Consumer Engagement Characteristics in Mobile Advertising

Lonneke Brakenhoff and Marco Spruit

Department of Information and Computing Sciences, Utrecht University, Utrecht, The Netherlands

Keywords:

Mobile Advertising, Persuasive Advertising, Consumer Engagement, MAEF, CRISP-DM.

Abstract:

Advertising on mobile devices is becoming increasingly more important as the possibilities regarding their

design and context become increasingly more extensive. This research focuses on the characteristics of design

and context regarding mobile advertisements, structured according to the CRISP-DM process model. First,

we describe their key concepts and relevant theoretical background. Then, we design the Mobile Advertising

Effectiveness Framework for Consumer Engagement (MAEF4CE), which relates medium types, creative at-

tributes, ad formats, device speciﬁc ads, and brand visibility as mobile advertisement characteristics. Finally,

we uncover the combination of characteristics that elicits optimal consumer engagement in mobile advertise-

ments in a real-time bidding dataset.

1 INTRODUCTION

The number of mobile device owners increased to

68% in 2015 for all ages (Deloitte, 2015). The biggest

growth in mobile device owners occurred in the age

category 55+. This indicates an attractive target group

available for the purpose of mobile advertising. The

advertising world is using mobile devices to reach

their target group by a speciﬁc medium: the smart-

phone. Since mobile devices are personalized in a

high degree advertisers are able to reach this target

group more efﬁciently. Another difference between

mobile advertising and traditional advertising is the

possibility to use the capabilities of the device to spec-

ify a target, such as location-based data.

Less is known about the factors that trigger a

user’s behavior when they get served an advertise-

ment (ad). In the case of mobile advertising we ﬁrst

distinguish two types of behavior. First, the user can

or does not interact with the ad. This means that the

user performs no action related to the banner and is

therefore not engaged with the ad. The second be-

havior does involve the user performing an action, for

example clicking on the ad. This interaction is called

an engagement. This research focuses on the second

behavior.

In a recent research Grewal, Bart, Spann, & Zubc-

sek (2016) looked at the broader picture of mobile ad-

vertising. Components related to mobile advertising

were mapped into the Mobile Advertising Effective-

ness Framework (MAEF) as separated components

inﬂuencing the mobile advertising process. In this

work we customize and extend MEAF to model the

inﬂuence of ad design characteristics on ad engage-

ment in mobile advertising: the MAEF4CE frame-

work. The goal of this project is to increase the num-

ber of consumer engagements. Which will be mea-

sured based on different interaction types. The fol-

lowing research question is formulated:

Which combination of mobile advertisement char-

acteristics can optimally improve consumer engage-

ment?

To answer this question propositions are deﬁned

based on the literature and the MAEF. This research is

structured according to the CRoss Industry Standard

Process for Data Mining (CRISP-DM). CRISP-DM is

a very commonly used method for knowledge discov-

ery processes (Spruit et al., 2014). The related work

is described in the following section, Section 2. This

section also contains the explanation of some deﬁni-

tions that are important for the remainder of the re-

port. Next, a theoretical background is set in Section 3

including the explanation of the MAEF and the in-

troduction of the MAEF4CE. The research approach

(Section 4) is structured according to the CRISP-DM

phases: Domain Understanding, Data Understanding,

Data Preparation, Modeling, Evaluation, and at last

the Deployment phase. The results are discussed in

Section 5 followed by the conclusion and suggestion

of further research in Section 6.

Brakenhoff L. and Spruit M.

Consumer Engagement Characteristics in Mobile Advertising.

DOI: 10.5220/0006499602060214

In Proceedings of the 9th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (KDIR 2017), pages 206-214

ISBN: 978-989-758-271-4

2 KEY CONCEPTS

Consumer Behavior. Generally, consumer behavior

is based on the consumer‘s buying behavior. Since

the purpose of mobile advertising is not persuading

a user to directly buy a product, the approach of this

research is focused on user awareness.

Mobile Campaign. An ad is part of a campaign.

A campaign is a gathering of one of more ads that

are used to persuade a target group to engage with a

brand, product or service. An advertisement is de-

ﬁned by Flores, Chen & Ross (2014) as “an online

advertising space that typically consists of a combi-

nation of graphic and textual content and contains an

internal link to target ad pages or an external link to

the advertiser’s website via a click through URL“. A

mobile application can be added as a medium to dis-

play ads for mobile advertising besides websites. An

ad that is made viewable in a website or application is

called an impression. The ads within a campaign can

be different. This means that the format, design and

layout can differ between the ads of a campaign.

