How Brands Can Make Smarter Decisions in Mobile Marketing Strategies for Improved Media-Mix Effectiveness and Questions for Future Research

Home , Advertising research, Brand

How Brands Can Make Smarter Decisions in Mobile Marketing Strategies for Improved Media-Mix Effectiveness And Questions for Future Research

Vassilis Bakopoulos Editors’ Note: Mobile Marketing In 2014, the Mobile Marketing Association launched a research initiative to help individual brands Association improve the efficacy of their mobile-marketing efforts. Each case study addresses marketers’ core questions [email protected] in a unique way: What share of their overall advertising spend should be allocated to mobile marketing? How should they use mobile formats and targeting methods more efficiently to maximize the performance John Baronello of their media investments? Although the authors acknowledge that their findings “might not provide Allstate Insurance definitive answers of long-term effect for all marketers,” they do offer insight into how mobile marketing John.Baronello@ can be optimized on a case-by-case basis. In 2017, campaign case studies with Allstate Insurance and a allstate.com major U.S. fast-food, or quick service restaurant (QSR), chain were the latest additions to a body of work with AT&T, the Coca-Cola Company, MasterCard, Walmart, and Unilever. The most striking results Rex Briggs came from the QSR study, which estimated an optimal allocation to mobile for that campaign at 33 percent Marketing Evolution of the total media mix—the highest allocation ever recommended in this research program. In the pages rex.briggs@ that follow, the authors describe their methods and findings, and propose best practices and questions for marketingevolution.com future research.

INTRODUCTION do not have a formal mobile strategy for their Since the arrival of the smartphone in 2008, con- brand, and only one out of three say they are ready sumers have embraced mobile technology faster for mobile adoption.5 Marketers, moreover, are than any previous technology.1 Smartphones unclear about the degree to which mobile drives have become the most important device to access revenue and profitability6 and its impact within the Internet.2 Consumers increasingly use these the context of their overall advertising mix. Under- mobile devices to make purchases; mobile usage standing the value of mobile is part of a much accounts for at least half the traffic and one-third bigger challenge of accurate attribution across of the revenue of e-commerce.3 Mobile’s share of all marketing channels—the process of assigning total advertising spending has increased rapidly value to a set of events or touch points that contrib- and is projected to reach 36 percent by 2020, sur- ute in some manner to a desired outcome. passing television’s share of spending, according Traditional top-down, aggregated approaches, to eMarketer.4 such as marketing-mix modeling, have addressed The rapid pace of change has left many marketers the topic of budget allocation for decades. unprepared for the new environment; two-thirds About 80 percent of marketers currently use this approach, according to Mobile Marketing Associa- 1 Benedict Evans for Andreesen Horowitz. (2016, March 29). “Presenta- 7 tion: Mobile is eating the world.” Retrieved May 3, 2017, from http://ben- tion (MMA) data. More recently, some marketers evans.com/benedictevans/2016/3/29/presentation-mobile-ate-the-world. 2 Office of Communications, United Kingdom. (2016, February). 5 WARC. (2017, May 17). “State of Industry: Mobile Marketing in “Ofcom Nations & Regions Tracker: Main set.” Retrieved Octo- North America.” Retrieved May 18, 2017, from http://bit.ly/2rx3hsY. ber 4, 2017, from https://www.ofcom.org.uk/__data/assets/pdf_ 6 CMO Survey. (2016, August). “CMO survey report: Highlights file/0030/68358/ofcom_technology_tracker_h1_2016.pdf. and Insights.” Retrieved May 3, 2017, from https://cmosurvey.org/wp- 3 Benedict Evans for Andreesen Horowitz. (2016, March 29). content/uploads/sites/11/2016/08/The_CMO_Survey-Highlights_and_ “Presentation.” Insights-Aug-2016.pdf 4 eMarketer. (2016, November 1). “US Ad Spending: eMarketer’s 7 Mobile Marketing Association. (2016, October). “Marketer Research Updated Estimates and Forecast for 2015–2020.” Retrieved May 3, Study: Marketing Productivity Assessment Attitudes.” Retrieved May 2017, from https://www.emarketer.com/Report/US-Ad-Spending- 3, 2017, from http://www.mmaglobal.com/files/documents/marketing_ eMarketers-Updated-Estimates-Forecast-20152020/2001915. productivity_assessment_marketer_study_july_2016.pdf.

DOI: 10.2501/JAR-2017-052 December 2017 JOURNAL OF ADVERTISING RESEARCH 447 How Brands Can Make Smarter Decisions in Mobile Marketing

have striven to measure media effective- • How should we plan and execute for of various mobile tactics. An extensive ness at a more granular level,8 favoring mobile in the context of an integrated review of published research (Grewal, Bart, multi-touch attribution approaches that campaign? Spann, and Zubcsek, 2016) provides an appear more relevant in a cross-platform • Should we use mobile for branding, overall framework to better understand the context. Although these multi-touch attri- direct response, or both? role of mobile in the mix and the key factors bution methods show some promise, they • Are there mobile tactics that are more that influence it, including the following: have their own limitations of validation, suitable for the upper funnel versus the data quality, transparency, and the abil- lower funnel? • specific tactics, such as mobile pro- ity to unify data across channels.9 Adding • Which specific key performance indica- motions (Andrews, Luo, Zheng, and to the attribution problem is the down- tors (KPIs) should we measure and opti- Ghose, 2015), mobile gaming (Hofacker, side of mobile innovation. The increas- mize against, and how? Manchanda, Ruyter, Donaldson, and ing number of formats, platforms, and Lurie, 2016), mobile-shopper marketing targeting methods has made it extremely Acknowledging these knowledge gaps, the (Shankar et al., 2016), mobile coupons difficult for marketers to decide where MMA in 2014 initiated an industry-wide in different contexts (Ghose, Han, and to focus. research program called Smart Cross- Park, 2013), and mobile-display adver- Mobile’s key opportunity for market- Marketing Effectiveness Research (SMoX), tisements (Bart, Stephen, and Sarvary, ers—data that can tell them when, where, a series of individual-brand case studies. 2014); and how to communicate with consumers In total, 11 have been conducted in four • the importance of context (environmen- in order to maximize the impact on their countries, including AT&T, MasterCard, tal, technological, or consumer), and decision making—has made budgeting the Coca-Cola Company (four studies), how location, time, and weather influ- that much more complex. The “when,” Walmart (two studies), and Unilever. Stud- ence consumers’ reactions to mobile “where,” and “how” represent different ies with Allstate Insurance (both auto and advertising (Molitor, Reichhart, and dimensions of advertising delivery, and home) and a major U.S. fast-food or quick Spann, 2014). Research that preexisted each comes with a price (for media, data, service restaurant (QSR)10 followed in 2017. the mobile “revolution” has examined production, technology, etc.) that often is Ford Motor Co. and MillerCoors studies the impact of other variables, such as difficult to justify. Marketers thus ask: are in progress for 2018, and future work weather, on consumer behavior (Hirsh- is earmarked with a major U.S. bank and a leifer and Shumway, 2003); • What share of our overall advertising leading fashion retailer. • how mobile-display advertisements spend should be allocated to mobile Each study measures the effectiveness of work in different industries (high ver- marketing? a real cross-marketing campaign against its sus low involvement), yet with conflict- • How should we use these formats and own marketing goals and media approach. ing results (Bart et al., 2014; Shankar and targeting methods more efficiently to The current article focuses on results from Balasubramanian, 2009); maximize the performance of our media the Allstate and QSR studies. References to • the impact of mobile on conversion investments? some of the earlier SMoX studies provide goals, mainly in the context of promo- context and comparison. tional campaigns (Grewal et al., 2016). Questions about measurement, attribution, and mobile fragmentation might have dif- BACKGROUND There also is some evidence about the role ferent answers, depending on the type of What We Know that mobile can play as part of the overall campaign or even the industry and prod- Mobile still is considered an emerging advertising mix, and the multiple benefits uct type: channel within a growing body of research. of advertising across multiple platforms Researchers have explored, among other (with or without mobile) versus a single

