Using Internet Behavior to Deliver Relevant Television Commercials

INTMAR-00124 No. of pages 11 4C Avai tryouting groundle online at www. sciencedirect. com diary of Interactive marketing xx (2013) xxx xxx www. elsevier. com/ sink/intmar Using earnings way to contain Relevant video Commercials St until now bellman a,? & Jamie Murphy b, d & Shiree Treleaven-Hassard a & James OFarrell c & Lili Qiu c & Duane Varan a a Audience Re count Labs, Murdoch University, 90 South Street, Murdoch, WA 6150, Australia Australian School of Management, Level 1, 641 Wellington Street, Perth, WA 6000, Australia Business School, University of Western Australia, 35 Stirling mettlesomeway, Crawley, WA 6009, Australia dCurtin Graduate School of Business, 78 Murray Street, Perth, WA 6000, Australia b c Abstract Consumer footprints left on the meshing help advertisers show consumers applicable sack up ads, which ontogenesis aw arness and click-throughs. This proof of concept experiment illust positions how Internet appearance throw out identify relevant telev ision mercenarys that join on ad- installiveness by raising wariness and ad pictorial matter. output engagement and foregoing check ikon, however, complicate military forceive Internet-targeting. Ad relevancy matters to a greater extent for low- interest returns, which have a footling pre-purchase count process.For the same reason, victimisation Web shop behavior to make inferences or so current ad relevancy is much accurate for low-involvement crossroads. Prior brand depiction reduces schooling- prise, all the same for relevant commercials, and therefore dampens ad relevancys pith on attention and ad exposure. 2013 Direct merchandise Educational Foundation, Inc. Published by Elsevier Inc. All fulls reserved. Keywords Consumer search behavior advertise Ad relevance Product involvement wayal targeting Attention Ad invalidateance Television Internet Experiment Heart rate IntroductionTelevision, declining in value for advertisers in recent years, is shr inking as a mass medium due to the proliferation of networks and consequent audience fragmentation. At the same age, digital video recorders (DVRs) change TV ad avoidance (Wilbur 2008). Finally, advertising budgets are shifting to other media such(prenominal) as the Internet, where interest- base targeting has increased measurement ad authorization by 65% (Goldfarb and Tucker 2011). Addressability, heralded decades ago, uses technology to track customer preferences and subsequently tailor advertising (Blattberg and Deighton 1991).Sending ads wholly to interested kinspersons improves advertisings value for consumers by increasing its relevance, and for advertisers by trim down wastage (Gal-Or and Gal-Or 2005 Gal-Or et al. 2006 Iyer, Soberman, and Villas-Boas 2005). advertize addressability ? Corresponding author. E-mail addresses s. emailprotected edu. au (S. bellman), jamie. emailprotected com (J. Murphy), emailprotected com (S. Treleaven-Hassard), emailprotected com (J. OF arrell), lili. emailprotected edu. au (L. Qiu), emailprotected com (D. Varan). establish on consumer Web behavior could apply to other media nd devices such as television, quick-witted phones, tablet devices and satellite radio (Shkedi 2010). Although search engine keywords and online social network data could augment targeting based on Web search behavior (Delo 2012 Jansen and Mullen 2008 Jansen et al. 2009), this available advertising proof of concept paper uses whole Web search behavior. Currently, TV advertisers target relevant commercials based on location, lifestyle and get data (Marcus and Walpert 2007). A cable company, for instance, might use subscriber information to send different ads to different ethnic groups (Vascellaro 2011b).But information in these databases can be months or years old. Current crossroad and brand interest based on Internet behavior could add a new layer to a targeting database. intimately all (85%) of the United States population are Intern et users (Pew Internet and American Life Project 2012), leaving digital footprints that suggest product interest. Cable companies that tract cable and broadband Internet services, Comcast for example, could align household Internet and TV- see data to increase the relevance of marketing communication. The basic intuition behind targeting TV ads based on Web rowsing behavior is that cartridge clip spent browsing pages in a 1094-9968/$ -see front matter 2013 Direct merchandise Educational Foundation, Inc. Published by Elsevier Inc. All rights reserved. http//dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 Please state this condition as Steven Bellman, et al. , Using Internet Behavior to forgo Relevant Television Commercials, diary of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 S. Bellman et al. / diary of Interactive Marketing xx (2013) xxxxxx 2 certain product stratum increases interest in commercials for brands in that category.This i ntuition necessarily empirical testing, and the literature on consumer search suggests that differences among product categories may complicate applying this intuition (Richins and Bloch 1986). This paper opens with our conceptual framework, which distinguishes ad relevance from product involvement (Batra and radiation therapy 1983). Consumers tend to use an ongoing search process (Bloch and Richins 1983) for amply-involvement products buying the wrong brand entails greater financial, social, or psychological risks than for low-involvement products (Ros orderr and Percy 997). Internet obtain strategies differ, therefore, amidst high- and low-involvement products (Moe 2003). These differences in involvement, along with prior brand exposure, lead to four hypotheses about the performances of TV ad relevance disc everywhereed via Web-browsing behavior. After a discussion of the methodological abstract and results, the paper closes with implications, limitations and afterlife res earch avenues. Conceptual Framework Ad Relevance and the Consumer look for Process advertize has relevance before, du fudge, and after purchase (Vakratsas and Ambler 1999).Consumer pre-purchase search has two phases, exploratory and purposeful search (Janiszewski 1998). Consumer information needs change from generic product information (e. g. , hotels) to brand-specific information (e. g. , Hilton), including advertising by these brands (Rutz and Bucklin 2011). In St. Elmo Lewis continent AIDA model (Strong 1925), exploratory search begins with awareness consumers archetypal recognize their need for a product. As interest grows, they explore options in the category and seek information from friends and the media, including the Internet. In the later oal-directed search phase, they desire a particular product or brand. Finally, they put that desire into action and buy a specific brand. Ad relevance for a product is highest during purposeful search, freeze off during explorator y search, and practically non-existent with consumers unaware of a product need. Product interest and Web Browsing Behavior Moe (2003) illustrates how useful matching ads to Web browsing behavior can be, and the complications associated with product involvement. Most products are low-involvement, attracting attention only during the pre-purchase search process (Bloch and Richins 983). Since pre-purchase search for these products by and large ends in a purchase, the search process for low-involvement products has an warm purchasing horizon. But the risks associated with high-involvement products lead many consumers, especially product enthusiasts, to engage in ongoing search, to unendingly update their knowledge or just for enjoyment (Richins and Bloch 1986). Examples of such products include automobiles, computers, and fashion items (see Table 2 later). A search for information about a high-involvement product may not end in a purchase, and often has a future urchasing horizon. Moe (2003) utilize two dimensions, low versus high ad relevance (exploratory vs. goal-directed search) and low versus high involvement (immediate vs. future purchasing), in a 2 ? 