When a user interacts with an ad, such as by click-

ing or swiping, the user is engaged with the ad. The

speciﬁc engagement is measured and therewith avail-

able for further analysis. The engagement measure-

ments say something about the performance of an ad.

The Click-Through-Rate (CTR) is the measurement

that is most used in mobile advertising and indicates

the performance by providing a ratio of the number of

times the ad is served and the number of times the ad

is actually clicked.

Mobile User. A mobile user is referred to as a person

that owns and uses a mobile device. This research

will be mainly focused on smartphone users. These

mobile users are able to receive advertisements (ads)

during the use of an application and/or mobile Web

browser.

Brand Awareness. In this research brand aware-

ness is deﬁned according to the deﬁnition of Hoyer &

Brown (1990) and is as follows: brand awareness is

as a rudimentary level of brand knowledge involving,

at least recognition of the brand name.

3 THEORETICAL BACKGROUND

3.1 Persuasive Technology

New technologies are more and more focused on in-

ﬂuencing the behavior of consumers, which is gen-

erally referred to as persuasive technology (Fogg,

2002). According to Fogg (2002) , the Internet had

a great inﬂuence on the growth of persuasive tech-

nologies. As an example he mentions the example

of persuading Webshop visitors by making sugges-

tions to buy more products based on their prefer-

ences (i.e. up-selling). The Internet also enables an-

other medium of approaching consumers: advertis-

ing. Fogg (2002) states that persuasive technology

is more effective when it is interactive. This charac-

teristic distinguishes computer technologies from tra-

ditional technologies, such as television commercial

and paper advertisements. Computer technology al-

lows its users to interact, and therefore is able to adapt

its content to the user. Additionally, computer tech-

nology is ubiquitous and versatile, and is able to gen-

erate and store lots of data (Fogg, 2002). Lane (2010)

states that mobile devices with its sensors can create

user proﬁles for advertising purposes.

Persuasive Advertising. Persuasive Advertising is

deﬁned by Armstrong (2010) as advertising that in-

tends to inﬂuence other through all types of media.

This media includes both traditional as ‘new‘, mod-

ern media, like the Internet.

Mobile Advertising. The effectiveness of mobile

advertising is measured by Goh, Chu & Wu (2015)

by looking at the search behavior and advertising re-

sponse of people who got served with a mobile ad.

Goh et al. (2015) describe the beneﬁts of mobile ad-

vertising, compared to the Internet market, as a tar-

geted marketing strategy. Additionally, nowadays a

lot of people own a smartphone (Lane et al., 2010).

Location targeting and being personal are the main

beneﬁts of mobile devices due to the extensive num-

ber of sensors in mobile devices (Grewal et al., 2016;

Lane et al., 2010). Goh et al. (2016), Wong, Tan,

Tan and Ooi (2015) , and Lane (2010) conclude that

mobile devices therefore are the most relevant media

regarding advertising since it allows the advertiser to

serve more tailor made advertisements to their con-

sumers.

3.2 MAEF

As mentioned earlier Grewal et al. (2016) de-

signed the Mobile Advertising Effectiveness Frame-

work (MAEF). The MAEF is a framework that maps

the components involved in the setup and design of

an advertisement and its purpose is to create a re-

search agenda. The components deﬁned by Grewal

et al. (2016) are: context, consumer, ad goal, market

factors, ad elements, and outcome metrics.

The following Section (3.3) will ﬁrst simplify the

framework to better ﬁt the more narrow scope of this

research. Then, the MAEF is expanded with some

additional ad elements and metrics.

3.3 MAEF for Consumer Engagement:

MAEF4CE

The MAEF model shown is too extensive and unfo-

cused for our research application. Therefore we ﬁrst

customize it by removing various components that are

not applicable to the domain of this research. For ex-

ample, the ‘Market Factors‘ are being disregarded due

to the low impact on the ad goal and the ad elements

to limit the scope due to page restrictions.