8 Coalition of Innovative Media Measurement. (2017, Feb- themes, how this channel influences the platform. A recap of such benefits (Neijens ruary). “Current Practices in Attribution and ROI Analy- purchase-decision processes (Shankar and Voorveld, 2015) includes the ability to sis” (white paper). Retrieved May 3, 2017, from http:// cimm-us.org/wp-content/uploads/2012/07/CIMM-4As- and Balasubramanian, 2009), and there is Whitepaper_Current-Practices-in-Attribution-and-ROI- increasing evidence about the effectiveness • increase reach (e.g., Briggs, Krishnan, Analysis_February-2017.pdf and Borin, 2005; Enoch and Johnson, 9 Mobile Marketing Association. (2016, October). “Mar- keter Research Study.” 10 The QSR company requested its name not be disclosed. 2010; Fulgoni and Lipsman, 2014; Taylor,

448 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

Source: Mobile Marketing Association, SmoX Study

Figure 1 Overview of Campaigns and Media Tested in the Smart Cross-Marketing Effectiveness Research Program (SMoX)

Kennedy, McDonald, Larguinat, et al., emphasized the role of mobile (Snyder and return on investment (ROI), with mobile 2013); Garcia-Garcia, 2016): video advertising being more effective • take advantage of unique strengths of in driving ROI than desktop video (with individual media (e.g., Dijkstra, Buijtels, • Mobile particularly is effective for some exceptions, e.g., financial services) and van Raaij, 2005; Okazaki and Hirose, established brands, which consumers and mobile video delivering higher ROI 2009; Tsao and Sibley, 2004); have less need to research or validate. compared with mobile banners. • facilitate information encoding in a more This aligns with previous work (Steele, • Mobile banners, although less effective complex way (Laroche, Kiani, and Econ- Jacobs, Siefert, Rule, et al., 2013) sug- than video, can benefit when placed in omakis, 2013; Stammerjohan, Wood, gesting that online environments are contextually relevant environments (e.g., Chang, and Thorson, 2005; Tavassoli, less able to invoke nonconscious emo- related magazine or newspaper article). 1998; Vandeberg, Murre, Voorveld, and tional connections, which are important Smit, 2015; Voorveld, Neijens, and Smit, components of media-delivered brand What We Don’t Know 2011; Voorveld and Valkenburg, 2015); equity. There has been far less research answering • reduce wearout (e.g., Navarro-Bailon, • Advertising in multiple platforms is the question of optimal media allocation 2012; Stammerjohan et al., 2005); more effective than advertising on a derived from analysis at the individual • create synergy, in terms of recall due to single platform. These findings specifi- user level. When marketers try to build exposure to multiple media (e.g., Chang cally emphasize the impact of the order a zero-based budget, what should they and Thorson, 2004; Dijkstra, 2002; Edell of exposure, with stronger results when do with specific mobile tactics in the con- and Keller, 1989; Voorveld et al., 2011). television came before mobile. text of the overall media mix to optimize • Different types of digital advertising their business results? What percentage Further evidence about the benefits of (i.e., desktop, mobile) and formats (i.e., of their total advertising budget should advertising across multiple platforms has banner, video) deliver different levels of go into mobile, and which advertising

June 2017 JOURNAL OF ADVERTISING RESEARCH 449 How Brands Can Make Smarter Decisions in Mobile Marketing

