic can also buy ad spots with a fixed cost.

DSP (Demand Side Platform) [27]

Advertisers storing their campaigns on ad servers. The ad server is connected to an DSP (Demand Side Platform). DSP is helping advertisers to find the right publishers and make an offer to ad exchange for an ad spot.

SSP (Supply Side Platform) [27]

DSP is the platform for advertisers, SSP (Supply Side Platform) is the platform for publishers.

SSP helps the publisher to reach out and to get millions of impressions each day. SSP sells advertisers’ spot to the highest bidding.

DMP (Data Management Platform) [27]

Information about clients could be provided through DMP (Data Management Platform). DMP gathers first, second- and third-part information via cookies and deliver information about the client’s interest, gender, geographic, etc. to DSP or SSP.

2.7 User stories Theoretical Framework 12

FIGURE 3.DIGITAL ADVERTISING PROCESS.

2.7 USER TEST

To determine if a design is user-friendly or not, user tests need to be conducted. User tests can be done in multiply ways. This study will do a case study to evaluate if the design has any flaw. The most efficient number of participants in a user test is 5 according to Nielsen [28].

2.7.1 CASE STUDY

Case study is a test where a couple of tasks are given to users. The test is supervised, and notices and observations are made through the test. The test aims to find out flaws in the design. After each test with a participant, a table of how the participant have performed being filled in. After the test is done with all participants a success rate is calculated [29]. In this study it will be number of success/testcases.

2.8 USER STORIES

What are relevant user stories and how can they be used to design a user-friendly interface?

The real name for user stories is much longer, “natural language description of one or more features of a software system”. User stories are written in the perspective of the user and it is about different features that the software system is needed. The format of user stories is.

2.9 Google ads Theoretical Framework 13

• As a <role> I can <activity> so that <business value>

Where:

1. Role - Is who is the performer, like a user, teacher, musician, etc.

2. Activity – what should happen in the system.

3. Business value – why should it happen.

User stories define the need for the software, this is good for the developer to understand the need for an application. In this case, is good when creating a design for interface [30].

2.9 GOOGLE ADS

The biggest advertiser on internet is Google [6]. Inspecting how Google work with advertising and the statistics given back to the user is a good guideline. This thesis will look into Google ads. Google ads is an advertising tool created by Google. The ads added in Google Ads, are shown when a Google search result matches the ad.

To add a campaign in Google ads, first, you have three choices for your goals!

• Get more phone calls

• Sell more on the website or get more registrations

• Increase traffic to your website

If “Sell more on the website or get more registrations” is chosen, there are the following four sides to fill out.

Your company and your target group: Here you can enter your,

• company name and web place.

• Geolocation on potential customers, and a number of potential reaches is shown.

• Define product or service. Like the language of ad, show which type of clients, name of products services which marketing in the ads.

Your ads:

• Title of ad, description and where it should link.

Budget and inspection:

• How much are you ready to pay each month for the ads. Here is 3 alternatives on pay grade.

• Inspect your ad.

Configure invoice:

• How to pay for the ad [31].

After an ad is made, Google has another tool called Google analytic which can help to evaluate the ad.

2.10 Google analytic Theoretical Framework 14

2.10 ANALYSIS DATA [32]

Analyzing a campaign can be done in several ways. One of these ways is described below in four steps.

Goal: The most important aspect of analysing advertising is to figure out why advertisers make this ad. A good way to do this is to define an objective and to define goals for the campaign.

Some example of goals are as follows.

• Audience to target.

• Brand safety.

• Increase website traffic.

• Get more clickthrough threw site.

• Make a better customer relationship.

• Improve the upselling.

• Increase sales.

Creating a campaign it is important to set up goals to achieve. This is to specify if the campaign has been good or bad. Goals can show if the campaign is on the right track or if something should be changed to reach these goals [32].

Setup: There is a substantial of analytics tool, the world’s most common is Google analytic as described in section 2.11. To get information about a campaign and to analyze it, there need to be a setup with an analytic tool.

Monitor: To get some kind of feedback on how the campaign is running and to see statistics about the campaign.

Analyse: There is substantial of data to analyse when a campaign is finished. Here it's important to have set up goals to easily see if they have been fulfilled [32].

2.11 GOOGLE ANALYTIC

A free tool to use and analyse data traffic is Google analytic. From Google analytic there is overwhelming of information available. For this thesis, some information will be explained.

This information will be useful for the work with a hi-fi prototype.

Goal: When a campaign is set, there can be some milestones that are important for the campaign, like after 1000 clicks on the campaign the advertiser is notified. Goal is a good measurement to see if the campaign was successful.

Demographics: Information about gender and age of users.

Geographic: Information about where the website is shown geographically. Typical statistics shown from a geographic area is bounce rate, page per session, avg session duration, number of new users, etc.

2.11 Google analytic Theoretical Framework 15

Audience overview: Graphs on how many views the page has and when the views occur.

Audience overview also shows which site has directed the most users to the destination website.