3.3.1 Context and Consumer

The MAEF includes two components that inﬂuence

the ad goal directly. First, they split the ‘context’ of

an ad in two types: environmental context and tech-

nology context. The environmental context contains

factors, such as location, time, weather, and events,

that described the environment in which the mobile

device is located at the moment an ad is, or can be

displayed (Grewal et al., 2016). The second type of

context is technology context and is based on the spec-

iﬁcations of the mobile device, such as the size of the

screen and the location (website or application) they

come from. They found that the context in which an

ad is served is very important regarding the behavior

of the consumer towards the ad.

The second component ‘consumer‘ contains infor-

mation about the consumer. Besides the information

of the current state of the consumer in the ‘customer

journey‘, it also contains information about the his-

tory of the consumer and possibly demographic in-

formation (Grewal et al., 2016).

The context and consumer components contribute

to the determination of the goal. The components al-

low an advertiser to specify its target based on data

that is generated by mobile devices. These compo-

nents are not further speciﬁed in this project, but are

mentioned in the model since their inﬂuence, pro-

vided that they are chosen accurately, can be of great

inﬂuence on ad performance (see column A of Fig-

ure 1).

3.3.2 Ad Goal

The ad goal is what the initiator of a mobile campaign

aims for. It determines the target of the campaign.

The MAEF mentions multiple ad goals, however sev-

eral elements are not included in this project. The fo-

cus is on the ‘engagement goal’ and therefore the rest

of the outcome metrics are removed from the frame-

work (see column B of Figure 1).

3.3.3 Ad Elements

The Ad Elements in the MAEF are elements that char-

acterize the look and feel of an ad and from now on

referred to as design characteristics. The characteris-

tics given by Grewal et al. (2016) are the ad medium,

media type, pull versus push, interactive versus static,

and promotional elements.

Ad Medium. Grewal et al. (2016) mentions that a

mobile device can function as different ad medium.

First, the mobile device can be the only screen the

consumer is working with, but can also be the sec-

ond screen besides, for example, a television. In this

project we assume that the mobile device serves the

purpose of the ﬁrst and only screen and therefore ad

medium is removed from the MAEF.

Pull/Push. In the MAEF distinction is made between

pull and push methods for advertising. With the pull-

based method the consumers pull the advertisement

by opening a speciﬁc mobile website or application

(Grewal et al., 2016). The second method is the push-

based method, and relies on the delivery methods

such as SMS. This project is only focused on the pull-

based methods and therefore the pull/push element is

removed from the MAEF.

Promotional Element. An ad might contain a pro-

motional element. This means that an ad can contain

discount offers for the consumer (Grewal et al., 2016).

This element is not taken into account in this project

since the purpose of the ads are more often to create

trafﬁc to the website of the client.

The ad elements described below are included in

the framework, as shown in column C of Figure 1.

Medium Type. The medium type is the channel

through which the ad is served to a consumer. The

types can be a mobile Webpage (site) or a mobile ap-

plication (app). Both are opened by the consumer it-

self. The content of the Webpage or application can

inﬂuence the perception of an ad (Grewal et al., 2016).

Since the content of an app is more personalized, the

user installed the app itself (Grewal et al., 2016), we

expect that the ad is more personalized and therefore

more likely to be interacted with.

Proposition 1. An ad that is served in an app elicits

better consumer engagement than an ad served on a

site.

Creative Attributes. Grewal et al. (2016) distin-

guishes four types of creative attributes (IAB Ned-

erland, 2014) in mobile ad: static, dynamic, interac-

tive, and ads containing a video. In the MAEF the

creative attributes are indicated as interactive / static.

The type of creative in an ad can be important regard-

ing how inviting the ad is for interaction (IAB Ned-

erland, 2014). According to de Sa, Navalpakkam, &

Figure 1: MAEF4CE: Mobile Advertising Effectiveness Framework for Customer Engagement.

Churchill (2013) an in-ad animation leads to a nega-

tive impact on user experience than ads that are not

animated. Based on Sa et al. (2013) we expect that a

static ad will elicit the highest consumer engagement

compared to a interactive, dynamic, or video ad due

to the possible negative impact of animated ads.

Proposition 2. A static ad will elicit better consumer

engagement compared to a dynamic ad.

Ad Format. According to a study performed by de Sa

et al. (2013) is an smaller ad preferred over a larger

ad. We expect that large ad sizes elicit better con-

sumer engagement since a large banner size is com-

pletely ﬁlling up the device screen.

