Mobile usage as a proportion of Internet traffic continues to skyrocket, but mobile advertising often gets the rap for being ineffective. In this article, Professor Anindya Ghose unravels the myths and paradoxes of mobile marketing through findings from a series of studies conducted in four different countries − the United States (US), Germany, China and South Korea. By conducting various randomised experiments and deploying sophisticated econometric models on historical data, brands can determine the right time, right place and right device to reach the right customer.
One of the most exciting developments in the digital media arena is the explosion of mobile phone data and the possibilities for analysing it. Brands, marketers and strategists have invested considerable time and money in trying to decipher such data and develop insights from it. Mobile data provides us the opportunity and the ability to ask a plethora of interesting and sometimes highly intuitive questions. This article presents a series of studies that were undertaken in four different countries − the US, Germany, China and South Korea. These studies helped us access detailed granular atomic datasets that enabled us to comprehend and identify actionable insights about what brands and marketers can do with mobile data.
The Mobile Phenomenon
Increasingly, all evidence points to the fact that we are more and more engaged in one way or the other with our mobile phones. Whether it is a smartphone or a feature phone, the mobile device is very personal, with almost constant accessibility, and it is a veritable storehouse of data. As a result, the device is constantly storing data on what we might be doing at a given point in time, for instance, walking into a restaurant and checking in, using a mobile application (“app”) or writing a review of the restaurant and sharing updates with friends. The smartphone may be a device for customer engagement, but it has had a visible impact on various other parts of the digital media industry and on a fragmented mobile ecosystem made up of several different institutions.
Mobile traffi c accounts for more than 50% of all Web traffi c in a number of countries. India, the world’s second most-populous country, sees 48.24% of its mobile traffi c coming from mobile phones (Information Week 2012). With the rapid growth in mobile content generation, the chances of our lives being affected by mobile apps in one way or another will increase. Furthermore, the practice of “showrooming” − where consumers examine and compare merchandise in a traditional brick and mortar retail store and then shop for the product online at a lower price − is gaining prominence in mobile commerce as well.
Consumer Travel Patterns and Mobile Purchases
In an effort to gauge consumer behaviour across devices and platforms, our projects looked at data collected from different countries. Although there are differences in the manner in which data is generated and stored across these countries, at a fundamental level, the underlying factors affecting consumer behaviour remain the same
A fascinating aspect about the data we gathered is that it records interactions on a second-to-second basis. In other words, for the sample of people under observation, we can capture every single interaction through the smartphone, including locational movements and interactions with other people. My co-author, Sangpil Han, and I were interested in examining what insights on users’ content consumption and creation behaviour could be derived from consumer data on the mobile Internet (that is on mobile optimised sites of brands). For instance, in Seoul, South Korea, consumer data from two of the leading telecommunications operators, KT Corporation and SK Telecom, yielded a great deal of information about consumption and mobile content creation. By collating information on mobile content generation on Youtube, Facebook or Twitter, content consumed via the Maeil Business Newspaper mobile app and user location data, we can gain a tremendous amount of insight into a consumer’s behaviour.
To better comprehend the implications of the data observed, imagine a hypothetical consumer X in Hyderabad, who on a regular day commutes between her residence in Begumpet and her office in Jubilee Hills. Her routine includes a client visit to Banjara Hills. On certain days, her routine includes additional client visits to Gachibowli, followed by Madhapur, and additional visits unrelated to her work. By studying the mobile usage and daily travelling habits of a large panel of consumers like X above, we examined how brands could reach a potential customer at the right time, at the right place and on the right device. Using this data, we tested our hypothesis that mobile advertising was more effective when a consumer deviated from the standard path of travel than when she adhered to it. In other words, the consumer is more likely to respond to mobile advertising and make a purchase when she has to make additional visits resulting in deviations in standard travelling patterns than on a regular day when she sticks to her usual travel patterns. And consistent with this hypothesis, we found that variance in the user’s travel patterns is a much stronger predictor of mobile Internet usage than the mean (Ghose and Han 2011).