formats, targeting methods, and other tac- • The Coca-Cola Company (2015–16): Findings from previous tests with tics should they use? Marketers need more Mobile video and social advertising are Coca-Cola and Walmart demonstrated empirical evidence to address these ques- very efficient drivers of purchase intent, that richer advertising experiences, such tions for their own campaigns, but attribu- a finding validated in multiple studies as mobile audio and video, usually more tion across marketing channels remains a (four video-related, two social media- than justified their price premium in rela- big challenge,11 and its solution becomes related) conducted in North America, tion to mobile display when it comes to even more difficult with the increased pro- Brazil, the United Kingdom, and China. shifting perceptions and driving purchase liferation of advertising channels. • Unilever Magnum Ice Cream (2016): intent. Yet, in some instances, simple ban- The MMA’s case-study work with Mobile social video advertising can be ners were more efficient in terms of driving individual brands has filled some of this a strong driver of ROI. In this case, it awareness (new awareness or top-of-mind knowledge gap. One could argue that was much more efficient than video in awareness, in the sense of reminding con- there are limitations to making broad other screens, including desktop and sumers of a brand message). empirical statements, given that these television. Findings in 2017 from two companies studies span three years and a number of in very different categories—Allstate auto product categories. The state of the market Research Questions and home insurance (high-involvement, and how marketers approach mobile also The researchers posed three key questions: infrequent purchase) and the fast-food continue to evolve (See Figure 1), changing (QSR) chain (lower-involvement, frequent the nature of the questions and the context RQ1 Path to purchase: How can impulse purchase)—offered additional of some of these tests. a brand most effectively use insight into the path-to-purchase ques- The previous studies conducted as part mobile advertising tactics to tion. The goal was to provide an even more of this series reached the following conclu- engage consumers across the granular read into how these formats affect sions (with limitations; for deeper discus- funnel? attitudinal (upper funnel) or actual behav- sion on AT&T, Walmart, and Unilever, see ioral (lower funnel) metrics—namely sales pages 455–458): RQ2 Size, depth, and repetition: How and, in the QSR’s case, foot traffic. can a brand most effectively • AT&T (2014): Mobile display advertising communicate a message on a Size, depth, and repetition. Conven- can be a very efficient driver of brand mobile platform? tional wisdom suggests that attention awareness for a new product launch. span in mobile advertising is shorter, yet The findings justified a 16 percent allo- RQ3 Targeting: What is the value of the screen also is smaller. The size, depth, cation to mobile in the total mix of that different data signals—digital or and repetition question aimed to reveal the campaign. physical—for improving target- implications for the advertising units that • MasterCard (2015): A combination ing in mobile? marketers use in mobile video and display of mobile display, mobile video, and while trying to communicate their adver- mobile social advertising can be very Path to purchase. Marketers have two tising message. Should they design their efficient in terms of driving brand image, goals: strengthen consideration for the mobile video to be shorter and their mobile even among an older demographic of brand by reinforcing image perceptions display to be more “visible”? What does nesters and empty nesters. (top of the purchase funnel), and drive that mean in terms of frequency of expo- • Walmart (2015): Mobile advertising can sales (lower funnel). There are different sure and consumer experience? be a very efficient driver of sales; in this options provided by mobile in order to The AT&T study provided early evi- study, it was twice as efficient as the support these goals, often with trade-offs. dence that larger banners were more effi- average of that campaign. Proximity Simple, lower-cost mobile banners are cient for driving awareness, compared location targeting, when matched with limited in their ability to communicate a with the smaller “pencil” mobile ban- expandable mobile display units, also compelling message, yet richer experi- ners. Findings from AT&T, Coca-Cola, improves the impact of advertising in ences, such as mobile video, are priced at and Walmart also illustrated that banners terms of driving foot traffic. a premium. How can marketers use these tended to have a “linear” relationship with tactics to drive different KPIs? Should the frequency, which meant that each addi- 11 Mobile Marketing Association. (2016, October). “Mar- keter Research Study.” tactics vary by product category? tional exposure continued to build impact,

450 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

even at a higher frequency, especially in • measure mobile and other media down optimization—including specific target- relation to awareness. For the Allstate and to the individual tactic and message; ing using location, web behaviors, time QSRs studies, the goal was to measure • analyze the interactions among differ- of day, weather, and other variables— and assess more advertising formats and ent media of very different individual the analysis needed to connect directly subformats, especially richer formats (e.g., brands; to person-level media-planning and video lengths of 15 seconds versus 30 sec- • be applied across a wide range of prod- buying systems. onds), and to gain perspective from differ- uct categories, each of them bringing its • Specifically for Allstate, given that the ent product categories. own complexity and data availability. purchase cycle for insurance can be long and the decision considered, the Targeting. Mobile advertising exempli- To have a meaningful degree of certainty approach needed to include measure- fies what is possible in targeting, given about the causality, the method needed to ment of advertising’s influence on con- that people carry their phones with them sideration, as well as on behaviors. most of the time. With that in mind, in • link a person’s activities across devices • Because the product (insurance) can be addition to considering traditional demo- (e.g., a consumer who is exposed to a sold through the physical location (in graphic, income, or other consumer char- message on a mobile device, who sub- this case, a local agent), the analysis acteristics, marketers increasingly think sequently calls an insurance agent, and needed to incorporate location analysis in terms of “need states,” occasions, and uses a desktop computer to request a and location-based media optimization. “key moments” in the customer journey quote); (Moran, Muzellec, and Nolan, 2014). Iden- • incorporate the design of experiments to The Limitations of Marketing-Mix tifying those moments and understanding control and isolate the impact of expo- Modeling the “customer journey” is an important sure to mobile media. The MMA concluded that traditional first step, yet marketers need evidence marketing-mix modeling approaches about these moments’ impact on ROI. The Allstate Insurance study focused could not address all the above goals. As By investigating targeting at individual on understanding the path to purchase, Forrester noted in its Fall 2016 Measure- brands, the current researchers could which required measuring both attitudinal ment & Optimization Wave: “Market- assess a variety of different “data signals” survey-based metrics (consideration) and ing mix models aren’t fast or detailed used in mobile. The Walmart study in behavioral metrics (sales). enough…These models can’t provide 2015 shed light on the value of location There were other methodological con- campaign performance detail that drives targeting, and in 2016 Unilever’s Magnum siderations across the various brands: media buying: the networks, programs, study assessed the impact of weather tar- publishers, and sites that will optimize geting for an ice cream brand. In 2017, the • Because of the nature of mobile, the return on ad spend.” Allstate and QSR studies have built on method needed to rely on something Moreover, as noted in a 2013 Coun- that knowledge covering a combination of more permanent than cookies, because cil on Research Excellence report, given data signals —some of them digital (i.e., mobile users often delete cookies,12 and, that mobile is still a relatively small part online search behavior, genre of content therefore, the connection between expo- of the mix, marketing-mix modeling in which the advertisement appears), and sure and sales might be lost. often suffers from what is referred to some physical (i.e., location history and • To measure mobile in-app advertis- as the small-reach problem.13 If, on the movement patterns)—and their impact ing (which comprises 90 percent of the one hand, only a small fraction of peo- on marketing KPIs in relation to their mobile advertisements and does not ple is reached and the impressions are premium. accept cookies), it was important to con- fairly evenly spread across the country, a nect a device ID to a person. marketing-mix model unlikely will pick CASE-STUDY METHODOLOGY • Because the ultimate action from the up the impact purely, because the impact The MMA research team evaluated a wide measurement would be a detailed media occurs among a small percentage of the range of vendors in search of a method 12 Davis, W. (2013, January 23). “Study: 44% of Adults 13 Sequent Partners. (2013, July). “Current State of Mar- that would best address the goals of the Opt Out of Targeted Ads, 66% Delete Cookies.” From keting Mix Models: A Report for the Council for Research MediaPost website: https://www.mediapost.com/publica- Excellence.” Retrieved May 30, 2017, from http://sequent- current case-study research program. In tions/article/191809/study-44of-adults-opt-out-of-targeted- partners.com/wp-content/uploads/2016/11/CRE-Modeling- particular, the method needed to ads-66-d.html. White-Paper-CRE-Branded.pdf.