Proposition 3. A ‘large’ medium type will elicit

better consumer engagement compared to the other

medium types.

Device Speciﬁc Ads. The size of the mobile device

limits down the space for mobile ads (Grewal et al.,

2016). Since there are major differences in screen size

of mobile devices it can be useful to create ads that are

speciﬁc for certain screen sizes.

Proposition 4. An ad that is screen speciﬁc will result

in a higher consumer engagement than an ad that is

not screen speciﬁc.

Brand Visibility. Ghose & Todri (2015) concluded

that advertising contributes to an increased interest in

a brand. This ad element researches if the presence of

a brand logo contributes to the recognition of a brand.

According to Flores et al. (2014) does an ad contain-

ing images of the brand of the product positively in-

ﬂuenced attitude towards the brand.

Proposition 5. A visible logo within an ad results in

a higher consumer engagement than an ad without a

logo.

3.3.4 Outcome Metrics

The outcome metrics are used to determine the result

of serving a viewable ad to consumers. The metrics

are explained below and shown in column D of Fig-

ure 1.

Viewability. This outcome metric is not described

in the MAEF. Viewability is whether the ad is di-

rectly viewable, without performing an action by the

consumer, or is viewable as a result of the consumer

scrolling down. In this project all the ads used for

analysis are viewable. This means that the ad is view-

able the moment the consumer opens an application

or site. As a result of the guaranteed viewability we

can assume that the ad is seen by the consumer, be-

cause the ad was visible on a consumer’s screen area

(Ghose and Todri, 2015). An ad is considered to be

viewable when at least 50% of the ad is visible for at

least 1 second, according to the Media Ratings Coun-

cil (MRC) (Ghose and Todri, 2015; IAB Nederland,

2015).

Interaction. Here, an interaction is considered as a

measurable type of engagement. When an ad is actu-

ally visible to users.

The ratio is a composition of multiple interaction

types. The interactions measured in this research are:

i) installs, ii) clicks, and iii) events.

Installs. An install is an engagement in the form of

an application installation as a result of an interaction

with an ad. When the application is actually installed

an ‘install’ event is registered.

Clicks. Clicks are the simplest form of interaction

of the user and a served ad. When a user clicks on

the ad a click event is registered. This ‘click through’

event often results in the navigation to another (web)

page of the advertiser. In this project a click is cate-

gorized as an engagement.

Events. Besides install and click events other in-

teractions can be elicited, such as dragging and swip-

ing.

4 RESEARCH APPROACH

Quantitative research design is used to determine

the relationship between advertisement characteris-

tics and consumer engagement. Multiple statistical

tests are performed to determine this relationships.

The structure of the methodology, as described in the

Introduction, is according to CRISP-DM. The follow-

ing order is used for the research approach descrip-

tion: Domain Understanding, Data Understanding,

Data Preparation, Modeling, and Results (evaluation).

The last, deployment, phase of the CRISP-DM will be

absent as this paper is an exploitative research without

an implementation.

4.1 Domain Understanding

Two parties can be distinguished in the process of

serving an ad in a mobile device context (see Fig-

ure 2). First the one which provides the ad, the ad-

vertiser, and second, the party that offers an ad space

in an application or site, the publisher. However, be-

tween the advertiser and the publisher are two service

parties that make sure the ad is served in the right

space, the bidding service and the auction service.

When an advertiser is going to use mobile adver-

tising, the ad needs to be served in mobile contexts.

Before a mobile ad, or impression, is served in a mo-

bile application or mobile website, a Real-Time Bid-

ding (RTB) process is performed. The Interactive Ad-

vertising Bureau (IAB) creates and maintains stan-

dards and speciﬁcations in order to improve the ad

network landscape and describes RTB as bidding for

individual impressions in real-time (IAB Nederland,

2014).

The advertiser is not directly involved in the auc-

tion, but is outsourcing the bidding process to a ser-

vice on the Demand-Side Platform (DSP). Companies

on the DSP compete in auctions to obtain an unique

impression. This means that the advertiser is waiting

until the auction for one unique banner is ﬁnished in

order to serve an ad (IAB Nederland, 2014). The auc-

tion is ﬁnished when the highest bidder is determined.