Best Keyword: It is shown which keyword for a website has performed best.

Bounce rate: Bounce rate is how many users visit the website and go leave from it without browse through it. A low bounce rate is good because then the user stays and looks through the website. On Google, there is a feature showing which website has the highest bounce rate [33].

2.11 Google analytic Method 16

3 M

To develop a good UX design, it is favourable to have a working process to follow. When creating a User-friendly design, there are many different methods, processes, and adaptations to consider. The working process this thesis has followed is an adaptation of a popular UX design process called "the Wheel". The Wheel is a well-used design process to create a UX design [37]. In the adapted version of the Wheel, there are six stages in total, three of those stages iterate over, if needed, until a final design can be determined. The adapted version of the Wheel is shown in Figure 4. The diagram in Figure 4 uses the following blocks.

Research: To create a design for the advertising/reporting tool, suitable theory has been studied. This work has investigated what advertising on internet is, what contextual advertising is, a few parts of UX design and also techniques to categorise large amounts of data. One interview with one experienced UX designer has been held. Interviews with advertisers were supposed to held, but due to the outbreak of covid-19, this was not feasible.

Due to the cancelled interviews a form was sent out to advertisers. One advertiser had time to answer the form. The research stage has been fundamental for the work in this study.

Analysis: In this stage What, How and Why have been determined.

Design stage: In this stage a conceptual design was produced, this included the look, feeling and could be redesigned after evaluation. An investigation of what users needed determined the placement of features. This stage has a creative design thinking part included.

Prototype stage: During this stage, a hi-fi prototype was created. This stage was based on the result of the design stage.

Evaluation stage: User testing determined whether the design was good or whether an iteration was needed to be carried out again.

Final design suggestion stage: After the iteration a final hi-fi prototype was produced.

FIGURE 4.AN ADAPTED VERSION OF “THE WHEEL”[34].

3.1 Research Method 17

3.1 RESEARCH

During the first weeks of working with this thesis, research was done to collect the theory needed to progress into the analysis stage. The research includes, a literature study, one interview with an UX designer, card sorting, interviews with advertisers, and a layout based on the research.

3.1.1 INTERVIEW WITH UX DESIGNER

After research, it was shown that analysis of needs was useful to determine the design. The needs analysis involved interviewing advertisers. It was challenging to make contact with advertisers, which meant that the work remained at a standstill. In order to progress, an interview with an experienced UX designer was carried out. The UX designer provided some advice on what to do while waiting for interviews and also some general tips to consider when making a UX design.

3.1.2 CARD SORTING

Two card sorting methods were implemented. One method included making a statistic page and the other method included making a page that can add a campaign.

Information that has been gathered to use for card sorting has been used to categorize data in order to make a statistics page. These statistics come from Google analytic [33], meeting with the UX designer, one interview form, and the CEO at Adlede. The information from research was used to design cards for card sorting on behalf of statistic page.

Needs analysis for adding a campaign was done with help of Google Ads, the interview form and meeting with the CEO at Adlede. With the information gathered, a card sorting method for adding a campaign could be established.

3.1.3 INTERVIEWS WITH ADVERTISERS

To understand what the users request for in a management tool there was an interview form prepared, the form can be found in appendix 1. Due to covid-19 no advertiser was interested or had time for an interview. A Google form was created with the same questions and sent out to media agencies. One agency had time to answer this form.

3.2 Analysis Method 18 3.1.4 LAYOUT

To create the layout a study on competitors, such as, Google analytic, Google ad, Bidtheatre, books about marketing on internet, interviews with professionals on advertising, and card sorting to categorize data were made. This layout was created in Balsamiq.

The design decision for adding campaign is made based on card sorting, how Google ad [31]

works with their adding campaign tool, the book “the art of digital marketing: the definitive guide to creating strategic, targeted, and measurable online campaigns” [32], which points out the importance of setting good goals for campaigns. Additionally, there was a study on where web users look on websites. This study showed that 80 % of the time is spent by looking on the left side of the screen when using a webpage. Moreover an interesting conclusion from this study is that the optimal margin from the left side of screen is 600 pixels, this test is made for a 1920 pixel width screen [38]. Using all of this information, a layout for adding campaign was created.

To design the statistic page and how the information should be shown for a campaign, the card sorting was considered, and Google analytic was evaluated. Google analytic has a great deal of information that is overwhelming to navigate through. Also, the book “the art of digital marketing: the definitive guide to creating strategic, targeted, and measurable online campaigns” [32] was helpful to comprehend Google analytic. A DSP (Demand Side Platform) named Bidtheatre has been a useful reference for statistic page layout [39].

3.2 ANALYSIS

Information gathered from interviews with advertisers was supposed to be the basis to answer what, how and why. Due to covid-19 no interviews with advertisers could be implemented. Therefore what, how and why were answered by the CEO at Adlede and an expert in contextual advertising. These individuals have substantial knowledge about what advertisers are looking for when adding a campaign, as well as which information advertisers request from a statistic page.