June 2017 JOURNAL OF ADVERTISING RESEARCH 451 How Brands Can Make Smarter Decisions in Mobile Marketing

Table 1 Test Requirements and Methods’ Capabilities Comparison Requirement Media-Mix Modeling SMoX/Marketing Evolution Method Measure detail down to the individual tactic and message No Yes Measure at a speed at which insights could be acted on while the No Yes campaign was still live Ability to link behaviors of individuals across devices (e.g., exposure on No Yes mobile, sale via call center) Ability to overcome the small-reach problem and use control–exposed No Yes measurement for clear read on causality or validation of models Ability to measure the effect of advertising on consideration and other Possible, but not with Yes perceptions over a long purchase cycle the same precision as person-level analysis Ability to measure the relationship between brand perceptions and No Yes purchase behavior Method can track person exposures and behaviors over time, without N/A Yes losing data as a result of cookie deletion Use of specific physical location data (spatiotemporal analysis) No Yes Output of analysis directly connected to person-level targeting, digital No Yes buying segments, and other media implementation

Note: SMoX = Smart Cross-Marketing Effectiveness Research. overall population. On the other hand, use of design of experiments with logistic advertising servers to target people with when it is possible to run an experiment regression and elastic net regularization. known identities to run experiments. in which those exposed and those given Specifically, the approach combines Working with personally identifiable infor- a control can be known, one can put a logistic regression statistical analysis—to mation (PII) safe-harbor identity-matching magnifying glass on the small percent- isolate the contribution of media expo- firms (e.g., LiveRamp and Neustar), Mar- age of people and see the impact that was sure as it applies to driving sales (or other keting Evolution and MMA enlisted missed with a mix model. Although any KPIs)—with controls for behavioral and mobile-advertising platforms to integrate analytic technique can be adapted to deal other demographic characteristics in the identity data in a PII-protected way. This with special cases, marketing-mix mode- sample. The model’s elastic net is applied allowed the marketer to measure tactics ling was not an obvious fit for this study’s to identify the advertising messages and for which the reach would have been too requirements. the people influenced by those messages. small to measure with other methodolo- This information is used to optimize mes- gies. The approach accordingly overcame An Analytics-Driven Approach: sage rotation, targeting, and, ultimately, the small-reach problem that would have People Data Focus media mix. Experiments are used to vali- plagued mix-modeling measurement with The MMA researchers selected Marketing date the model and to measure small-reach targeted exposed–control media delivery. Evolution’s ROI Brain™ analytic platform activities. This field experiment approach had the to conduct the omnichannel attribution combined benefits of a laboratory experi- and optimization analysis for each case Addressing Small Reach ment’s test–control design and of real- study. The person-level analysis allowed Although the method included many of world interactions with other media and measurement of every message in every the capabilities the researchers desired, marketing activity. medium and included location analysis, the small-reach problem was a concern The role of individual-level data and the brand-perception measurement, and sales with respect to measuring certain tactics, use of PII safe harbors simplifies the abil- measurement. An expanded measurement especially in mobile. The vendor thus ity to merge data from different sources. model, furthermore, integrated continuous revived the technique of connecting with This approach, moreover, can lead to the

452 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

creation of detailed datasets that include media-exposure inputs along with behavioral and attitudinal outcomes, including sales data or store visitation. Without this approach and the ability to integrate mobile media sellers, link device IDs to people, and target them with specific media, the researchers believe they could not have measured mobile in-app advertis- Note: An MMA-derived index measures the relative efficiency of each advertising ing accurately. medium as it drives consideration or sales per advertising dollar spent. Consideration By using this analytics-driven, individual- measured respondents’ answers to a survey question, while sales were measured by level data approach, the researchers could direct matching to Allstate data. Source: SMoX Allstate study. overcome some of the limitations of traditional marketing-mix model methods. Figure 2 Allstate: Relative Efficiency of Mobile Formats Additional adjustments were made to Versus Campaign Average (Index Measures) address some of the unique challenges of mobile (e.g., cookies) and to measure tactics that typically have too small a reach to be media approaches of marketing to con- number of people “reached” by each measured in the field with other analytic sumers in a high-involvement, infrequent- medium (as measured by cost per 1,000 techniques (See Table 1). purchase category—in this case, home impressions on a web page [CPM]) but and auto insurance. Given the high brand in relation to how these media influenced RESULTS awareness of most established brands in the actual KPIs of each campaign. In that In this section, the authors address their the industry, the two main communication sense, the impact of advertising that was research questions by comparing results opportunities for Allstate were attributed to each medium was divided by from case studies of three different adver- the cost of these media. The authors meas- tisers (references to the earlier AT&T and • to reinforce brand image and build ured impact by comparing the lift between Walmart studies add further context): consideration among broader groups the exposed group and the control group. of consumers, who might not be in the For the purposes of confidentiality, the • Allstate, for which the product (insur- market at this very moment but would authors agreed not to reveal actual cost and ance) requires high-involvement pur- be soon in the future; sales information for each brand. Instead, chase decisions; • to identify consumers at the right they calculated an efficiency index for • QSR—fast food—a frequent, low- moment—when they are in the market each study, to illustrate how different tac- involvement purchase, for which driv- for auto insurance—and use media to tics performed without releasing sensitive ing consumers to physical locations is trigger immediate response and drive information (See Figure 2). The campaign key; acquisition. average includes all the media—digital • Unilever’s Magnum Ice Cream, a low- and traditional—in relation to the specific involvement impulse brand. Allstate’s media approach combined high- KPI (brand consideration versus sales). impact messaging to drive consideration The findings of the study validate the RQ1 Path to purchase: How can (at the top of the path-to-purchase funnel) overall media approach used for Allstate a brand most effectively use and more direct, response-focused mes- (combining high-impact messaging with mobile-platform options to saging for acquisition and sales (at the direct response) and suggest the following engage consumers across the lower part of the funnel). Within that con- for mobile: funnel? text, Allstate wanted to understand better how to utilize its existing mobile-platform • At the top of the funnel (influencing Allstate Study Findings options. consumers’ perceptions and driving The Allstate study methods were repre- The current authors defined “media consideration for brand), the richer sentative of the specific conditions and efficiency” not simply in relation to the mobile formats (audio and video)