Impact of Geographic Distance
We also studied the impact of the distance between the consumer and a store on the consumer’s response to mobile coupons. While redemption rates increased as the customer’s proximity to a store increased, this occurred in a non-linear manner. It is not always the case that the closer a customer is to a store, the more likely she is to respond to a mobile coupon, all else being equal.
In the case of the mobile, its limited screen size is an additional factor. An iPhone is about two inches smaller than an Android device and both are more than a few inches smaller than a personal computer (PC). This significantly reduces the number of offers one can view on the screen. Thus, a customised or targeted advertisement becomes more critical on a mobile phone. Its position on the mobile screen and the distance between a customer and the nearest store advertised produces an interesting set of three-way interactions for marketers to exploit. For example, if the Barista coffee shop outlet you regularly pass by is able to figure out when you are at an optimal distance from it, it can send you a targeted coupon of the right value on your phone. Many retailers will thus jostle for your attention via their mobile coupons but there is only limited space on the mobile screen. The topmost slots on the mobile screen will therefore have greater value for retailers and be priced at a premium.
Consumers were divided into multiple groups, where one group was able to access information related only to distance, another only to price and a third to both distance and price. The order of the sale was randomised to see how the rank of the offer on the screen incentivised consumers. The products on sale were all high frequency retail products such as coffee, sandwiches, food, books and movie tickets. The data revealed that, on average, if the discount was increased by 10%, it was the same as reducing the distance between the consumer and the store by 131 metres.
To further examine these interactions, my coauthors (Dominik Molitor, Martin Spann and Philipp Reichhart) and I worked closely with a set of German fi rms and consumers spread across multiple cities. As part of the experiment, the retailers had to advertise through a particular mobile app, which would also be made available to all the consumers on their smartphones. The app would provide details on the most recent discounts from the retail stores and the consumers’ distance from the various stores based on their exact position in real time.
The consumers were divided into multiple groups, where one group was able to access information related only to distance, another only to price and a third to both distance and price. The order of the sale was randomised to see how the rank of the offer on the screen incentivised consumers. The products on sale were all high frequency retail products such as coffee, sandwiches, food, books and movie tickets. The data revealed that, on average, if the discount was increased by 10%, it was the same as reducing the distance between the consumer and the store by 131 metres. In other words, if the consumer receives a coupon when she is 131 metres away from the nearest Starbucks outlet, then she is as likely to redeem that coupon as she is to avail of a 10% discount on her coffee. Similarly, every one unit additional improvement in rank of appearance on screen produces the same effect as a price reduction of about 12%. Our study (Molitor et al., 2014) clearly indicates the remarkable possibilities of reaching out to the mobile shopping segment while experimenting with distance, face value of the coupon and the rank of the mobile offer on the screen.
Impact of Geo-fencing
The above research targets advertisements or coupons based on distance but not on time. Say, for instance, that a person is walking past a Starbucks outlet at 2 P.M. His friend is walking past a pub at the same time. Both retailers send discount offers: Buy one coffee and get one free at 2 P.M., or buy one beer and get one beer free at 2 P.M. Which offer is more likely to be redeemed? In all probability, it will be the Starbucks coffee offer. However, if the time of the offers is switched from 2 P.M. to 8 P.M., the consumer is probably more likely to go to the pub and redeem the beer offer. Mobile targeting should take context into account to increase its effectiveness. This means that combining location with time is a much more powerful strategy than using either one of them individually.
Narrowing down or defi ning the periphery within which the consumer is most likely to be at a particular time is known as geo-fencing. The idea behind geofencing is to combine location-based and time-based targeting. Another recent academic study on this topic was conducted by Luo et al. (2014) on tickets to a movie theatre advertised via a mobile app in Shanghai, China. In Shanghai, people often buy movie tickets through a movie app on their smartphones. The experiment involved targeting people with discount offers using the two dimensions of time and location.