After answering what, how and why, the layout was shown to the expert in contextual advertising. The expert offered suggestions about the layout. The CEO at Adlede also checked the layout, and in collaboration with the expert, a mock-up for the adding campaign was created.

For the statistic page there were some limitations in what kind of information Adlede was able to provide to advertisers. Therefore, some information was removed. Adlede also informed that their system is cookie-free and does not store any user data.

3.3 First iteration Method 19

3.3 FIRST ITERATION

After the analysis the iteration was initiated.

3.3.1 DESIGN

The design for adding a campaign was based on the mock-up as well as the graphic shapes and the colour Adlede have in their brand. This design was created in collaboration with a Graphic designer at Codemill. The tool used to make this design was Adobe Photoshop.

Photoshop is an image processing software that can be helpful for designing.

The statistic page for a campaign was designed based on layout, interview form and expert advice from Adlede. This design was made in Photoshop.

3.3.2 FIRST PROTOTYPE

The prototype for adding a campaign was implemented based on the design for adding a campaign. The tool used in the prototyping process was Pycharm (python development tool) with Django as framework and Wagtail as CMS. This prototype was published on the internet.

For the statistic page the design was shown to the advertiser expert and the CEO from Adlede.

After showing the design, feedback was given and with this feedback a prototype could be established. The prototype was made in Figma. Figma is an online prototyping tool.

3.3.3 EVALUATE

To evaluate the adding tool for the campaign, the prototype was published at Adlede’s website and the URL was sent out to partners and customers of Adlede. The feedback was given to the CEO at Adlede, and ways to improve were detected.

The evaluation part for the statistic page and dashboard was made through user testing. The testing technique used was a case study [29]. The approach of the test was to give the users some tasks to do. While working on the task, the users were instructed to “think out loud”.

While the users performed the test, they were observed to see whether the users

completed the task or missclicked or were confused. After eight tasks, each user was asked a question about the design and his/her opinion of the design. After the testing was done, a table of successful/unsuccessful tasks was produced. From this table a success rate was calculated.

3.4 Corona Virus Method 20

3.4 SECOND ITERATION

The second iteration was made on the adding part. For statistic the final design decision was made.

3.4.1 DESIGN

From the evaluation of adding a campaign some small design changes were made. This design changes were done based on the feedback that partners and customers had given to CEO at Adlede.

3.4.2 SECOND PROTOTYPE

The new prototype was created and published on internet.

3.4.3 EVALUATION

For adding a campaign a user was satisfied with the prototype.

3.5 FINAL DESIGN SUGGESTION

A final design decision for adding a campaign was made and it went to publication at Adledes website.

The final design decision for statistic page was not implemented. The result from evaluation was investigated and the decision to make a final design was taken.

3.6 CORONA VIRUS

In the end of 2019, a new virus from coronavirus family was discovered in Kina and it is called covid-19. Covid-19 is a recently discovered virus, therefor little research has been done on this virus [34]. In 2020 covid-19 had reach nearly the entire world and Mars 17, 2020 had 179 111 cases with 7426 death been confirmed [35]. The primary factor for death is high age in combination with other diseases such as high blood pressure, cardiovascular disease, lung disease or diabetes [36]. However, this rapid contagion of covid-19 have

3.7 Method 21 resulted in actions from the government. The government recommend people that if they feel a little sick or have the ability to work from home, they should do so [36].

Due to corona some complications came up, there was no advertiser able to make the interviews also they had no time to fill out a form. This led to some changes in the work.

Information gathered to make this tool came from Adlede and also from research instead of interviews.

4.1 Research Results 22

4 R

The results of, Research, First Iteration, Second Iteration, and Final design suggestion are summarized next.

4.1 RESEARCH

The research is based on meeting with UX designer, Card sorting technique, google form sheet and layout.

4.1.1 MEETING WITH UX DESIGNER

Meeting with UX designer was realized. The meeting was a one-hour interview. The UX designer has many years of experience in developing UX friendly applications at Codemill.

With this experience, the designer came up with some guidelines and recommendation for advancing with the work. According to the UX designer, it important to start with a good objective to know what the work is aiming for and also include what the company intend to mediate. To get inspired a recommendation was to look at similar tools and see what was good and bad with those tools. If there is too little information gathered for the tool there is no idea to make a lo-fi because it is just a waste of time and the final product will probably have a lot of flaws. The designer gave advice about how to test the product without real users and also gave some advice to make a card sorting on gathered information. The main suggestion was to always be able to motivate the design decision made in the tool. If there is difficult to get information from real users in the start, a layout with the already gather information could be established. The layout is beneficial for the position of buttons and content according to UX designer.

Summarizing the interview, the following guidelines were given by the UX designer.

• Have a good objective.

• What does Adlede intend to mediate.

• Check other similar tools and evaluate what's good and what's bad about those.

• Do not do a lo-fi if there is too little information gathered.

• Make your test about the information you can gather without the needs of users.

• Always be able to motivate the design selection.

• Make some card sorting.

• Create a layout of the site.