The decisions made regarding the participation in the

auction and its result is based on computer algorithms,

so there is no direct human input required (Yuan et al.,

2013).

On the other hand, opposite to the advertiser and

the DSP, are the publisher that offers the impressions,

and the Supply-Side Platform (SSP) that conducts an

auction among bidders on the DSP per impression.

4.2 Data Understanding

The data used in the research is made available by a

company operating in the Dutch mobile advertising

landscape, Mobile Professionals (MobPro). MobPro

has build its own bidding server in order to participate

in the RTB process. MobPro is mostly operating on a

DSP. The data that is needed for this research is a re-

sult of three types of information: i) bid request infor-

mation, ii) design characteristics, and iii) engagement

information.

First, each impression has information that was

generated during the RTB process consisting of infor-

mation about the context of the prospective impres-

sion, such as the location of the user and device spec-

iﬁcations. This information is stored in the (1) ‘auc-

tion’ table in Google BigQuery. BigQuery enables us

to perform SQL-queries for data analyses. Second,

some visual information is available of the ad that

is shown during the impression, (2) the ‘view’ table.

These are the design characteristics. The ad is only

considered an impression when the design character-

istics, speciﬁed by the advertiser and the designer,

meet the speciﬁcations of the request. It is only pos-

sible that a certain design is shown to a user if the

design ﬁts the available ad space. Finally, data that

is generated during the impression is stored in the (3)

‘click’ or ‘event’ table. This is the actual data about

how the ad is perceived by the user and contains in-

formation about an engagement. The ﬁrst two cate-

gories are constructing the independent variables, the

last one forms the dependent variables.

4.3 Data Preparation

MobPro uses a system to store information about all

the incoming bid requests. A bid request is the call

from the bidding servers that an ad space is available.

The data used in this research is only containing won

bid requests, since only the bids that are won can be

ﬁlled with an ad. This means that an incoming bid

request is accepted, bid on, and the bid is won ac-

cording to the highest bid respectively. This process

takes place in the RTB process explained earlier.

Due to the large amount of data the dataset is that

is used is reduced to limit number of bid request.

First, we are using the auction data table (the incom-

ing bid requests) of one day (2017-01-26). Since we

are only interested in the ads that were actually view-

able to users, the auction table is combined with the

table containing the views. Joining the auction table

and the view table results in a table with 3.16 million

won bids on that day. In this step, for ad format and

medium type, the design characteristics are directly

ﬁltered from the auction table. This means that the

total numbers of views can differ per dataset for each

hypotheses. Overall, the number of viewable ads was

more than one million.

In the next step a distinction was made between

the characteristics. For the creative attributes and

brand visibility a list with campaigns was manually

checked to assign the right value for each ad (N =

151). This means that each ad was checked and cat-

egorized for with the right characteristic. Since each

Figure 2: Summary of the Real-Time Bidding process and the involved parties.

ad in the dataset contained a logo or other brand in-

dication, a distinction was made between whether a

logo was visible more than 90% of the time (‘long’)

or less than 50% of the time (‘short’).

For medium type and ad format the following ap-

proach was used. A new table was created for each

for each possible value of a category while combin-

ing the views with the engagement data based on ad

ID. Staying with the example of the ad format, three

tables were created for each ad size and for each ad

size an extra column was added with a ‘1’ or ‘0’ for

indicating the an engagement or not.

Unfortunately, the data for the device speciﬁc ads

could not be gathered from the available resources

due unavailability of data and time limit. As a result

that proposition 4 can not be tested.

4.4 Data Modeling

Both independent variables (clicking) and depen-

dent variables (ad characteristics) are categorical data.

First, to test the propositions deﬁned in Section 3.3

null hypotheses and alternative hypothesis are formu-

lated for each proposition. To analyze the hypotheses

the signiﬁcance level is 0.05 is used. Using the sam-

ple data, we will conduct a chi-square test for inde-

pendence. The statistics are calculated using Python

packages. The following packages are used: pandas

for data preparation and visualization, numpy for pro-

cessing calculations, scipy as statistics library, and

matplotlib for visualization of the results.

5 RESULTS

In this section the evaluation phase of the CRISP-DM

is described by examine the stated propositions by ap-

plying the previous explained statistics.