June 2017 JOURNAL OF ADVERTISING RESEARCH 453 How Brands Can Make Smarter Decisions in Mobile Marketing

video, mobile video, and television). And Eﬃciency Index on although mobile overall emerged as the Store Visitaon most efficient driver of foot traffic compared to other media (digital and tradi- Targeted Mobile Display 357 tional), not all mobile formats performed Social Mobile 127 the same. (See Figure 3). Mobile Video 62 Among the QSR study findings: Total Media (digital and traditional) 100 • Targeted mobile display placements Note: Index measures the relative efficiency of each media’s ability to drive foot traffic emerged as the most efficient driver per for QSR marketer per advertising dollar spent. Foot traffic was passively tracked and advertising dollar spent for both Allstate matched to ad exposures. Source: SMoX QSR study. and the QSR. The authors propose that recency and context of message were Figure 3 Fast-Food (QSR): Relative Efficiency of Media more important drivers of conversion, Compared with Campaign Average which is why targeted mobile display and radio ranked at the top of MMA’s efficiency index. resulted in greater ROI (See Figure 2). In other words, a simple, direct-response • Richer video experiences (in mobile or When the authors accounted for the banner message, delivered to those con- television) were not as efficient, given number of people affected in “brand sumers who already were in the market that the message was easy to commu- consideration” by each type of mobile for auto insurance, was a more efficient nicate and there was preexisting aware- format and factored in the cost of buy- approach for driving short-term sales. This ness about the promotion. ing the specific media, they found the was the case despite the proven effective- • Using location and other data signals following: ness of mobile video (Snyder and Garcia- allowed the QSR company to target key Mobile audio was about 30 percent Garcia, 2016) to shift perceptions and drive segments of consumers who much more more efficient than the campaign consideration for the brand. likely would respond positively to the average (which included all media, promotional offer, such as commuters, including television) for driving con- Fast-Food Chain (QSR) Findings who tend to travel the same route every sideration for Allstate. The results from the QSR research were day. Despite the additional cost of data Mobile video was 85 percent more directionally aligned with the Allstate to improve the targeting of these place- efficient than the campaign average, study—sharing a marketing goal for lower ments, advertisements that were served given its very high effectiveness and funnel conversion. But fast food, a low- to these segments were more than five lower cost compared with television. involvement category, is a very different times more cost effective at driving • At the bottom of the funnel (driving market from insurance, requiring differ- foot traffic than the campaign average. sales), targeted mobile banner units ent metrics to evaluate mobile marketing Similarly, targeting a segment of coupon emerged as more efficient—by 12 per- effectiveness (foot traffic versus sales for users (based on personal level data that cent, on average—compared with the insurance). The QSR tested a promotional showed these consumers had redeemed campaign average. In this case, the campaign for choosing a handful of menu coupons in the recent past), performed analysis took into account the total sales options at a discounted price. The specific almost as well—about five times higher that were attributed to each medium promotion had preexisting awareness from than the average. and the cost of buying the media. By a previous launch. The goal of this cam- contrast, paign was to use the same promotional The most important result came in deter- mobile video appeared less efficient platform in order to drive additional foot mining the optimal allocation for mobile compared with the campaign average traffic to various restaurant locations. in this QSRs campaign media mix. The (index of 76), and Similarly to the Allstate campaign, a researchers took into consideration the per- mobile audio was close to the average broad mix of advertising media was used formance of the various mobile tactics and (index of 98). (targeted mobile display, radio, desktop its contribution to the total, as it was derived

454 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

Allstate, AT&T, and Unilever: Size, Depth, and Repetition Findings Video emerged as a strong driver of consideration for Allstate, but not all video units and assets performed the same. Length and placement made a difference, but mobile video and mobile audio showed varied outcomes of driving con- Note: Index measures the relative efficiency of each media’s ability to drive consideration sideration, in relation to the length of the for Allstate per advertising dollar spent. Consideration was measured by a survey creative asset that was used (See Figure 4). question. Source: SMoX Allstate study. The authors proposed that because attention span is shorter for mobile adver- Figure 4 Allstate: Relative Efficiency of Video and Audio by Length tising, marketers could expect stronger results if they can articulate their product or brand story in a shorter time. A 15-second commercial was more efficient, particularly in the case of mobile video. The gap between 15 and 30 seconds was a lot larger when it came to mobile video. This possibly suggests that when consumers were watching video on their phone, they had a lower tolerance for longer commercial interruptions. For mobile audio advertising, the 15- to 30-second difference Note: Index measures the relative efficiency of each media’s ability to drive purchase was less significant, likely because a listen- intent for Unilever’s Magnum per advertising dollar spent. Purchase intent was ing activity does not require consumers to measured by a survey question. Source: SMoX Unilever study. hold the device in their hands. Size and scale mattered in different ways Figure 5 Unilever Magnum: Price and Effectiveness Comparison in the Allstate, AT&T, and Unilever campaigns. Although shorter appeared to be Of Large and Small Mobile Banners more effective for mobile video, bigger was better for mobile banners. The authors com- by the logistic regression analysis. They also because this specific “use case”—combi- pared two different-sized banners that were compared the cost of mobile to other media, nation of product category (low involve- tested in the Unilever Magnum Ice Cream and they ran various optimization scenarios ment, frequent purchase), campaign study (See Figure 5). The results showed in order to understand the optimal media message (simple, promotional, immediate) that the larger banner, by far, was more combinations that would maximize the per- and KPI (foot traffic)—uniquely played to effective and efficient—the most dramatic formance of the QSR campaign. mobile’s strengths. Nevertheless, part of difference observed among the “bigger is The researchers recommended for the the upside also came from the ability of the better” pattern found among the other stud- QSR campaign an estimated optimal company to experiment with various data ies in this research program. A possible rea- allocation of 33 percent of the media mix signals and test innovative tactics and data son for this is the type of product category to mobile—significantly higher than the combinations. (impulse) and the way the advertisement actual allocation that the company had real estate was utilized to create desire with used, (mid-20s percentage), and the high- RQ2: Size, depth, and repetition: How impactful imagery. By contrast, the differ- est recommended allocation among all can a brand most effectively ence in banner size effectiveness that was previous studies in this program. The communicate a message on a reported for telecoms (AT&T study) was authors proposed that this was largely mobile platform? much lower, although still compelling,

June 2017 JOURNAL OF ADVERTISING RESEARCH 455 How Brands Can Make Smarter Decisions in Mobile Marketing

advertising, because the results are not completely consistent across all of the brands. The preponderance of findings, however, suggests that optimal ROI is achieved with less frequency with mobile video and other highly noticeable advertising units. This might be explained by the engagement level with video consumed on a mobile screen, and it is consistent with empirical research on matters of frequency. Note: Index measures the relative efficiency of each media’s ability to drive An Advertising Research Foundation consideration for Allstate per advertising dollar spent. Consideration was measured by study in 2016 found that “while impres- a survey question. Source: SMoX Allstate study. sion levels delivered to consumers via digital can range in the hundreds per month, Figure 6 Allstate: Relative Efficiency of Targeting Approaches more than 40 impressions per month can actually have a negative impact on a In Mobile Video brand’s goals” (Snyder and Garcia-Garcia, 2016, p. 361).

RQ3 Targeting: What is the value of different data signals, digital or physical, for improving targeting in mobile?