2 matrix to define four Web browsing strategies utilize by Internet shoppers (Table 1). Moe (2003) categorized visitors to a real stores Web site, which sold nutrition products such as vitamins, into these four strategies. Shoppers interested in a low-involvement product with an immediate purchasing horizon adopt a hedonic browsing dodging during exploratory search, and advertising has low relevance.They use the directed buying strategy during goal-directed search, and advertising has high relevance. Shoppers use the other two strategies for a high-involvement product with a future purchasing horizon. Advertising for high-involvement products should have relatively lower relevance for shoppers using the exploratory knowledge building strategy, compared to shoppers using the goal-directed search/ delibera tion strategy. Table 1 also reports the average Web browsing clipping for these four strategies. These data suggest that long versus short Web browsing time can signal high ad relevance for low-involvement products.Directed buyers averaged over 36 minutes visiting the online store. In contrast, hedonic browsers spent one one-fifth as much time on the site, about seven minutes. big versus short Web browsing time, however, may not signal high ad relevance for high-involvement products. First, average Web browsing time is nearly 3? multiplication longer for high- rather than low-involvement products due to the ongoing nature of search for these products (Richins and Bloch 1986). Second, Moes (2003) data suggest that the opposite pattern of Web browsing times leave behind indicate low versus high ad relevance for high-involvement products.In line with theory that predicts an inverse-U order of product experience on search activity (Moorthy, Ratchford, and Talukdar 1997), knowledge- building shoppers (low ad relevance) preserve the longest Web browsing times, nearly two hours in a single seance. Shoppers with a search/deliberation strategy (high ad relevance) and extensive category knowledge focus their search time on specific products or brands and record relatively shorter Web browsing times, about the same duration as directed buyers. Table 1 Influence of ad relevance and product involvement on Web browsing behavior. Product involvementAd relevance low-pitched (exploratory search) humiliated (immediate purchasing horizon) High (future purchasing horizon) High (goal-directed search) SHORT Hedonic browsing (641) LONG Knowledge building (11147) LONG Directed buying (3633) SHORT Search/ deliberation (3759) NOTEAdapted from Moe (2003). metrical composition in parentheses are the average Web site browsing time for separately of the four Internet shopping strategies (minutesseconds). Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, ledger of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. ntmar. 2012. 12. 001 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx The next section uses this conceptual framework to propose four hypotheses about the effects of ad relevance, indicated by Web browsing behavior, on attention and ad exposure. Hypotheses Moderating Effect of Product elaboration According to the conceptual framework above, Web browsing behavior can suggest ad relevance. A long time browsing information about a product indicates a consumer seeming in goal-directed search for that product brand advertising has high relevance, just only for low-involvement products.For highinvolvement products, Web browsing behavior is unrelated to ad relevance, or the opposite pattern, short rather than long Web browsing time, is likely to signal greater ad relevance. When advertising is relevant, that is, a consumer is in the goal-directed pha se of product search, a TV commercial for that product should receive above average attention. When people pay attention to external stimuli, their sum rate goes down, most likely to littleen interference with information-intake (Lacey 1967). In other words, greater attention to relevant ads will associate with a decrease in heart rate.Ad relevance should also increase ad exposure, by reducing ad avoidance. As viewers may avoid TV commercials mechanically by channel-changing or fast-forwarding, addressable commercials interest TV advertisers as a method to combat ad avoidance. This ad exposure is better measured in viewing time, which conveys more information than a innocent binary measure of ad avoidance (Gustafson and Siddarth 2007). Single-source data that match a households commercial viewing time to its purchase history suggests viewers are more likely to abide by relevant ommercials, that is, commercials for products the household buys, as opposed to irrelevant commerci als (Siddarth and Chattopadhyay 1998). A recent field trial found that addressable TV ads can reduce ad avoidance by 32% (Vascellaro 2011a). Less ad avoidance means longer viewing times for commercials, and therefore high ad relevance commercials will increase ad exposure. According to the conceptual model in Table 1, high versus low product involvement is likely to chink the reliability of Web browsing time as an indicator of high versus low ad relevance, attention, and ad exposure.High involvement with a product is likely to show into high interest in advertising by brands of that product during both(prenominal) exploratory and goal-directed search. For high-involvement products, therefore, TV commercials could have high ad relevance, attention, and ad exposure, whether or not Web browsing behavior has been recently observed. Furthermore, for high-involvement products, short rather than long Web browsing time could indicate relatively greater ad relevance. Consumers, however, a re less likely to seek information online or offline about low-involvement products (Bloch andRichins 1983 Bloch, Sherrell and Ridgway 1986). This suggests that Web browsing for low-involvement products is highly valuable for behavioral targeting, as pre-purchase search for these products is for an immediate need (Moe 2003). For low-involvement products, Web browsing behavior should be a 3 highly reliable indicator of ad relevance, attention and ad exposure for TV commercials, but this will not be the case for high-involvement products. Thus, product involvement will moderate the effects of ad relevance indicated by Web browsing behavior H1.Ad relevance based on Web browsing behavior will increase attention to commercials for low-, but not for high-involvement products. H2. Ad relevance based on Web browsing behavior will increase ad exposure to commercials for low-, but not for high-involvement products. Moderating Effect of Prior steel mental picture Another variable likely to moderate addressability effects is prior exposure to advertising for a brand. Prior