1. Sander Buitelaar
Master Thesis Defense
Business Information Management
September 21, 2012
Coach: Wolf Ketter
Co-reader: Stefano Puntoni
2. Outline
• Introduction
• Research Questions
• Methodology
• Conceptual Model
• Results
• Conclusion
First mobile phone:
Motorola 8000x
The Rise of Mobile Advertising 21/9/2012 2
3. Introduction
• Mobile is changing the Web
o “More users will connect to the Internet over mobile
devices than desktop PCs [by 2015].” – Mary Meeker
(2012)
• How does this change online advertising?
The Rise of Mobile Advertising 21/9/2012 3
4. Research Questions
• RQ 1: What factors impact the click-through rate
(CTR) of an online search advertisement?
• RQ 2: Does the CTR differ based on the type of
device?
• RQ 3: How do these factors differ based on the
type of device?
The Rise of Mobile Advertising 21/9/2012 4
5. Methodology
• Online advertising campaign in Google AdWords
• Laptop bags, laptop sleeves, phone cases and iPad
cases.
The Rise of Mobile Advertising 21/9/2012 5
13. Mobile Devices x Product x Price
The Rise of Mobile Advertising 21/9/2012 13
14. Conclusion
• Generalizations
o The medium matters
• Managerial Implications
o Do not include price
o Separate mobile devices and tablets
• Limitations
o Dutch market, small business, unknown brands.
• Future Research
o User behavior in mobile devices
o Tablet devices vs. smartphones
The Rise of Mobile Advertising 21/9/2012 14
15. Thank you
I will now take your questions
The Rise of Mobile Advertising 21/9/2012 15
Notes de l'éditeur
Good morning everyone.I would like to present the work I have done for my thesis – The Rise of Mobile Advertising: exploring advertising effectiveness with mobile devices.
The agenda for today will be as following, I will give some context for this topic of online advertising and mobile advertising and my motivation for this research. Then I will go over the specific research questions and the methodology that I used to answer the research questions. I will briefly explain the conceptual model and spend most of the presentation explaining the results of my experiment.I will end with generalizations, future research suggestions and managerial implications.As a small aside, the picture is of the first mobile phone, the Motorola 8000X, from 1983.
It might be difficult to see, but we are in the midst of a technological revolution. People are changing their primary computing device from PCs to smartphones. The adoption rate of mobile internet is faster than radio, TV, and desktop internet.In about 2 to 3 years more people will access the internet through mobile phones than through desktops. More people in the world will have smartphones than computers.This is important because businesses need to know how to effectively advertise and sell their products to consumers that are viewing content on smartphones, tablets, laptops and desktops.This topic is also relevant for academia because it is so recent. Smartphones and tablets are very young. Publishing a paper might take more than a year. The pace of technology is incredibly fast, it is much faster than the pace of research.For these reasons I was interested in exploring the differences in online advertising between desktops and mobile devices.
I specifically wanted to research the factors that would impact the effectiveness – or click-through rate – of an online search advertisement. For those that don’t know what the click-through rate is, that is the rate of clicks per impressions. An impression is simply the showing of an ad. So an ad that is shown once is one impression, if the ad is shown once and clicked on, has a click-through of 100% If the ad is shown ten times and clicked on once, it has a click-through rate of 10%Moreover, I wanted to see if mobile devices have a different click-through rate and if the factors differ based on the type of device.The factors that I wanted to test are: price information, product information and a call-to-action phrase, which I will explain later.
In order to answer the research questions I chose to do a field experiment. I created the online search advertising campaign using Google AdWords, the advertising program of Google. The campaign was built for a small Dutch company that sold laptop bags, laptop sleeves, phone cases and iPad cases among other accessories. These are some of the products they sell. The advertising campaign ran from the 1st of May to the 22nd of May this year. In the campaign I advertised four different product categories (laptop bags, laptop sleeves, phone cases and iPad cases) in two different types of devices with three different factors. The experimental design was a 2x2x2x2 factorial model.This might be easier to understand if you see the conceptual model.
We can see from this conceptual model, the three factors that I included in the experiment: pricing information, product information, and the call-to-action phrase. A call-to-action phrase is simply a phrase such as “order now” that implies for the user to take some action.Mobile devices would act as a moderating variable, the click-through rate is the dependent variable and is a proxy for ad effectiveness.We can also see how the advertisements appeared. For example, this ad would show for laptop sleeves with specific product information, the size of the sleeve and the price information. This other ad shows product information, price information and the call-to-action phrase.I did not include the control variable in this conceptual model, but it is worth mentioning that the average position of the ad was controlled for. Because the average position has an effect on the click-through rate.
During the three weeks there were almost 25,000 impressions and 541 clicks in total. From this total roughly two thirds went to computers and laptops and one third to mobile devices. The overall click-through rate was about 2.2%These absolute measures of impressions and clicks do not tell us much about the effectiveness, for that we need to look at the click-through rate.Smartphones had the highest CTR, followed by tablets and computers and laptops had the lowest CTR. These results don’t tell us much about the significance. In order to do that I ran an ANCOVA model with average position as the control variable.
Note that this table does not show all the factors in the model, it only shows the significant results. Note that all of these factors are significant at a p-level of 0.01. The F statistic is a measure of the significance of the result. The Partial Eta Squared is a measure of the variance attributed to the factor, thus the model itself ‘explains’ about 30% of the variance in the data. The average position explains about 23% of the variance, which is a substantial amount. Price and mobile devices were significant. We also see that there are significant interactions between product and price and also three-way interactions between product, price and mobile devices and price, CTA and mobile devices.These effects are easier to see with graphs.
From the ANCOVA results we know that price is significant and mobile devices are significant. This is easy to see here with this graph.First look at the introduction of price information in both computers and mobile devices. The difference is significant. We can also see the difference in the two lines, the top one is computers and laptops and the bottom one is mobile devices. The difference is also significant.The difference in click-through rate between mobile devices and computers is also apparent in other factors.
When we look at the product information we can still see that the difference between computers and mobile devices is quite different. Almost by a percentage point.
And also in the call-to-action,mobile devices exhibit a significantly higher gross CTR than computers and laptops.
This graph shows a significant interaction between product and price.If the ad does have product information, price increases the CTR, if the ad does not have product information, price decreases the CTR.
What is more interesting though is the two-way interaction effect between price, product and device type.On the left we have computers and laptops, on the right we have mobile devices.Previously we saw an interaction effect between price and product. This interaction itself is dependent on the device type.For computers there is no interaction between product and price. Yet on mobile devices there is an interaction between product and price.
What can we generalize from this research? First of all, medium does matter, mobile devices have a significantly higher click-through rate than computers and laptops and moderate some of the factors. The factors that impact the click-through rate of an online ad are different when mobile devices are introduced.Some implications for businesses are that price should not be included in the advertisement. And being aware that mobile devices are different should be reflected in the campaign. This simply means separating the campaigns and experimenting to see what changes. Limitations of this study are that this campaign was done for the Dutch market, for a small business with unknown brands. The results might and probably would be different in another country, with different products and with other brands.Future research should look at the question of why mobile devices exhibit a higher click-through rate, and what are the differences between tablet devices and smartphones.