The researchers would send targeted offers to consumers on three occasions: the day the movie was going to be screened, one day prior and two days prior. The other targeting dimension was the consumer’s distance from the movie theatre: 200 feet or so (“near”), 200 feet to 500 feet (“medium”) and up to a kilometre (“far”). One would expect that a discount offered on the day of the movie screening when the consumer was within 200 feet of the theatre would be more likely to be redeemed. However, they found, interestingly, that one-day prior offers actually had the highest rate of redemption. Interestingly, for those who get these discount coupons up to a kilometre away from where they are in real time, instead of the same day offers, the one-day offer yield the highest redemptions. Why exactly is this happening? Does it have something to do with how people plan?
Offers are costly, so businesses need to make them as targeted as possible. The idea is to narrow down the scope and the geo-fencing perimeter beyond which they are unlikely to reach a consumer. I am working with several different companies to execute additional experiments that would help us answer questions like the ones above.
App demand increases with the in-app purchase option wherein a user can complete transactions within the app. On the contrary, app demand decreases with the inapp advertisement option where consumers are shown advertisements while they are engaging with the app. The direct effect on app revenue from the inclusion of an in-app purchase and in-app advertisement option is equivalent to offering a 28% price discount and increasing price by 8%, respectively.
Mobile apps may appear to be a trivial part of the digital media ecosystem, but they force us to think about deep and strategic questions. For example, how would you price an app? Two-thirds of the apps on iTunes are free and roughly three-fourths of the apps on Android are free. If you are an app developer, one way to go is to charge for the app and make it advertisement free since a lot of people are annoyed by advertisements. The other option is to make the app free, but monetise it by showing advertisements. In 2013, the global mobile app market was estimated at over US$50 billion and is expected to grow to US$150 billion in the next two years (Ghose and Han 2014).
In a recently published research project (Ghose and Han 2014), we used data from both Apple and Android platforms and estimated the optimal price of an app that would maximise the developer’s revenues. How should you price an app so that you make the same revenues as a developer whether you are giving it for free and then monetising it through advertisements or charging a price for the app and promising not to show any advertisements? We built a structural econometric model to quantify the vibrant platform competition between mobile (smartphone and tablet) apps on the Apple iOS and Google Android platforms. We found that app demand increases with the in-app purchase option wherein a user can complete transactions within the app. On the contrary, app demand decreases with the in-app advertisement option where consumers are shown advertisements while they are engaging with the app. The direct effect on app revenue from the inclusion of an in-app purchase and in-app advertisement option is equivalent to offering a 28% price discount and increasing price by 8%, respectively. We also fi nd that a price discount strategy results in a greater increase in app demand in Google Play compared to Apple App Store. Using the estimated demand function, we fi nd that mobile apps have enhanced consumer surplus by approximately US$33.6 billion annually in the US, and discuss various implications for mobile marketing analytics, app pricing and app design strategies.
Interestingly, app usage is challenging television usage: average app usage across the world is about 128 minutes and average television consumption is about 164 minutes. For advertisers interested in allocating their budgets across traditional and digital media, this is a useful statistic to keep in mind as they devise their future digital marketing strategies.
Future Mobile Marketing Experiments
There are new opportunities globally for exploring the tremendous impact of mobile technology. The Shanghai government has decided to start introducing Wi-Fi on buses, both public and private. This is a marketer’s dream: once we know the bus route and where potential customers get on and off, we can incentivise them with the right kind of advertisements based on that knowledge. We can customise the advertisements based on the route of the bus. For example, if there is a particular shopping mall on the bus route, we can send targeted advertisements based on which stores are in that shopping mall. This large scale availability of internet access on public transportation is going to be a key feature of China’s “smart cities.” The technology infrastructure required for this is not especially complex and we have seen early implementation of the core idea in Canada. My prediction is that we will soon see it in countries like India and the US as well.