brand exposure reduces a commercials information value, even when that information is relevant (Campbell and Keller 2003 Pechmann and Stewart 1989).Prior exposure should therefore reduce a viewers willingness to pay attention to the commercial (Potter and Bolls 2012), or to choose ad exposure over ad avoidance (Bellman, Schweda, and Varan 2010 Woltman Elpers, Wedel, and Pieters 2003). Hypotheses 3 and 4 predict that prior brand exposure moderates the effects of ad relevance and involvement on attention and ad exposure H3. Prior brand exposure reduces the effect of ad relevance on attention to commercials for low-involvement products. H4. Prior brand exposure reduces the effect of ad relevance on ad exposure to ommercials for low-involvement products. The next section describes the experiment to test these four hypotheses. Methodology Overview To test the concept of using Internet behavior to part wi th relevant TV commercials, this experiment drew on two seemingly unrelated lab sessions. In the first lab session, apiece participants Web browsing behavior was analyzed to discover highly relevant products. In the second lab session, this knowledge was used to individually customize the playlist of TV commercials shown to each participant. Sample and Design The experiment was a 2 ? 2 ? 2 mixed design. Prior brand xposure (yes/no) was a between-participants factor. The yes group saw Web banner ads in the first lab session, exposing them to visual aspects of the TV commercials for the same brands shown in the second lab session. All TV commercials were for U. S. brands unavailable in the test market, Australia, ensuring no prior brand exposure in the no group. Ad relevance (high/low) and Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. int mar. 2012. 12. 001 4 S.Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx A. The crustal plate page for the sise high-involvement product categories. B. The home page for a subcategory of high-involvement products credit cards. Fig. 1. The Web site used to unobtrusively measure interest in 12 product categories. A. The home page for the cardinal high-involvement product categories. B. The home page for a subcategory of high-involvement products credit cards. Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. oi. org/10. 1016/j. intmar. 2012. 12. 001 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx product involvement (high/low) were both within-participants factors for the TV commercials shown in the second lab session. A total of 211 members of an audience panel, representative of the Australian public, earned $30 (AUD) to participate in two lab sessions totaling 90 minutes. These participants were randomly assigned to the two between-participants groups (yes, prior brand exposure = 109, no = 102). Half the sample (49%) were women, and ages ranged from 19 to 78 years (M = 45, SD = 15).All had high levels of Internet experience (Venkatesh and Agarwal 2006). Careful cognitive operations, such as describing the two lab sessions as separate studies, helped ensure that participants were unaware that their Web browsing behavior in the first lab session influenced the TV commercials served in the second lab session. Lab Session 1 In the first lab session, participants evaluated the fictitious Consumer Choices Web site (Fig. 1A), which displayed information about six high- and six low-involvement product categories, identified from published classifications (Kover and Abruzzo 1993 Ratchford 1987 Rossiter, Percy, and Donovan 991 Vaughn 1986). Each product category had trinity subcategories (Table 2). The c ardinal pages of subject field for each of these 36 subcategories were matched crossways products for depth, breadth and reading level to allow meaningful time-in-category comparisons. Participants had four minutes to explore the six highinvolvement categories, and another four minutes to explore the six low-involvement categories (the order, high- or lowinvolvement first, was randomized). Browsing time in each category was logged. For each participant, the two product ategories (one high- and one low-involvement) browsed for the longest time were that participants two high ad relevance categories. The two corresponding low ad relevance categories (one high- and one low-involvement) were randomly selected from the participants categories with the shortest browsing times (e. g. , 0 seconds). For participants in the prior brand exposure group, banner advertisements were at the top of each page. In the no prior brand exposure group, a generic photo-montage of the same size occupied t his ad space. Each of the 36 subcategories advertised a different brand.For each participant, one brand was chosen randomly to represent its subcategory across both stages of the experiment (e. g. , great(p) One, Fig. 1B), from the two brands available for each subcategory, a total of 72. The duration of prior exposure to a brand was the time the participant spent viewing pages of content about the brands subcategory (i. e. , prior exposure was higher for high ad-relevance categories). Lab session 1 ended after participants completed an extensive online survey about the Web sites usability (Agarwal and Venkatesh 2002 Venkatesh and Agarwal 2006). This survey reated a 20-minute delay, realistically replicating the process of identifying ad relevance based on Web browsing behavior, and subsequently delivering a set of customized commercials to a TV set-top box. 5 Lab Session 2 Participants went to a different laboratory for the second lab session, in which they evaluated new TV progra ms. Participants first verified their shout and date of birth displayed on the TV screen, to ensure no miss-targeting of the customized ads (Gal-Or et al. 2006). They then practiced using the TV remote control to select programs and mechanically avoid ads.Participants selected one of four new one-hour U. S. television programsdrama, comedy, reality or documentaryto evaluate for potential airing in Australia. They were told these programs had been recorded off-air in the U. S. , with ads included. This selection social function successfully eliminated differences in program like (Coulter 1998), which can affect advertising response (Norris, Colman, and Aleixo 2003). Each program had five ad breaks, with five 30-second ads in each break. The ads shown in the first four breaks were individually customized based on the ad relevance information discovered in the first lab session.The four test ads for two high ad-relevance products (one high- and one low-involvement) and two low ad-re levance products (one high- and one low-involvement)were counterbalanced across the first four breaks, always appearing in the middle rig to avoid primacy and recency effects (Pieters and Bijmolt 1997). The remaining eight product categories each contributed two filler ads, the 16 required for the first four ad breaks. The fifth ad break, which always showed the same five filler ads, created a natural delay before measuring brand recall. While participants seeed their chosen program, the two ependent variable measures were collected unobtrusively. Attention was heart rate decrease relative to each participants pre-program baseline heart rate (Potter and Bolls 2012). The slowest heart rate during a commercialrepresenting the peak of attention (Lang et al. 1993)was subtracted from the participants slowest resting-baseline heart rate (Wainer 1991). Heart rate was measured via pulse photoplethysmography at two places the lobule of the ear and the distal phalanx of the non-dominant han ds ring finger. The signal, ear or finger, with the fewest artifacts (mainly caused by movement) was retained.Sixty-four participants (30% of 211, women = 47%, age range 19-75 yrs) consented to this procedure and yielded usable heart rate data. None of these participants was on medication that affects heart rate (Andreassi 2007). Thanks to an efficient mixed-level design, the size of this sub-sample was sufficient to test the two attention hypotheses with 99. 