Allstate, Unilever, and Walmart Targeting Strategies Allstate tested a variety of targeting approaches leveraging a wide spectrum Note: Index measures the relative efficiency of each media’s ability to drive sales of data signals, both digital and physical. for Allstate per advertising dollar spent. Sales were measured by direct matching to There were differences in the impact of Allstate data. Source: SMoX Allstate study. those approaches, depending on the KPI (consideration versus actual sales). Figure 7 Allstate: Relative Efficiency of Targeting Approaches The researchers compared the efficiency of behavioral targeting and contextual In Mobile Banners targeting approaches used to deliver mobile video to the right consumers and with the larger unit being more than twice • Larger size is better, especially for drive consideration for Allstate (See Fig- as effective as the smaller unit. image-driven categories, when it comes ure 6). The study demonstrated that both Findings from Allstate, Unilever, and to banners. approaches justified their incremental cost AT&T suggest that different rules apply and improved the performance of mobile when it comes to using mobile video and “If bigger generally is better,” the research- video. The authors specifically found the mobile banners to deliver a message. ers asked, “is more also better?” The following: impact of frequency was different for • Shorter clearly is better for video and is banners versus video and other, richer • Brands used contextual targeting to preferable, although not as much of an advertising units. More research is needed identify consumers who were brows- issue, for audio. into the optimal frequency and pacing of ing relevant content about the category

456 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

• By contrast, when historical location data were used to define an audience (of past shoppers), the impact on store visitation was lower for the expandable (larger) advertising unit, and there was zero incremental impact for the small, static banner (“pencil unit”).

Whereas Allstate and Walmart used loca- Note: Store visitation lift was measured using actual foot traffic data. Advertiser did tion data to define audiences (for Walmart, not approve disclosure of the exact lifts, only the directional findings illustrated above. to target consumers relative to their proximity to a store), Unilever focused on the weather conditions in a given location. For Figure 8 Walmart: Store-Visitation Lift Caused by Location- Unilever’s Magnum Ice Cream, mobile Targeting Approach and Advertisement-Unit Combination banners, even with no weather targeting, had emerged as an efficient driver of on their mobile devices. This approach for video advertisements, did not really brand awareness (75 percent more efficient improved the results of mobile advertis- move the needle on sales when it was compared with the campaign average; See ing by more than 90 percent. used with mobile banners. The authors Figure 9). There was no impact in terms of • Brands also used behavioral targeting, proposed that this approach was more driving short-term sales in the target group relying on first-, second-, and third-party suitable for upper-funnel metrics (e.g., of that campaign, which resulted in zero data about key behaviors that illustrate image, consideration), especially when ROI. When the same advertisements were that a person is in-market to buy home matched with a rich experience, such as served targeting locations where the out- or auto insurance. This approach fur- mobile video. door temperature was above 80°F, however, ther improved results of mobile video • Location targeting emerged as another the results changed, and a sales impact was advertising by more than three times, efficient driver of sales, next to retarget- measurable in the short term. The reason compared with when no behavioral tar- ing and behavioral targeting. In this case, for the drastic turnaround, the researchers geting was applied. it improved the results by two times. believe, was relevance. This is particularly important, because it The weather-targeting approach The authors also compared the efficiency means that marketers can utilize physi- ensured that the mobile banners matched of various targeting approaches that were cal data signals, such as location data, a relevant “need state” predicted by used to deliver mobile banners, with the in order to define elaborate audiences the high temperature. Combined, the ultimate goal of driving sales (See Figure and expand the reach of targeting “in- weather-targeting strategy and a message 7). Behavioral targeting emerged as the market” consumers beyond the behav- of indulgence led the mobile banners to most powerful tool to identify consum- ioral and retargeting approaches. become one of the most efficient drivers ers in the right moment when they are of short-term sales—about 50 percent in-market for auto and home insurance The findings added to knowledge (from more efficient in relation to the campaign and to drive acquisition. Retargeting (i.e., the previous case studies in the current average. Recall that the analytic platform targeting consumers who previously were program) about location-targeting meth- methodology employed in this research interested in Allstate) also delivered great ods for driving foot traffic to physical store program includes use of control groups. results, more than twice the return as when locations. The Walmart study (See Figure This enabled the researchers to assess it was not applied. 8) specifically suggested the following: that the same advertisement, when served The authors, moreover, found the without weather targeting, had not gener- following: • The strongest impact on foot traffic came ated a sales lift compared with the control. from proximity targeting when combined This suggests that it was the combination • Contextual targeting, although an effi- with expandable banner units, generating of the mobile advertising and weather tar- cient driver of consideration when used the strongest lifts in foot traffic. geting that generated sales.

June 2017 JOURNAL OF ADVERTISING RESEARCH 457 How Brands Can Make Smarter Decisions in Mobile Marketing

The interaction between need state and advertising is not a new marketing concept. Yet traditional media are limited in their ability to be present in places and times where these need states material- ize—for instance, the increased desire for ice cream when temperature goes above 80°, as in the Magnum Ice Cream exam- ple. In these situations, mobile advertising has a unique advantage because of the pervasiveness of mobile phones, which Note: Index measures the relative efficiency of each media’s ability to drive awareness allows marketers to act on their consumer or sales for Magnum, per advertising dollar spent. Awareness was measured by a insights and engage consumers in the survey question, while ROI takes into consideration the estimated profit due to media, moment. using actual sales lift analysis. Source: SMoX Unilever study.

BEST PRACTICES IN MOBILE MARKETING Figure 9 Unilever Magnum: Relative Efficiency of Mobile The findings from this body of research can serve as a guide to marketers who Banners in Relation to Campaign Average want to improve the effectiveness of their mobile-advertising investments. Although of the above levers “in-market” through take into account the impact of advertis- the rules for mobile-marketing efficiency programmatic advertising. Starting with a ing and how it drives their campaign KPIs can be applied in different ways depend- broader menu of possible ways to engage when considering cost. Most of the data- ing on the brand, product category, and with mobile-specific content, advertis- targeting approaches assessed in the cur- medium, the authors suggest the follow- ers can run massive experiments with rent case-study series more than justified ing best practices: dynamic creative optimization, measure their premium (cost) and significantly what works, and optimize accordingly improved the impact of mobile advertis- Produce Creative Content while the campaign is live.14 ing per dollar spent. This validates the Specifically for Mobile theory that marketers can find great value Mobile’s smaller screen offers the oppor- Target More Deeply in understanding the customer journey tunity for more purposeful engagement, Mobile creates a much richer dataset than and operationalizing planning insights to so repurposing creative content from other traditional media, allowing marketers target “moments of relevance” across the platforms misses the opportunity to fully to capture the context, intents, and need customer journey. In those instances, the leverage context and customize accord- states of individuals. The historic location results from applying such targeting not ingly. Previous research provided evidence of a consumer can reveal a great deal about only are significantly greater but also cre- about the benefit of unifying creative strat- his or her passion points (visited a golf ate a “multiplier” effect and drastically egy across different platforms but also cus- course, went skiing, attended a live music change the impact of a given creative asset. tomizing execution to the platform level performance), income (lives in an afflu- (Snyder and Garcia-Garcia, 2016). The cur- ent neighborhood), or even current need Adapt to Continuous Optimization rent research validates that marketers can state (find out today’s weather). Unifying Mobile targeting optimization may prompt maximize the impact of their advertising and activating these data potentially can marketers to ask, “Can we act fast enough when they align creative concept, format, increase the actual CPM paid by advertis- on the optimization insights?” Analyt- advertising unit, data, and delivery to com- ers when targeting broader demographic ics technology does the heavy lifting, yet municate a brand message that is relevant audiences. Advertisers, however, should the practice of changing in the middle of in the specific moment in the customer 14 Fondon, C. J. (2014, October 20). “Dynamic Crea- a campaign might be new to a marketer. journey. tive Optimization: What is it?” Retrieved May 26, Once a marketer begins to optimize mobile 2017, from ProgrammaticAdvertising.org’s web- Alternatively, marketers increasingly can while the campaign is live, he or she might site: http://programmaticadvertising.org/2014/10/20/ assess the impact of multiple combinations dynamic-creative-optimization-what-is-it/. want to optimize as many other media as