Medium Type

The following hypotheses are formulated to test if

medium type has an inﬂuence on consumer engage-

ment (see Proposition 1):

. Consumer engagement and medium type are in-

dependent.

. Consumer engagement and medium are not inde-

pendent.

The number of degrees of freedom is 4 which

gives critical value of 9.49 at 95% signiﬁcance level.

The chi square statistic of 373.8271 (n = 1016988)

exceeds the critical value and the P −value (0.0) less

than the signiﬁcance level (0.05), we reject the ﬁrst

hypothesis. Thus, we conclude that there is a rela-

tionship between medium type and clicking.

Figure 3 shows the relative frequency of the two

medium types. The ‘site’ medium is clicked more of-

ten compared to the ‘app’ medium (see Figure 3).

Figure 3: The distribution of medium type in app and site.

Creative Attribute

The following hypotheses are formulated to test if

the type of creative attribute has an inﬂuence on con-

sumer engagement (see Proposition 2):

. Consumer engagement and type of creative at-

tribute are independent.

. Consumer engagement and type of creative at-

tribute are not independent.

The number of degrees of freedom is 6 which

gives critical value of 12.59 at 95% signiﬁcance level.

The chi square statistic of 42.9243 (n = 306450) ex-

ceeds the critical value and the P − value (0.0) less

than the signiﬁcance level (0.05), we reject the ﬁrst

hypothesis. Thus, we conclude that there is a rela-

tionship between creative attribute and clicking.

Figure 4 shows the relative frequency of the three

types creative attributes. The ‘interactive’ advertise-

ment is slightly clicked more often compared to the

’video’ and ’dynamic’ ads (see Figure 4).

Figure 4: The distribution of creative attribute in dynamic,

interactive, and video.

Advertising Format

The following hypotheses are formulated to test if

advertising format has an inﬂuence on consumer en-

gagement (see Proposition 3):

. Consumer engagement and advertising format

are independent.

. Consumer engagement and advertising format

are not independent.

The critical value is 12.59 at 95% signiﬁcance

level when number of degrees of freedom is 6.

Calculating the chi-square distribution for P(X

2554, 77) = 0 (n = 1321890). The chi-square exceeds

the critical value and the P − value (0.0) less than the

signiﬁcance level (0.05), we reject the ﬁrst hypothe-

sis. We conclude that there is a relationship between

advertising format and consumer engagement.

Figure 5 shows the relative frequency of the three

ad formats. The ‘large’ ad format is clicked more of-

ten compared to a ‘small’ and ‘medium’ ad.

Figure 5: The relative frequency of the three banner for-

mats.

Brand Visibility

The following hypotheses are formulated to test if

brand visibility has an inﬂuence on consumer engage-

ment (see Proposition 5):

. Consumer engagement and brand visibility are

independent.

. Consumer engagement and brand visibility are

not independent.

The number of degrees of freedom is 4 which

gives critical value of 9.49 at 95% signiﬁcance level.

Since the chi square statistic of 77.04 (n = 306.450)

exceeds the critical value we reject the ﬁrst hypothe-

sis. We conclude that there is a relationship between

brand visibility and consumer engagement.

Figure 6 shows the relative frequency of the ‘long’

or ‘short’ visibility of a brand. This means that the ﬁg-

ure shows the percentages of clicks compared to the

total number of views for each speciﬁc brand visibil-

ity option.

The ads where the brand is shown less than 50%

of the time are clicked more often compared to the ads

where the brand was visible almost the whole time the

ad was served (see Figure 6).

Figure 6: The clicks in percentages for on (non-)visibility

of a recognizable brand in an advertisement.

6 CONCLUSION

All characteristics are not independent with consumer

engagement, which means that they have a relation-

ship with consumer engagement.

Proposition 1 stated that an ad in an app will elicit

better consumer engagement, however, according to

the results advertisements on a mobile site did gen-

erate more clicks. This proposition turned out to be

false. In Proposition 2 we assumed that a static ad

would be clicked more often. However, no static adds

were found on in the data selection. Interactive ads

did elicit the best consumer engagement. It turned

out that an ad with a short brand visibility generated

more consumer engagement compared to being the

brand always visible (Proposition 5). Proposition 3

turned out the be true. The large ads elicits, com-

pared to small and medium formats, more consumer

engagement. At last, short brand visibility receives

more clicks than long brand visibility. Proposition 5

can not be answered as true or false since all ads did

show a logo.