9% power (Faul et al. 2007). Ad exposure was the number of seconds that the commercial displayed on the screen before avoidance. Participants avoided ads by pressing the remote controls skip button, which jumped to the next ad or program segment.In this experiment skipping was impossible during the program and during the first five seconds of each commercial, to ensure that each skip decision was on the merits of the ad rather than a general goal of avoiding all commercials. A matched sample (n = 81) substantiate that this pro cedure added a non world-shattering 1. 67 seconds of ad exposure, compared to participants able to skip at any time. Although previous studies have used ad viewing time to measure ad attention (Olney, Holbrook, and Batra 1991), in this ask Please cite this article as Steven Bellman, et al. Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx 6 Table 2 Product categories and subcategories. Involvement Category Subcategories High Automotive 1. Luxury Cars 2. Compact 4WDs 3. Sedans 4. Credit Cards 5. Financial Planning 6. Retail Banking 7. Digital Televisions 8. calculators 9. Kitchen and Laundry Appliances 10. Jewellery 11. Casual stretch out 12. Sportswear 13. sept Insurance 14. Automotive Insurance 15. Life Insurance 16. Deodorant 7. Hair Care 18. Allergy Medication 19. Hamburgers 20. Mexican Food 21. jaundiced 22. Household Cleaners 23. Laundry Detergent 24. Cleaning Tools 25. Gardening 26. Tools 27. Pest Control 28. Chocolate Bars 29. Mints 30. Chewing Gum 31. Soft Drinks 32. Energy Drinks 33. burnt umber 34. Frozen Meals 35. Packaged Meats 36. Desserts Financial Services Technology Fashion Apparel Insurance Health & Well-Being Low Fast Food Home Cleansers Home Maintenance Candy Beverages Packaged Food NOTEFor every subcategory, two brands were available for selection (i. e. , 72 brands). attention and ad exposure were uncorrelated (r = ? 06, p = . 665), justifying the use of both measures. After watching the one-hour program, participants completed a second online survey on the same flat screen monitor used to watch the program. In line with the cover story for lab session 2, this survey began by measuring program liking (Coulter 1998 Cronbachs alpha = . 96). The survey went on to measure treatment checks of ad relevance and product involvement, and managerially relevant ou tcomes associated with greater attention and ad exposure (see the Appendix A). After completing this survey, participants were debriefed, hanked, and given their gift-card. products for which they were in the goal-directed search phase. This was confirmed by significant differences in self-reported purchasing horizon, measured in the post test (Table 3). Products classified as high ad-relevance, based on Web browsing time, were more likely to be used or purchased in the next month than those classified as low ad-relevance (Mlow ad-relevance = 3. 65 times per month vs. Mhigh ad-relevance = 6. 78). As predicted by the conceptual framework in Table 1, a significant two-way interaction between ad relevance and product involvement ualified this Internet-targeting main effect (Table 3). Using Web browsing time, ad relevance was inferred more accurately for low- rather than high-involvement products. For high-involvement products, purchase/usage was more likely for products inferred as low ad-relevance, based on Web browsing time (Mlow ad-relevance = . 20 times per month vs. Mhigh ad-relevance = . 10). Failure to observe Web browsing did not indicate low ad-relevance for high-involvement products, and as shown in Table 1, short rather than long Web browsing time could indicate relatively greater ad relevance.Also in line with Table 1, low-involvement products had a significantly shorter purchasing horizon compared to highinvolvement products (Mlow-involvement = 10. 28 times per month vs. Mhigh-involvement = . 15 Table 3). Product Involvement The manipulation of product involvement was also successful, measured by self-reported product involvement (Mlow-involvement = 4. 02 on a 7-pt scale vs. Mhigh-involvement = 4. 93, p b . 001, partial ? 2 = . 27), even without individual customization. No other effects were significant (e. g. , ad relevance Mlow ad-relevance = 4. 40 vs.Mhigh ad-relevance = 4. 55, p = . 213, partial ? 2 = . 007). Table 3 ANOVA results. Effect Within -participants effects Ad relevance Product involvement Purchasing horizon (monthly frequency) Attention (heart rate decrease) Ad exposure (viewing time in seconds) 10. 08** (. 05) 122. 15*** (. 37) 10. 78** (. 05) 1. 26 (. 01) .19 (. 001) 1. 40 (. 01) 3. 67 (. 06) 1. 34 (. 02) 1. 64 (. 03) 2. 17 (. 03) .27 (. 004) 4. 64* (. 07) 7. 14** (. 03) 2. 42 (. 01) 1. 90 (. 01) .38 (. 002) 2. 47 (. 01) 1. 02 (. 005) .17 (. 001) 209 .01 (b . 001) 62 .56 (. 003) 209 Independent varying ChecksAd relevance ? product involvement Ad relevance ? prior brand exposure Product involvement ? prior brand exposure Ad relevance ? product involvement ? prior brand exposure mingled with-participants effect Prior brand exposure via Web banner ads Error degrees of freedom Ad Relevance The validity of the ad relevance factor depends critically on whether participants spent more time in lab session 1 looking at NOTESF ratios (hypothesis degrees of freedom = 1). Numbers in parentheses are effect sizes (partial ? 2) small = . 01, medium = . 06, large = . 14. Significant effects in bold. p = . 06, * p b . 05, ** p b . 01, *** p b . 001. Results Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx Fig. 2B shows that, in line with H1, ad relevance based on Web browsing time increased attention to commercials for low-, but not for high-involvement products. Attention was measured by heart rate decrease (HRD) the greater the ecrease, the more attention to the commercial. But H1 was only partially supported, as this effect was significant only without prior brand exposure (H1 in Table 4), as predicted by H3 (see below). The effect of ad relevance on ads for low-involvement products generated a marginally significant main effect of ad relevance on attention (Tables 3 and 4). Similarly, planned contrasts (Winer 1991) showed that in line with H2, ad relevance based on Web browsing time increased ad exposure to commercials for low-, but not for high-involvement products (Fig. A and H2 in Table 4). Ad exposure was measured by ad viewing time how much of an ad was seen before pressing the skip button. A longer ad viewing time means more ad exposure and less ad-avoidance. This effect delivered a significant effect of ad relevance even after collapsing across low- and high-involvement products (Table 3). Moderating make of Prior Brand pic Hypotheses 3 and 4 The effect of ad relevance on attention to commercials for low-involvement products predicted by H1 was qualified by the