458 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

possible. And, as data-analysis and meas- typically requires a year or more of obser- the previous studies, these marketers will urement technology continue to evolve, vation. Each case study is a single cam- explore “unique-to-mobile opportunities,” the issue becomes marketers’ and agency paign, and, therefore, the benchmark including: orientation toward agility. This research, conversion applied is based on the ven- therefore, has implications for how a mar- dor’s database. The database of norms • the impact of “mobile-first” creative keter organizes teams and agencies around might not be applicable directly to any content and its impact on ROI. This continuous optimization, as opposed to an single brand and, as a result, might not will include the impact of “nonworking annual planning cycle. provide definitive answers of long-term cost”—in other words, the cost of pro- effect for all marketers. More emphasis is ducing the actual creative assets, which LIMITATIONS placed on the sales benefit achieved within often is left out of media-efficiency In the authors’ view, the biggest limita- one purchase cycle, and some might view analysis. tion to the current research program is the this as a limitation. • using mobile for integrated communi- match rate of identity data. With a current cations: how mobile can enhance televi- rate of 40 percent and growing, researchers FUTURE RESEARCH sion or out-of-home advertising and the should be careful to examine whether scal- More work is needed to better understand role of creative message and sequencing. ing up the 40 percent to 100 percent of the the increasing variety of mobile tactics at a This could build on knowledge from a population aligns with total sales. If so, the more granular level: 2016 Advertising Research Foundation identity data that can be matched can be study that found a “kicker effect” when viewed as representative, and a marketer • Factors such as auto-play versus opt-in, digital-advertising investment was can have more confidence in the findings. sound-on versus sound-off, and skip- added to television, resulting in an ROI If not, a researcher can consider weighting pable versus nonskippable advertising increase of about 60 percent (Snyder and but will need to understand whether cer- sometimes are intertwined with vari- Garcia-Garcia, 2016). tain groups are misrepresented and assess ables, such as video length and target- the impact of specific variables to achieve a ing method applied, creating an infinite Finally, there is an interest in further exam- projection that aligns with total sales. number of combinations. ining social media and other “walled gar- In the current case-study series, impact • Location targeting sometimes is viewed dens” that account for a large share of the was measured for both brand equity as “proximity targeting,” or geofencing, mobile-advertising spend. Some work in and sales behavior, but might it miss the and solely is associated with market- this case-study series (Unilever and the longer-term impact of advertising? The ers who want to drive foot traffic to a QSR) provides isolated cases of measur- method of the primary vendor selected physical location. There are promising ing some of these platforms, but further for this study series (Marketing Evolution) opportunities, however: using histori- investigation is needed. As the industry for quantifying the relationship between cal location patterns to define elaborate brand-equity measures and a person’s audiences around behaviors (e.g., people conversation about transparency and data purchase activity for months, and in some who commute using a specific route), sharing continues to evolve, marketers will cases years, has been applied to customers passion points (e.g., people who go to get better at predicting what is feasible and in the automotive, retail, and financial- music venues), or even income segments how advertising spend in these platforms services categories. Within each sale, the (e.g., affluent consumers who go to ski will be assessed in the years to come. method allows the researchers generally resorts). Some of these location-targeting to trace back to a set of beliefs about the tactics are assessed further in the context ACKNOWLEDGMENTS brand. When advertising enhances these of the QSR study. beliefs, sales follow. The researchers refer The Mobile Marketing Association thanks to these measures as leading indicators. More MMA-led studies will follow, the following companies that support The current research program meas- including Ford Motor Co. and Miller- its Smart Cross-Marketing Effectiveness ures brand beliefs as well as sales and Coors (results expected in 2018). Work Research (SMoX): Jun Group; PlaceIQ; can apply a formula of brand beliefs and with a major bank and leading fashion The Weather Company; xAd, Inc.; ESPN; their conversion to longer-term sales and retailer also is in progress. Beyond the Foursquare; Pandora; Turner; Ubimo; and profits. Arriving at this formula, however, more granular questions addressed in Verve.

June 2017 JOURNAL OF ADVERTISING RESEARCH 459 How Brands Can Make Smarter Decisions in Mobile Marketing

About the authors Briggs, R., R. Krishnan, and N. Borin. “Inte- Research Agenda.” Journal of Interactive Market- grated Multichannel Communication Strategies: ing 34 (2016, May): 3–14. Vassilis Bakopoulos is the head of insights and research Evaluating the Return on Marketing Objec- for the Mobile Marketing Association in New York tives—The Case of the 2004 Ford F-150 Launch.” Hirshleifer, D., and T. Shumway. “Good Day City. Previously, Bakopoulos worked for Digitas and Journal of Interactive Marketing 19, 3 (2005): 81–90. Sunshine: Stock Returns and the Weather.” Jour- for Kantar’s Added Value and Research International nal of Finance 58, 3 (2003): 1009–1032. divisions in New York and Europe, specializing in Chang, Y., and Thorson, E. “Television and Web