7 DISCUSSION

Two of the four tested propositions (1 and 2) were

stated incorrect. First, an ad in a mobile site is more

clicked than an app. So while an app is considered

to be more personal (e.g. having an app is the con-

sumer its own choice), its use is goal oriented since

it will ﬁt the purpose of the app. Therefore, an ad

opposed to browsing on a site, its content might vary

more and ads are thus only additional varying con-

tent. Secondly, since no ‘static’ ads were included in

the evaluation, another creative attribute elicited more

consumer engagement: ‘interactive’ ads. Since the

purpose of interactive ads is to engage interaction this

output seems valid. Additionally, an interactive ad is

relative newly creative type which can result in a lim-

ited coverage of the type in the literature.

7.1 Limitations

The data is only based on one day due to process lim-

itations. So, different days of the week are not taken

into account.

No information is available of the actions of the

consumer after the click-out. This means, that we

could not state anything on how effective the adver-

tisement was regarding the engagement with the ad-

vertiser outside the advertisement.

Additionally, inﬂuences of other advertising me-

dia for the same campaign at the same time is not

taken into account. For example, when a campaign

for a certain brand is served on both mobile and tele-

vision, brand awareness can be inﬂuenced by multiple

channels.

7.2 Further Research

The ‘context’ and ‘consumer’ components of the

MAEF are not tested in this project, but are highly in-

teresting within this scope. An additional study could

be the combination of the context or consumer char-

acteristics regarding ad performance.

REFERENCES

Armstrong, J. S. (2010). Persuasive advertising: Evidence-

based principles. Palgrave Macmillan.

de Sa, M., Navalpakkam, V., and Churchill, E. F. (2013).

Mobile advertising: evaluating the effects of anima-

tion, user and content relevance. In Proceedings of the

SIGCHI Conference on Human Factors in Computing

Systems, pages 2487–2496. ACM.

Deloitte (2015). Global mobile consumer survey 2015: The

smart mainstream. Technical report, Deloitte.

Flores, W., Chen, J.-C. V., and Ross, W. H. (2014). The

effect of variations in banner ad, type of product,

website context, and language of advertising on in-

ternet users attitudes. Computers in Human Behavior,

31:37–47.

Fogg, B. J. (2002). Persuasive technology: using computers

to change what we think and do. Ubiquity, 2002(De-

cember):5.

Ghose, A. and Todri, V. (2015). Towards a digital attribution

model: Measuring the impact of display advertising

on online consumer behavior.

Goh, K.-Y., Chu, J., and Wu, J. (2015). Mobile advertising:

an empirical study of temporal and spatial differences

in search behavior and advertising response. Journal

of Interactive Marketing, 30:34–45.

Grewal, D., Bart, Y., Spann, M., and Zubcsek, P. P.

(2016). Mobile advertising: A framework and re-

search agenda. Journal of Interactive Marketing,

34:3–14.

Hoyer, W. D. and Brown, S. P. (1990). Effects of brand

awareness on choice for a common, repeat-purchase

product. Journal of consumer research, 17(2):141–

148.

IAB Nederland (2014). OpenRTB API Speciﬁcation. IAB

Nederland, version 2.3 edition. Final Draft.

IAB Nederland (2015). State of viewabil-

ity transaction 2015. Retrieved from:

https://www.iab.com/guidelines/state-of-viewability-

transaction-2015/.

Lane, N. D., Miluzzo, E., Lu, H., Peebles, D., Choudhury,

T., and Campbell, A. T. (2010). A survey of mo-

bile phone sensing. IEEE Communications magazine,

48(9).

Spruit, M., Vroon, R., and Batenburg, R. (2014). Towards

healthcare business intelligence in long-term care: An

explorative case study in the netherlands. Computers

in Human Behavior, 30:698–707.

Wong, C.-H., Tan, G. W.-H., Tan, B.-I., and Ooi, K.-B.

(2015). Mobile advertising: the changing landscape

of the advertising industry. Telematics and Informat-

ics, 32(4):720–734.

Yuan, S., Wang, J., and Zhao, X. (2013). Real-time bidding

for online advertising: measurement and analysis. In

Proceedings of the Seventh International Workshop on

Data Mining for Online Advertising, page 3. ACM.