significant three-way interaction predicted by H3, among ad elevance, product involvement and prior brand exposure (Table 3). Prior brand exposure reduced the effect of ad relevance on attention to commercials for low-involvement products, most likely because prior brand exposu re reduced their information-value. After prior brand exposure, viewers salaried equal attention to the test commercials, no matter what their ad relevance (Fig. 2B and H3 in Table 4). Prior brand exposure also reduced the effect of ad relevance on ad exposure to commercials for low-involvement products, as predicted by H4. After prior brand exposure, ad exposure DiscussionThis analyse tested the effectiveness of Internet-targeted TV advertising, using recent Web browsing to identify a households relevant TV commercials. The results suggest that this method of Internet-targeting significantly increases attention and ad exposure, even when based only on Web browsing behavior rather than search-engine keywords. These results echo similar field trials of addressable TV ads (Vascellaro 2011a) and single-source data (Siddarth and Chattopadhyay 1998), which have shown how ad relevance can increase TV ad exposure. However, these results also show that product nvolvement and prior brand e xposure complicate Internettargeting of TV commercials. First, the overall effect of Internet-targeting on ad exposure in this study was due solely to its effect on commercials for A. No Prior Brand Exposure -5 Attention (heart rate decrease bpm) Effects of Ad Relevance Hypotheses 1 and 2 was not significantly longer for high- versus low ad-relevance commercials for low-involvement products (Fig. 3B and H4 in Table 4). The results of the four hypothesis tests are summarized in Table 5. -6 -5. 84 -7 -8 -7. 88 -8. 43 -9 -9. 11 -10 Low Ad Relevance -11 High Ad Relevance -12Low High Product Involvement B. Prior Brand Exposure -5 Attention (heart rate decrease bpm) Prior Brand Exposure Prior brand exposure, via Web banner ads, increased brand recall but not significantly (Mno = 4. 3% vs. Myes = 6. 8%, p = . 132, partial ? 2 = . 011). Prior brand exposure did, however, have a significant two-way interaction with ad relevance (p = . 017, partial ? 2 = . 027). When prior brand exposure was present, brand recall was significantly higher for high versus low ad-relevance TV commercials (Mlow ad-relevance = 3. 2% vs. Mhigh ad-relevance = 9. 6%, p = . 016, partial ? 2 = . 053).When prior brand exposure was absent, brand recall was not significantly different for high versus low ad-relevance commercials (Mlow ad-relevance = 5. 4% vs. Mhigh ad-relevance = 3. 9%, p = . 441, partial ? 2 = . 006). Since ad relevance was impelled by Web browsing time, participants who recorded zero browsing times for their low ad-relevance categories had no prior brand exposure. No other effects were significant. In particular, prior brand exposure did not interact with product involvement, suggesting no differences in cognitive avoidance of Web banner ads in the first lab session for lowversus high-involvement products. -6 -7 -8 -7. 76 -8. 07 -7. 84 -8. 51 -9 -10 Low Ad Relevance -11 High Ad Relevance -12 Low High Product Involvement Fig. 2. The effects of ad relevance and product involvement on attention to TV commercials, measured by heart rate decrease, for the two prior brand exposure groups (A) no prior brand exposure, and (B) prior brand exposure via Web banner ads. Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 S. Bellman et al. Journal of Interactive Marketing xx (2013) xxxxxx 8 Table 4 Cell means. Low ad relevance Variable ? 7. 55 Attention (heart rate decrease) No prior brand exposure Prior brand exposure Ad exposure (viewing time in seconds) No prior brand exposure Prior brand exposure High ad relevance Test Low product High product Low product High product involvement involvement involvement involvement H1 ? 6. 95 ? 7. 13x ? 5. 84x H3 ? 7. 96 ? 8. 07 H2 19. 99x 19. 18x ? 8. 32 ? 8. 43 ? 8. 44 ? 8. 49x ? 9. 11x ? 7. 88 ? 7. 84 ? 8. 14 ? 7. 76 ? 8. 51 20. 79 21. 23x 21. 22x 21. 25 19. 48x 18. 79 x H4 8. 14 ? 8. 19 20. 16 21. 01x 21. 70x 20. 33 20. 50 19. 58 21. 42 21. 46 20. 75 22. 17 NOTES elbow room in the same row with the same superscript letters differ significantly (p b . 05) using planned contrast tests (except p b . 06). which in turn increases ad liking (r = . 25, p b . 001). Although consumers have privacy concerns about targeted advertising (Spangler, Hartzel, and Gal-Or 2006), these concerns about Internet-targeted TV commercials could be alleviated if these commercials displayed the Digital Advertising Alliances Advertising Choices Icon and viewers could opt out from eceiving these commercials (youradchoices. com). For advertisers, these results support the concept of using Internet-targeting to reduce wastage in advertising budgets. Internet targeting also increases the effectiveness of TV commercials, by increasing ad exposure, which increases brand recall (r = . 14, p b . 05) and purchase intention (r = . 34, p b . 001). The results also show that Internet targeting is more critical for advertising low-involvement products, such as food, as opposed to high-involvement products like durables. Although changing the habitual nature of low-involvement onsumption is hard, commercials for low-involvement products may often suffer from bad timing. To combat this, many advertisers use continuous advertising (Ephron 1995), which is expensive and counterproductive by increasing prior brand exposure. Internet-targeting provides a way of continually monitoring household interest in low-involvement products, showing ads only when they are relevant and minimizing prior exposure. Relevance for habitual purchases, for which the A. No Prior Brand Exposure Implications Ad Exposure (ad viewing time seconds) 25 21. 70 20 0. 16 20. 33 18. 79 15 Low Ad 10 Relevance 5 High Ad Relevance 0 Low High Product Involvement B. Prior Brand Exposure Ad Exposure 30 (ad viewing time seconds) low-involvement products. But targeting-accuracy may not matter for high-invol vement products, such as durables. Meta-analysis shows that advertising is more effective, on average, for durables rather than non-durables (Sethuraman, Tellis, and Briesch 2011). Consumers often gather information about high-involvement products they are not planning to purchase immediately (Moe 2003 Richins and Bloch 1986).Commercials for high-involvement products attract consistently high levels of attention and ad viewing time, as sources of information during the ongoing search process for these products. For this reason, ad-relevance can be high for high-involvement products, whether or not Web browsing behavior is observed. Second, prior brand exposure reduces the information-value of advertising (Campbell and Keller 2003). Consumers pay less attention to TV commercials, evaluate them more negatively, and are more likely to avoid them (Bellman, Schweda, and Varan 2010 Woltman-Elpers, Wedel, and Pieters 2003).In this study, prior brand exposure dampens the effects of ad relev ance and product involvement. Relevant commercials for low-involvement products receive more attention and ad exposure only when prior brand exposure is not present. 30 25 20 19. 58 20. 75 21. 42 22. 