advertising effectiveness, customer segmentation, and Advertising Synergies.” Journal of Advertising 33, Hofacker, C., P. Manchanda, K. D. Ruyter, J. 2 (2004): 75–84. creative strategy. His research has been published in the Donaldson, and N. Lurie. “Gamification and Journal of Applied Marketing Analytics and the Journal of Mobile Marketing Effectiveness.” Journal of Dijkstra, M. “An Experimental Investigation Brand Strategy. Interactive Marketing 34 (2016, May): 25–36. of Synergy Effects in Multiple-Media Advertis- ing Campaigns” (doctoral dissertation). Tilburg John Baronello is director of marketing analytics at Laroche, M., I. Kiani, N. Economakis, and M. University, 2002. Allstate Insurance in Northbrook, IL. His career of O. Richard. “Effects of Multi-Channel Market- nearly 20 years has focused on developing analytics ing on Consumers’ Online Search Behavior: The Dijkstra, M., H. Buijtels, and W. F. van Raaij. solutions for organizations’ use of data to change Power of Multiple Points of Connection.” Jour- “Separate and Joint Effects of Medium Type on customer behavior and grow revenue. Before joining Consumer Responses: A Comparison of Televi- nal of Advertising Research 53, 4 (2013): 431–443. Allstate, Baronello held senior roles at Cardinal Path and sion, Print, and the Internet.” Journal of Business Walgreens, developing advanced retail analytics. Research 58, 3 (2005): 377–386. Molitor, D., P. Reichhart, and M. Spann. “Location-Based Advertising: Measuring the

Rex Briggs is founder and chief executive officer of Edell, J. A., and K. L. Keller. “The Information Impact of Context-Specific Factors on Consum- Marketing Evolution, a New York City-headquartered Processing of Coordinated Media Campaigns.” ers’ Choice Behavior” (working paper). Munich, analytics firm specializing in omnichannel marketing Journal of Marketing Research 26, 2 (1989): Germany, Ludwig Maximilian University, 2014. attribution and optimization. Briggs’s work can be found 149–163. in the Journal of Advertising Research and the Journal Moran, G., L. Muzellec, and E. Nolan. “Con- of Interactive Marketing, among other publications. He Enoch, G., and K. Johnson. “Cracking the Cross- sumer Moments of Truth in the Digital Con- is the coauthor of What Sticks: Why Advertising Fails Media Code: How to Use Single-Source Meas- text.” Journal of Advertising Research 54, 2 (2014): and How to Guarantee Yours Succeeds (New York: ures to Examine Media Cannibalization and 200–204. Kaplan, 2006), and he serves on the editorial board of Convergence.” Journal of Advertising Research 50, 2 (2010): 125–136. ESOMAR’s Research World magazine. Navarro-Bailon, M. A. “Strategic Consistent Messages in Cross-Tool Campaigns: Effects on Fulgoni, G., and A. Lipsman. “Digital Game Brand Image and Brand Attitude.” Journal of REFERENCES Changers: How Social Media Will Help Usher Marketing Communications 18, 3 (2012): 189–202. in the Era of Mobile and Multi-Platform Andrews, M., X. Luo, F. Zheng, and A. Ghose. Campaign-Effectiveness Measurement.”Journal “Mobile Ad Effectiveness: Hyper-Contextual Neijens, P., and H. Voorveld. “Cross-Platform of Advertising Research 54, 1 (2014): 11–16. Targeting with Crowdedness.” Marketing Science Advertising: Current Practices and Issues for the Future.” Journal of Advertising Research 55, 4 35 (2015): 218–233. Ghose, A., S. P. Han, and S. H. Park. “Analyzing (2015): 362–367. the Interdependence between Web and Mobile Bart, Y., A. T. Stephen, and M. Sarvary. Advertising: A Randomized Field Experiment” “Which Products Are Best Suited to Mobile (working paper). Leonard N. Stern School of Okazaki, S., and M. Hirose. “Effects of Dis- Advertising? A Field Study of Mobile Display Business, New York University, 2013. placement–Reinforcement between Traditional Advertising Effects on Consumer Attitudes and Media, PC Internet and Mobile Internet: A Intentions.” Journal of Marketing Research 51, 3 Grewal, D., Y. Bart, M. Spann, and P. P. Zubc- Quasi-Experiment in Japan.” International Jour- (2014): 270–285. sek. “Mobile Advertising: A Framework and nal of Advertising 28, 1 (2009): 77–104.

460 JOURNAL OF ADVERTISING RESEARCH June 2017 How Brands Can Make Smarter Decisions in Mobile Marketing thearf.org

Shankar, S., and S. Balasubramaniam. “Mobile and Previous Brand Attitudes and Knowledge.” Tsao, J. C., and S. D. Sibley. “Displacement Marketing: A Synthesis and Prognosis.” Journal Journal of Advertising 34, 4 (2005): 55–67. and Reinforcement Effects of the Internet and of Interactive Marketing 23, 2 (2009): 118–129. Other Media as Sources of Advertising Informa- Steele, A., D. Jacobs, C. Siefert, R. Rule, et al. tion.” Journal of Advertising Research 44, 1 (2004): 126–142. Shankar, V., M. A. Kleinjnen, S. Ramanathan, “Leveraging Synergy and Emotion in a Multi- R. Rizley, et al. “Mobile Shopper Market- Platform World: A Neuroscience-Informed Vandeberg, L., J. M. J. Murre, H. A. M. Voor- ing: Key Issues, Current Insights, and Future Model of Engagement.” Journal of Advertising veld, and E. G. Smit. “Dissociating Explicit and Research Avenues.” Journal of Interactive Market- Research 53, 4 (2013): 417–430. Implicit Effects of Cross-Media Advertising.” ing 34 (2016, May): 37–48. International Journal of Advertising 34, 5 (2015): Tavassoli, N. T. “Language in Multimedia: 744–764. Snyder, J., and Garcia-Garcia, M. “Advertising Interaction of Spoken and Written Information.” across Platforms: Conditions for Multimedia Journal of Consumer Research 25, 1 (1998): 26–37. Voorveld, H. A. M., and S. M. Valkenburg. Campaigns. A Method for Determining Optimal “The Fit Factor: The Role of Fit between Ads in Media Investment and Creative Strategies across Understanding Cross-Media Synergy.” Journal Taylor, J., R. Kennedy, C. McDonald, L. Lar- Platforms.” Journal of Advertising Research 56, 4 of Advertising 44, 3 (2015): 185–195. guinat, et al. “Is the Multi-Platform Whole (2016): 352–367. More Powerful Than Its Separate Parts? Measur- Voorveld, H. A. M., P. C. Neijens, and E. G. Stammerjohan, C., C. M. Wood, Y. M. Chang, ing the Sales Effects of Cross-Media Advertis- Smit. “Opening the Black Box: Understanding and E. Thorson. “An Empirical Investigation of ing.” Journal of Advertising Research 53, 2 (2013): Cross-Media Effects.” Journal of Marketing Com- the Interaction between Publicity, Advertising, 200–211. munications 17, 2 (2011): 69–85.

June 2017 JOURNAL OF ADVERTISING RESEARCH 461