17 15 Low Ad 10 Relevance 5 High Ad Relevance 0 For consumers, the results of this study suggest that Internet targeting can improve their TV viewing experience. Internet targeting increases ad relevance, which means TV commercials are worth watching rather than avoiding. In this study, greater ad relevance due to Internet targeting increases ad exposure, Low HighProduct Involvement Fig. 3. The effects of ad relevance and product involvement on ad exposure, measured by ad viewing time for the two prior brand exposure groups (A) no prior brand exposure, and (B) prior brand exposure via Web banner ads. Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 101 6/j. intmar. 2012. 12. 001 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx Table 5 Results of hypothesis tests. Hypothesis Accepted? H1. Ad relevance, based on Web browsing ehavior, will increase attention to commercials for low-, but not for high-involvement products. H2. Ad relevance, based on Web browsing behavior, will increase ad exposure to commercials for low-, but not for high-involvement products. H3. Prior brand exposure reduces the effect of ad relevance on attention to commercials for low-involvement products. H4. Prior brand exposure reduces the effect of ad relevance on ad exposure to commercials for low-involvement products. PARTIALLY (with no prior brand exposure) YES YES YES household does not search online, might be determined by knowledge of the households shopping cycle.For advertisers of high-involvement products, ad timing is less critical, and traditional databases derived from cable subscription data, or warranty cards, seem adequate for targeting. And advertising still plays a role outside the consumer search process, most importantly to create awareness and interest in new purchases (Vakratsas and Ambler 1999). Conclusions Limitations withstanding, this study demonstrates how Webbased targeting can deliver the right TV commercial to the right person, and at the right time. Timeliness is particularly important for low-involvement products, as their relevance may change aily or even hourly. punctual Internet activity data can help TV advertisers identify commercials that currently interest a consumer. Digital-targetings potential heightens as individuals and households increasingly add devices and applications for online multi-tasking (Pilotta and Schultz 2005). This article illustrates a viable technique to tempt marketing practitioners and academics, and fuel information privacy concerns. A framework for information privacy research builds on three broad dimensions (1) multiple publics, (2) information channel d evelopments, and (3) public responses to privacy ctions (Peltier, Milne, and Phelps 2009). Failure to address privacy concerns is one of several limitations to this study and a promising future research avenue. Limitations and time to come Research Suggestions This studys main limitation is customizing ad relevance individually rather than group-wise (Richins and Bloch 1986) in order to test the concept of Internet targeting. various(prenominal) differences provide alternative explanations and add noise to the observed ad relevance effect (Cook and Campbell 1979). Using over 30 product subcategories helps distribute this noise evenly. The procedure in this article resembles how Fazio et al. 1986) investigated attitude accessibility. In two experiments, they individually customized a list of 16 attitude objects on the 9 basis of each participants reception times in a pretest, and validated this procedure in a third experiment by obtaining identical results using manipulated stimuli . Future experiments could use a similar procedure to manipulate ad relevance (Perkins and Forehand 2012). Another limitation is using Web-browsing rather than search-engine keywords to identify ad relevance. Parameters for the former were more viable for a controlled experiment (e. g. only 72 commercials were needed). However, searchengine queries provide a more direct and accurate means of identifying the consumers stage in the search process (Rutz and Bucklin 2011). Future studies may find the benefits of using search-engine queries are greater (Langheinrich et al. 1999). Internet-based targeting for high-involvement products might be improved by using search-engine queries, and more sophisticated analysis of Web browsing behavior. For example, Cai, Feng, and Breiter (2004) identify travel sites as highly relevant when a visitor views pages conveying specific as pposed to general information. Moe (2006) demonstrates how clickstream data can be used to infer both the stage of the decision process and the decision rule, which together might help identify abnormally high ad relevance for highinvolvement products. This study used ad viewing time as a measure of ad exposure. But in other studies, especially field studies, the relationship between ad viewing time and effectiveness may not be positive (cf. Tse and Lee 2001). For example, Greene (1988) observed that an ad avoider in the field has to really watch the set to see/know/perceive what she or he is doing nd ends up with more commercial exposure value (p. 15). Future studies should attempt to repeat these findings in field trials. Also, ad exposure may have nonlinear threshold effects, 1 or be affected by differences between commercials (Woltman Elpers et al. 2003). A promising future research avenue is experimentally manipulating the content of ads (e. g. , Teixera, Wedel, and Pieters 2010), as well as their ad relevance. Ideally, other psychophysiological measures of attention (Potter and Bolls 2012) co uld have been used but in the current setting eart rate was the least invasive. The manipulation of prior brand exposure was too weak to generate a main effect on explicit memory, but did have a significant interaction effect. The explanation is most likely that prior brand exposure was manipulated by the bearing of Web banner ads and these ads tend to be processed preattentively or cognitively avoided (Chatterjee 2008 Dreze and Hussherr 2003). Future studies could manipulate prior exposure using more attention-getting stimuli, such as brand integrations in Web site editorial. If Web banners are used, implicit measures 1For example, brand recall may require a minimum ad exposure equal to 70% of an ads duration (21 s for a 30 s ad). To test for a non-linear threshold effect of ad exposure on brand recall, ad exposure was categorized into ? ve bins, 09 s, 1015 s, 1621 s, 2225 s, and 2630 s. This analysis revealed only a signi? cant linear trend (p b . 001, partial ? 2 = . 040) in the means for these bins 0%, 1. 6%, 2. 5%, 3. 9%, 10. 5%. This result may have differed, however, if the study had measured message recall. The authors thank an anon. reviewer for suggesting this analysis. Please cite this article as Steven Bellman, et al. Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 10 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx of banner ad effectiveness could be used as manipulation checks (Perkins and Forehand 2012). A final limitation of this study is investigating the effect of targeting ads solely by interest in a product category. Future studies could examine the effects of other personalization strategies, such as interest in specific brands, programs, fanciful execution styles, and offers (Verhoef et al. 010). Each of these strategies merits evaluation and comparison in order to determine effective methods of targetin g addressable TV advertising. Acknowledgments The authors would like to thank the editor and the two anonymous reviewers for their constructive feedback during the review process. The authors are also grateful to Adrian Duffell, Karl Dyktinski, Emily playing fielder, Michael Gell, Shannon Longville, and a team of research assistants for their considerable help in conducting the experiment reported here. This research was funded by the sponsors of the Beyond 30 project (www. beyond30. org). Appendix A.Manipulation-checks and other measures In addition to the two unobtrusive measures of attention and ad exposure collected during lab session 2, which were the main dependent variables, an online survey at the end of the second lab session collected self-report measures of manipulation checks and managerially relevant outcome measures. Except for product involvement (Mittal 1995 alpha = . 97), the survey used validated single-item measures (e. g. , ad liking Bergkvist and Rossiter 2007) . To accommodate the slightly different question wording required for each of the 72 brands, convinced(p) selecting only the articipants four test brands to ask questions about, the survey did not use a random order of questions, but the following fixed, minimally biasing order (Rossiter and Percy 1997). Brand recall (unaided correct brand recall = 1, else = 0) was measured after program liking. Purchase intention was measured next, using Justers (1966) 11-point scale for high-involvement products and Jamieson and low-pitcheds (1989) 5-point scale for low-involvement products. Ad liking was next, followed by product involvement, and finally purchasing horizon purchase/usage frequency per month, measured by different 8-point scales for low- and igh-involvement products (low never to 3 or more times a day high do not plan to purchase to within the next month Goldberg and Gorn 1987). For every measure except purchasing horizon, dont know options helped avoid over-use of scale mid-poi nts (Green, Goldman, and Salovey 1993). Missing data were replaced by the subjects mean, a conservative strategy (Blumenthal et al. 2005). References Agarwal, Ritu and Viswanath Venkatesh (2002), Assessing a Firms Web Presence A Heuristic evaluation Procedure for the Measurement of Usability, reading Systems Research, 13, June, 16886.Andreassi, John L. (2007), Psychophysiology Human Behavior and Physiological Response. 5th ed. Mahwah, NJ Lawrence Erlbaum Associates. Batra, Rajeev and Michael L. Ray (1983), Operationalizing Involvement as Depth and tonicity of Cognitive Response, Advances in Consumer Research, Vol. 10. Richard P. Bagozzi, Alice M. Tybout, editors. Ann Arbor, MI Association for Consumer Research, 30913. Bellman, Steven, Anika Schweda, and Duane Varan (2010), The Residual Impact of Avoided Television Advertising, Journal of Advertising, 39, Spring, 6781.Bergkvist, Lars and John R. Rossiter (2007), The Predictive Validity of Multiple-item Versus Single-item Measures of the Same Constructs, Journal of Marketing Research, 44, May, 17584. Blattberg, Robert C. and John Deighton (1991), Interactive Marketing Exploiting the Age of Addressability, Sloan Management Review, 33, Fall, 514. Bloch, Peter H. and Marsha L. Richins (1983), A Theoretical Model for the Study of Product impressiveness Perceptions, Journal of Marketing, 47, Summer, 6981. , Daniel L. Sherrell, and Nancy M.Ridgway (1986), Consumer Search An Extended Framework, Journal of Consumer Research, 13, June, 11926. Blumenthal, Terry D. , Bruce N. Cuthbert, Diane L. Filion, Steven Hackley, Ottmar V. Lipp, and Anton Van Boxtel (2005), Committee Report Guidelines for Human Startle Eyeblink Electromyographic Studies, Psychophysiology, 42, 1, 115. Cai, Liping A. , Ruomei Feng, and Deborah Breiter (2004), Tourist Purchase finish Involvement and education Preferences, Journal of Vacation Marketing, 10, April, 13848. Campbell, Margaret C. nd Kevin Lane Keller (2003), Brand Familiarity and Advert ising Repetition Effects, Journal of Consumer Research, 30, September, 292304. Chatterjee, Patrali (2008), Are Unclicked Ads Wasted? Enduring Effects of streamer and Pop-up Ad Exposures on Brand Memory and military capabilitys, Journal of Electronic Commerce Research, 9, February, 5161. Cook, Thomas D. and Donald T. Campbell (1979), Quasi-Experimentation Design & Analysis Issues for Field Settings. Chicago Rand McNally College Publishing Company. Coulter, Keith S. 1998), The Effects of Affective Responses to Media Context on Advertising Evaluations, Journal of Advertising, 27, Winter, 4151. Delo, Cotton (2012), Does Facebook Know Youre Pregnant? What It Knows Depends on Whom You Ask Social Network Says One Thing, Its Advertisers Another, Advertising Age, 83, September 10, 1820. Dreze, Xavier and Francois-Xavier Hussherr (2003), Internet Advertising Is Anybody Watching? , Journal of Interactive Marketing, 17, Autumn, 823. Ephron, Erwin (1995), More Weeks, Less Weight The Shelf-spac e Model of Advertising, Journal of Advertising Research, 35, May-June, 1823.Faul, Franz, Edgar Erdfelder, Albert-Georg Lang, and Axel Buchner (2007), G*Power 3 A Flexible Statistical Power Analysis Program for the Social, Behavioral, and Biomedical Sciences, Behavior Research Methods, 39, 2, 17591. Fazio, Russell H. , David M. Sanbonmatsu, Martha C. Powell, and Frank R. Kardes (1986), On the Automatic Activation of Attitudes, Journal of character and Social Psychology, 50, 2, 22938. Gal-Or, Esther and Mordechai Gal-Or (2005), Customized Advertising via a Common Media Distributor, Marketing Science, 24, Spring, 24153. ,, Jerrold H. May, and William E.Spangler (2006), Targeted Advertising Strategies on Television, Management Science, 52, May, 71325. Goldberg, Marvin E. and Gerald J. Gorn (1987), Happy and Sad TV Programs How they Affect Reactions to Commercials, Journal of Consumer Research, 14, December, 387403. Goldfarb, Avi and Catherine E. Tucker (2011), Privacy Regulation and Onl ine Advertising, Management Science, 57, January, 5771. Green, Donald Philip, Susan Lee Goldman, and Peter Salovey (1993), Measurement Error Masks Bipolarity in Affect Ratings, Journal of Personality and Social Psychology, 64, 6, 102941.Greene, William F. (1988), Maybe the Valley of the Shadow Isnt so Dark After All, Journal of Advertising Research, 28, October-November, 115. Gustafson, Paul and S. Siddarth (2007), Describing the Dynamics of Attention to TV Commercials A Hierarchical Bayes Analysis of the Time to Zap an Ad, Journal of Applied Statistics, 34, July, 585609. Iyer, Ganesh, David Soberman, and J. Miguel Villas-Boas (2005), The Targeting of Advertising, Marketing Science, 24, Summer, 46176. Please cite this article as Steven Bellman, et al. Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (2013), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001 S. Bellman et al. / Journal of Interactive Marketing xx (2013) xxxxxx Ja mieson, Linda F. and Frank M. Bass (1989), Adjusting Stated Intention Measures to Predict Trial Purchase of New Products A Comparison of Models and Methods, Journal of Marketing Research, 26, August, 33645. Janiszewski, Chris (1998), The Influence of Display Characteristics on Visual alpha Search, Journal of Consumer Research, 25, December, 290301.Jansen, Bernard J. and T. Mullen (2008), Sponsored Search An Overview of the Concept, History, and Technology, International Journal of Electronic Business, 6, 2, 11431. , Theresa B. Flaherty, Ricardo Baeza-Yates, Lee Hunter, Brendan Kitts, and Jamie Murphy (2009), The Components and Impact of Sponsored Search, Computer, 42, May, 98101. Juster, F. Thomas (1966), Consumer Buying Intentions and Purchase Probability An Experiment in Survey Design, Journal of the American Statistical Association, 61, 315, 65896. Kover, Arthur J. and Joseph Abruzzo (1993), The RossiterPercy Grid and excited Response to Advertising An Initial Evaluation, Journa l of Advertising Research, 33, November/December, 217. Lacey, John I. (1967), Somatic Response Patterning and focus Some Revisions of Activation Theory, Psychological Stress Issues in Research, Mortimer H. Appley, Richard Trumbull, editors. New York, NY AppletonCentury-Crofts, 1442. Lang, Peter J. , Mark K. Greenwald, Margaret M. Bradley, and Alfons O. Hamm (1993), Looking at Pictures Affective, Facial, Visceral, and Behavioral Reactions, Psychophysiology, 30, May, 26173.Langheinrich, Marc, Atsuyoshi Nakamura, Naoki Abe, Tomonari Kamba, and Yoshiyuki Koseki (1999), Unintrusive Customization Techniques for Web Advertising, Computer Networks, 31, May, 125972. Marcus, Claudio and Tara Walpert (2007), Emerging Applications and Challenges of Addressable Television Advertising, Journal of Integrated Marketing Communications, 2007, 117. Mittal, Banwari (1995), A Comparative Analysis of Four Scales of Consumer Involvement, Psychology and Marketing, 12, October, 66382. Moe, Wendy W. 2003), Buying, Searching, or Browsing Differentiating Between Online Shoppers Using In-store Navigational Clickstream, Journal of Consumer Psychology, 13, 1/2, 2939. (2006), An Empirical Two-stage Choice Model with Varying Decision Rules Applied to Internet Clickstream Data, Journal of Marketing Research, 43, November, 68092. Moorthy, Sridhar, Brian T. Ratchford, and Debabrata Talukdar (1997), Consumer Information Search Revisited Theory and Empirical Analysis, Journal of Consumer Research, 23, March, 26377. Norris, Claire E. , Andrew M. Colman, and Paulo A.Aleixo (2003), Selective Exposure to Television Programmes and Advertising Effectiveness, Applied Cognitive Psychology, 17, July, 593606. Olney, Thomas J. , Morris B. Holbrook, and Rajeev Batra (1991), Consumer Responses to Advertising The Effects of Ad Content, Emotions, and Attitude Toward the Ad on Viewing Time, Journal of Consumer Research, 17, March, 44053. Pechmann, Cornelia and David W. Stewart (1989), Advertising Repetition A C ritical Review of Wearin and Wearout, Current Issues and Research in Advertising, 11, 12, 285329. Peltier, James W. , George R. Milne, and Joseph E.Phelps (2009), Information Privacy Research Framework for Integrating Multiple Publics, Information persuades, and Responses, Journal of Interactive Marketing, 23, May, 191205. Perkins, Andrew W. and Mark R. Forehand (2012), Implicit Self-referencing The Effect of Nonvolitional Self-association on Brand and Product Attitude, Journal of Consumer Research, 39, June, 14256. Pew Internet and American Life Project (2012), Trend Data (Adults) Demographics of Internet Users, (August 2012 (accessed September 19, 2012 from) http//www. pewinternet. org/Static-Pages/Trend-Data-(Adults)/ Whos-Online. aspx). Pieters, Rik G.M. and Tammo H. A. Bijmolt (1997), Consumer Memory for Television Advertising A Field Study of Duration, Serial Position, and Competition Effects, Journal of Consumer Research, 23, March, 36272. 11 Pilotta, Joseph J. and Don Schul tz (2005), Simultaneous Media Experience and Synesthesia, Journal of Advertising Research, 45, March, 1926. Potter, Robert F. and Paul D. Bolls (2012), Psychophysiological Measures and Meaning Cognitive and Emotional Processing of Media. New York, NY Routledge. Ratchford, Brian T. (1987), New Insights About the FCB Grid, Journal of Advertising Research, 27, August-September, 2438.Richins, Marsha L. and Peter H. Bloch (1986), After the New Wears Off The Temporal Context of Product Involvement, Journal of Consumer Research, 13, September, 2805. Rossiter, John R. , Larry Percy, and Robert J. Donovan (1991), A Better Advertising Planning Grid, Journal of Advertising Research, 31, OctoberNovember, 1121. , and (1997), Advertising Communication and Promotion Management. 2nd ed. New York, NY McGraw-Hill. Rutz, Oliver J. and Randolph E. Bucklin (2011), From Generic to mark A Model of Spillover in Paid Search Advertising, Journal of Marketing Research, 48, February, 87102.Sethuraman, Raj, G erard J. Tellis, and Richard A. Briesch (2011), How Well Does Advertising Work? Generalizations from Meta-analysis of Brand Advertising Elasticities, Journal of Marketing Research, 48, June, 45771. Shkedi, Roy (2010), Targeted Television Advertisements Based on Online Behavior, United States Patent No. US 7,861,260 B2. Siddarth, S. and Amitava Chattopadhyay (1998), To Zap or not to Zap A Study of the Determinants of Channel Switching during Commercials, Marketing Science, 17, 2, 12438.Spangler, William E. , Kathleen S. Hartzel, and Mordechai Gal-Or (2006), Exploring the Privacy Implications of Addressable Advertising and Viewer Profiling, Communications of the ACM, 49, May, 11923. Strong, Edward K. (1925), The Psychology of Selling and Advertising. New York, NY McGraw-Hill. Teixera, Thales S. , Michel Wedel, and Rik Pieters (2010), Moment-to-Moment Optimal Branding in TV Commercials Preventing shunning by Pulsing, Marketing Science, 29, SeptemberOctober, 783804. Tse, Alan Ching Biu and Ruby P.W. Lee (2001), Zapping Behavior during Commercial Breaks, Journal of Advertising Research, 41, May-June, 259. Vakratsas, Demetrios and Tim Ambler (1999), How Advertising Works What Do We in reality Know, Journal of Marketing, 63, January, 2643. Vascellaro, Jessica E. (2011a), Calculating the Benefit of a Targeted TV Ad, Wall Street Journal Blogs Digits Technology News and Insights (March 7 (accessed March 7, 2011 from) http//blogs. wsj. com/digits/2011/03/07/calculatingthe-benefit-of-a-targeted-tv-ad/). (2011b), TVs Next wander Tuning in to You, Wall Street Journal, March 7, A1. Vaughn, Richard (1986), How Advertising Works A Planning Model Revisited, Journal of Advertising Research, 26, February-March, 5766. Venkatesh, Viswanath and Ritu Agarwal (2006), Turning Visitors Into Customers A Usability-centric Perspective on Purchase Behavior in Electronic Channels, Management Science, 52, March, 36782. Verhoef, Peter C. , Rajkumar Venkatesan, Leigh McAlister, Edward C.Mal thouse, Manfred Krafft, and Shankar Ganesan (2010), CRM in Data-rich Multichannel Retailing Environments A Review and Future Research Directions, Journal of Interactive Marketing, 24, May, 12137. Wainer, Howard (1991), Adjusting for Differential Base Rates Lords Paradox Again, Psychological Bulletin, 109, 1, 14751. Wilbur, Kenneth C. (2008), How the Digital Video Recorder (DVR) Changes Traditional Television Advertising, Journal of Advertising, 37, Spring, 1439. Winer, Ben J. (1991), Statistical Principles in data-based Design. rd ed. New York McGraw-Hill. Woltman-Elpers, Josephine L. C. M. , Michel Wedel, and Rik G. M. Pieters (2003), Why Do Consumers Stop Viewing Television Commercials? Two Experiments on the Influence of Moment-to-Moment Entertainment and Information Value, Journal of Marketing Research, 40, November, 43753. Please cite this article as Steven Bellman, et al. , Using Internet Behavior to Deliver Relevant Television Commercials, Journal of Interactive Marketing (20 13), http// dx. doi. org/10. 1016/j. intmar. 2012. 12. 001